{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 11187,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0002681684097613301,
      "grad_norm": 11.546087741490192,
      "learning_rate": 8.936550491510278e-09,
      "loss": 0.3111,
      "step": 1
    },
    {
      "epoch": 0.0005363368195226602,
      "grad_norm": 9.953375608712667,
      "learning_rate": 1.7873100983020556e-08,
      "loss": 0.2913,
      "step": 2
    },
    {
      "epoch": 0.0008045052292839903,
      "grad_norm": 10.083574369824506,
      "learning_rate": 2.6809651474530834e-08,
      "loss": 0.3191,
      "step": 3
    },
    {
      "epoch": 0.0010726736390453205,
      "grad_norm": 9.447970830397251,
      "learning_rate": 3.574620196604111e-08,
      "loss": 0.2578,
      "step": 4
    },
    {
      "epoch": 0.0013408420488066506,
      "grad_norm": 11.251499396258808,
      "learning_rate": 4.4682752457551387e-08,
      "loss": 0.3124,
      "step": 5
    },
    {
      "epoch": 0.0016090104585679806,
      "grad_norm": 9.761419389834565,
      "learning_rate": 5.361930294906167e-08,
      "loss": 0.2643,
      "step": 6
    },
    {
      "epoch": 0.0018771788683293108,
      "grad_norm": 10.706098225922805,
      "learning_rate": 6.255585344057195e-08,
      "loss": 0.2931,
      "step": 7
    },
    {
      "epoch": 0.002145347278090641,
      "grad_norm": 8.634989096819421,
      "learning_rate": 7.149240393208222e-08,
      "loss": 0.2391,
      "step": 8
    },
    {
      "epoch": 0.002413515687851971,
      "grad_norm": 9.36594310115394,
      "learning_rate": 8.04289544235925e-08,
      "loss": 0.244,
      "step": 9
    },
    {
      "epoch": 0.0026816840976133013,
      "grad_norm": 9.897222332347056,
      "learning_rate": 8.936550491510277e-08,
      "loss": 0.2507,
      "step": 10
    },
    {
      "epoch": 0.0029498525073746312,
      "grad_norm": 10.265339217923621,
      "learning_rate": 9.830205540661306e-08,
      "loss": 0.3087,
      "step": 11
    },
    {
      "epoch": 0.003218020917135961,
      "grad_norm": 8.984086419805022,
      "learning_rate": 1.0723860589812334e-07,
      "loss": 0.2468,
      "step": 12
    },
    {
      "epoch": 0.0034861893268972916,
      "grad_norm": 8.435693892454692,
      "learning_rate": 1.1617515638963361e-07,
      "loss": 0.2275,
      "step": 13
    },
    {
      "epoch": 0.0037543577366586215,
      "grad_norm": 10.277637040079355,
      "learning_rate": 1.251117068811439e-07,
      "loss": 0.282,
      "step": 14
    },
    {
      "epoch": 0.004022526146419952,
      "grad_norm": 8.68011596079924,
      "learning_rate": 1.3404825737265417e-07,
      "loss": 0.2324,
      "step": 15
    },
    {
      "epoch": 0.004290694556181282,
      "grad_norm": 8.284776298222052,
      "learning_rate": 1.4298480786416445e-07,
      "loss": 0.2202,
      "step": 16
    },
    {
      "epoch": 0.004558862965942612,
      "grad_norm": 9.401693902327883,
      "learning_rate": 1.5192135835567472e-07,
      "loss": 0.2571,
      "step": 17
    },
    {
      "epoch": 0.004827031375703942,
      "grad_norm": 9.550271277538208,
      "learning_rate": 1.60857908847185e-07,
      "loss": 0.2842,
      "step": 18
    },
    {
      "epoch": 0.005095199785465273,
      "grad_norm": 9.133814633288498,
      "learning_rate": 1.6979445933869527e-07,
      "loss": 0.2666,
      "step": 19
    },
    {
      "epoch": 0.0053633681952266025,
      "grad_norm": 10.390099110752441,
      "learning_rate": 1.7873100983020555e-07,
      "loss": 0.3296,
      "step": 20
    },
    {
      "epoch": 0.0056315366049879325,
      "grad_norm": 7.785194449037081,
      "learning_rate": 1.8766756032171582e-07,
      "loss": 0.2268,
      "step": 21
    },
    {
      "epoch": 0.0058997050147492625,
      "grad_norm": 10.33638714203637,
      "learning_rate": 1.9660411081322612e-07,
      "loss": 0.2862,
      "step": 22
    },
    {
      "epoch": 0.006167873424510592,
      "grad_norm": 9.11443524442616,
      "learning_rate": 2.055406613047364e-07,
      "loss": 0.2795,
      "step": 23
    },
    {
      "epoch": 0.006436041834271922,
      "grad_norm": 8.59863373964623,
      "learning_rate": 2.1447721179624667e-07,
      "loss": 0.2628,
      "step": 24
    },
    {
      "epoch": 0.006704210244033253,
      "grad_norm": 8.049790757653193,
      "learning_rate": 2.2341376228775695e-07,
      "loss": 0.245,
      "step": 25
    },
    {
      "epoch": 0.006972378653794583,
      "grad_norm": 7.239513277027356,
      "learning_rate": 2.3235031277926722e-07,
      "loss": 0.2461,
      "step": 26
    },
    {
      "epoch": 0.007240547063555913,
      "grad_norm": 7.1298417211879235,
      "learning_rate": 2.412868632707775e-07,
      "loss": 0.2581,
      "step": 27
    },
    {
      "epoch": 0.007508715473317243,
      "grad_norm": 7.779228481865567,
      "learning_rate": 2.502234137622878e-07,
      "loss": 0.2421,
      "step": 28
    },
    {
      "epoch": 0.007776883883078573,
      "grad_norm": 6.470433619076548,
      "learning_rate": 2.5915996425379804e-07,
      "loss": 0.2052,
      "step": 29
    },
    {
      "epoch": 0.008045052292839904,
      "grad_norm": 6.236463756139538,
      "learning_rate": 2.6809651474530835e-07,
      "loss": 0.2009,
      "step": 30
    },
    {
      "epoch": 0.008313220702601234,
      "grad_norm": 4.587968342061977,
      "learning_rate": 2.770330652368186e-07,
      "loss": 0.1814,
      "step": 31
    },
    {
      "epoch": 0.008581389112362564,
      "grad_norm": 5.024911407808915,
      "learning_rate": 2.859696157283289e-07,
      "loss": 0.2327,
      "step": 32
    },
    {
      "epoch": 0.008849557522123894,
      "grad_norm": 4.59761757749979,
      "learning_rate": 2.9490616621983914e-07,
      "loss": 0.1625,
      "step": 33
    },
    {
      "epoch": 0.009117725931885224,
      "grad_norm": 4.065074858815684,
      "learning_rate": 3.0384271671134944e-07,
      "loss": 0.1453,
      "step": 34
    },
    {
      "epoch": 0.009385894341646554,
      "grad_norm": 4.078798101124644,
      "learning_rate": 3.127792672028597e-07,
      "loss": 0.1847,
      "step": 35
    },
    {
      "epoch": 0.009654062751407884,
      "grad_norm": 4.323295936444864,
      "learning_rate": 3.2171581769437e-07,
      "loss": 0.1868,
      "step": 36
    },
    {
      "epoch": 0.009922231161169213,
      "grad_norm": 3.949963666581816,
      "learning_rate": 3.3065236818588024e-07,
      "loss": 0.1906,
      "step": 37
    },
    {
      "epoch": 0.010190399570930545,
      "grad_norm": 3.2811446803161313,
      "learning_rate": 3.3958891867739054e-07,
      "loss": 0.1658,
      "step": 38
    },
    {
      "epoch": 0.010458567980691875,
      "grad_norm": 2.919088266268021,
      "learning_rate": 3.4852546916890084e-07,
      "loss": 0.1501,
      "step": 39
    },
    {
      "epoch": 0.010726736390453205,
      "grad_norm": 4.237774178797634,
      "learning_rate": 3.574620196604111e-07,
      "loss": 0.1772,
      "step": 40
    },
    {
      "epoch": 0.010994904800214535,
      "grad_norm": 3.09519794159859,
      "learning_rate": 3.663985701519214e-07,
      "loss": 0.2102,
      "step": 41
    },
    {
      "epoch": 0.011263073209975865,
      "grad_norm": 2.4541059356482333,
      "learning_rate": 3.7533512064343164e-07,
      "loss": 0.1405,
      "step": 42
    },
    {
      "epoch": 0.011531241619737195,
      "grad_norm": 2.312503120150902,
      "learning_rate": 3.8427167113494194e-07,
      "loss": 0.1433,
      "step": 43
    },
    {
      "epoch": 0.011799410029498525,
      "grad_norm": 2.3968643746407965,
      "learning_rate": 3.9320822162645224e-07,
      "loss": 0.1169,
      "step": 44
    },
    {
      "epoch": 0.012067578439259855,
      "grad_norm": 1.845564836703066,
      "learning_rate": 4.021447721179625e-07,
      "loss": 0.1158,
      "step": 45
    },
    {
      "epoch": 0.012335746849021185,
      "grad_norm": 2.211840397481718,
      "learning_rate": 4.110813226094728e-07,
      "loss": 0.1478,
      "step": 46
    },
    {
      "epoch": 0.012603915258782515,
      "grad_norm": 1.9718038161636013,
      "learning_rate": 4.2001787310098304e-07,
      "loss": 0.158,
      "step": 47
    },
    {
      "epoch": 0.012872083668543845,
      "grad_norm": 2.2523119193681467,
      "learning_rate": 4.2895442359249334e-07,
      "loss": 0.1632,
      "step": 48
    },
    {
      "epoch": 0.013140252078305176,
      "grad_norm": 2.059556702923991,
      "learning_rate": 4.378909740840036e-07,
      "loss": 0.1268,
      "step": 49
    },
    {
      "epoch": 0.013408420488066506,
      "grad_norm": 2.0221668076822907,
      "learning_rate": 4.468275245755139e-07,
      "loss": 0.1747,
      "step": 50
    },
    {
      "epoch": 0.013676588897827836,
      "grad_norm": 1.3183282774039613,
      "learning_rate": 4.557640750670242e-07,
      "loss": 0.0998,
      "step": 51
    },
    {
      "epoch": 0.013944757307589166,
      "grad_norm": 1.9226372556275977,
      "learning_rate": 4.6470062555853444e-07,
      "loss": 0.1229,
      "step": 52
    },
    {
      "epoch": 0.014212925717350496,
      "grad_norm": 1.6578946859244836,
      "learning_rate": 4.7363717605004474e-07,
      "loss": 0.125,
      "step": 53
    },
    {
      "epoch": 0.014481094127111826,
      "grad_norm": 2.00885525013515,
      "learning_rate": 4.82573726541555e-07,
      "loss": 0.1275,
      "step": 54
    },
    {
      "epoch": 0.014749262536873156,
      "grad_norm": 1.511464235405912,
      "learning_rate": 4.915102770330653e-07,
      "loss": 0.1131,
      "step": 55
    },
    {
      "epoch": 0.015017430946634486,
      "grad_norm": 1.9971433448480032,
      "learning_rate": 5.004468275245756e-07,
      "loss": 0.1147,
      "step": 56
    },
    {
      "epoch": 0.015285599356395816,
      "grad_norm": 1.5847978657117172,
      "learning_rate": 5.093833780160858e-07,
      "loss": 0.1201,
      "step": 57
    },
    {
      "epoch": 0.015553767766157146,
      "grad_norm": 2.0949312852003965,
      "learning_rate": 5.183199285075961e-07,
      "loss": 0.1002,
      "step": 58
    },
    {
      "epoch": 0.015821936175918476,
      "grad_norm": 2.166412250989985,
      "learning_rate": 5.272564789991064e-07,
      "loss": 0.1451,
      "step": 59
    },
    {
      "epoch": 0.016090104585679808,
      "grad_norm": 2.081983124826213,
      "learning_rate": 5.361930294906167e-07,
      "loss": 0.1474,
      "step": 60
    },
    {
      "epoch": 0.016358272995441136,
      "grad_norm": 2.0203000295104703,
      "learning_rate": 5.451295799821269e-07,
      "loss": 0.13,
      "step": 61
    },
    {
      "epoch": 0.016626441405202468,
      "grad_norm": 1.4814879197707982,
      "learning_rate": 5.540661304736372e-07,
      "loss": 0.1256,
      "step": 62
    },
    {
      "epoch": 0.016894609814963796,
      "grad_norm": 1.7368842792391574,
      "learning_rate": 5.630026809651475e-07,
      "loss": 0.1156,
      "step": 63
    },
    {
      "epoch": 0.017162778224725127,
      "grad_norm": 1.5395558579212352,
      "learning_rate": 5.719392314566578e-07,
      "loss": 0.121,
      "step": 64
    },
    {
      "epoch": 0.01743094663448646,
      "grad_norm": 1.5812943147268939,
      "learning_rate": 5.808757819481681e-07,
      "loss": 0.1199,
      "step": 65
    },
    {
      "epoch": 0.017699115044247787,
      "grad_norm": 2.2486492517524708,
      "learning_rate": 5.898123324396783e-07,
      "loss": 0.1176,
      "step": 66
    },
    {
      "epoch": 0.01796728345400912,
      "grad_norm": 1.5989523680967617,
      "learning_rate": 5.987488829311886e-07,
      "loss": 0.1054,
      "step": 67
    },
    {
      "epoch": 0.018235451863770447,
      "grad_norm": 1.7234831440536167,
      "learning_rate": 6.076854334226989e-07,
      "loss": 0.1416,
      "step": 68
    },
    {
      "epoch": 0.01850362027353178,
      "grad_norm": 2.158302389030971,
      "learning_rate": 6.166219839142092e-07,
      "loss": 0.1474,
      "step": 69
    },
    {
      "epoch": 0.018771788683293107,
      "grad_norm": 2.478784731943137,
      "learning_rate": 6.255585344057194e-07,
      "loss": 0.1171,
      "step": 70
    },
    {
      "epoch": 0.01903995709305444,
      "grad_norm": 1.360559845020528,
      "learning_rate": 6.344950848972298e-07,
      "loss": 0.0987,
      "step": 71
    },
    {
      "epoch": 0.019308125502815767,
      "grad_norm": 1.1204522642522026,
      "learning_rate": 6.4343163538874e-07,
      "loss": 0.0744,
      "step": 72
    },
    {
      "epoch": 0.0195762939125771,
      "grad_norm": 2.6753643379873027,
      "learning_rate": 6.523681858802503e-07,
      "loss": 0.1074,
      "step": 73
    },
    {
      "epoch": 0.019844462322338427,
      "grad_norm": 1.7157580160480599,
      "learning_rate": 6.613047363717605e-07,
      "loss": 0.0985,
      "step": 74
    },
    {
      "epoch": 0.02011263073209976,
      "grad_norm": 1.7968912989210934,
      "learning_rate": 6.702412868632709e-07,
      "loss": 0.1301,
      "step": 75
    },
    {
      "epoch": 0.02038079914186109,
      "grad_norm": 2.0129220271235435,
      "learning_rate": 6.791778373547811e-07,
      "loss": 0.093,
      "step": 76
    },
    {
      "epoch": 0.02064896755162242,
      "grad_norm": 2.089724817770505,
      "learning_rate": 6.881143878462914e-07,
      "loss": 0.0997,
      "step": 77
    },
    {
      "epoch": 0.02091713596138375,
      "grad_norm": 1.438308295940405,
      "learning_rate": 6.970509383378017e-07,
      "loss": 0.1016,
      "step": 78
    },
    {
      "epoch": 0.02118530437114508,
      "grad_norm": 2.154762741384592,
      "learning_rate": 7.05987488829312e-07,
      "loss": 0.1048,
      "step": 79
    },
    {
      "epoch": 0.02145347278090641,
      "grad_norm": 1.7224331562153552,
      "learning_rate": 7.149240393208222e-07,
      "loss": 0.1234,
      "step": 80
    },
    {
      "epoch": 0.02172164119066774,
      "grad_norm": 1.482899071115104,
      "learning_rate": 7.238605898123326e-07,
      "loss": 0.1352,
      "step": 81
    },
    {
      "epoch": 0.02198980960042907,
      "grad_norm": 3.6555661574975833,
      "learning_rate": 7.327971403038428e-07,
      "loss": 0.1377,
      "step": 82
    },
    {
      "epoch": 0.0222579780101904,
      "grad_norm": 1.50266018779977,
      "learning_rate": 7.417336907953531e-07,
      "loss": 0.1072,
      "step": 83
    },
    {
      "epoch": 0.02252614641995173,
      "grad_norm": 1.3726027929133422,
      "learning_rate": 7.506702412868633e-07,
      "loss": 0.1054,
      "step": 84
    },
    {
      "epoch": 0.022794314829713058,
      "grad_norm": 1.2830790779857988,
      "learning_rate": 7.596067917783737e-07,
      "loss": 0.074,
      "step": 85
    },
    {
      "epoch": 0.02306248323947439,
      "grad_norm": 1.465152769407914,
      "learning_rate": 7.685433422698839e-07,
      "loss": 0.1108,
      "step": 86
    },
    {
      "epoch": 0.02333065164923572,
      "grad_norm": 1.4158571606334536,
      "learning_rate": 7.774798927613941e-07,
      "loss": 0.09,
      "step": 87
    },
    {
      "epoch": 0.02359882005899705,
      "grad_norm": 1.509351898824975,
      "learning_rate": 7.864164432529045e-07,
      "loss": 0.1205,
      "step": 88
    },
    {
      "epoch": 0.02386698846875838,
      "grad_norm": 1.2708777586285185,
      "learning_rate": 7.953529937444148e-07,
      "loss": 0.1265,
      "step": 89
    },
    {
      "epoch": 0.02413515687851971,
      "grad_norm": 1.3011507433475809,
      "learning_rate": 8.04289544235925e-07,
      "loss": 0.0941,
      "step": 90
    },
    {
      "epoch": 0.02440332528828104,
      "grad_norm": 1.925125276184661,
      "learning_rate": 8.132260947274352e-07,
      "loss": 0.1126,
      "step": 91
    },
    {
      "epoch": 0.02467149369804237,
      "grad_norm": 2.163626181375269,
      "learning_rate": 8.221626452189456e-07,
      "loss": 0.1236,
      "step": 92
    },
    {
      "epoch": 0.0249396621078037,
      "grad_norm": 1.2094321310616027,
      "learning_rate": 8.310991957104558e-07,
      "loss": 0.1022,
      "step": 93
    },
    {
      "epoch": 0.02520783051756503,
      "grad_norm": 2.0119224304097574,
      "learning_rate": 8.400357462019661e-07,
      "loss": 0.1265,
      "step": 94
    },
    {
      "epoch": 0.02547599892732636,
      "grad_norm": 1.0132683770266218,
      "learning_rate": 8.489722966934765e-07,
      "loss": 0.0843,
      "step": 95
    },
    {
      "epoch": 0.02574416733708769,
      "grad_norm": 1.321449959342236,
      "learning_rate": 8.579088471849867e-07,
      "loss": 0.1177,
      "step": 96
    },
    {
      "epoch": 0.02601233574684902,
      "grad_norm": 1.2803596390894127,
      "learning_rate": 8.668453976764969e-07,
      "loss": 0.1053,
      "step": 97
    },
    {
      "epoch": 0.026280504156610353,
      "grad_norm": 1.3875316429294406,
      "learning_rate": 8.757819481680072e-07,
      "loss": 0.091,
      "step": 98
    },
    {
      "epoch": 0.02654867256637168,
      "grad_norm": 1.598445796830463,
      "learning_rate": 8.847184986595175e-07,
      "loss": 0.1115,
      "step": 99
    },
    {
      "epoch": 0.026816840976133013,
      "grad_norm": 1.3669246808889461,
      "learning_rate": 8.936550491510278e-07,
      "loss": 0.1133,
      "step": 100
    },
    {
      "epoch": 0.02708500938589434,
      "grad_norm": 1.1443200549299357,
      "learning_rate": 9.02591599642538e-07,
      "loss": 0.0865,
      "step": 101
    },
    {
      "epoch": 0.027353177795655673,
      "grad_norm": 1.01461740344635,
      "learning_rate": 9.115281501340484e-07,
      "loss": 0.0859,
      "step": 102
    },
    {
      "epoch": 0.027621346205417,
      "grad_norm": 1.6783856953916776,
      "learning_rate": 9.204647006255586e-07,
      "loss": 0.1044,
      "step": 103
    },
    {
      "epoch": 0.027889514615178333,
      "grad_norm": 1.7192371055136828,
      "learning_rate": 9.294012511170689e-07,
      "loss": 0.099,
      "step": 104
    },
    {
      "epoch": 0.02815768302493966,
      "grad_norm": 1.4339125746991326,
      "learning_rate": 9.383378016085791e-07,
      "loss": 0.0845,
      "step": 105
    },
    {
      "epoch": 0.028425851434700992,
      "grad_norm": 1.1823270250600704,
      "learning_rate": 9.472743521000895e-07,
      "loss": 0.0818,
      "step": 106
    },
    {
      "epoch": 0.02869401984446232,
      "grad_norm": 1.2149966227513322,
      "learning_rate": 9.562109025915998e-07,
      "loss": 0.1017,
      "step": 107
    },
    {
      "epoch": 0.028962188254223652,
      "grad_norm": 2.1827174599443433,
      "learning_rate": 9.6514745308311e-07,
      "loss": 0.0856,
      "step": 108
    },
    {
      "epoch": 0.029230356663984984,
      "grad_norm": 0.9497273289254222,
      "learning_rate": 9.740840035746204e-07,
      "loss": 0.0737,
      "step": 109
    },
    {
      "epoch": 0.029498525073746312,
      "grad_norm": 1.7965517534461946,
      "learning_rate": 9.830205540661306e-07,
      "loss": 0.1048,
      "step": 110
    },
    {
      "epoch": 0.029766693483507644,
      "grad_norm": 1.667039042581746,
      "learning_rate": 9.919571045576408e-07,
      "loss": 0.0842,
      "step": 111
    },
    {
      "epoch": 0.030034861893268972,
      "grad_norm": 1.673985178498018,
      "learning_rate": 1.0008936550491512e-06,
      "loss": 0.0994,
      "step": 112
    },
    {
      "epoch": 0.030303030303030304,
      "grad_norm": 1.1084966539963994,
      "learning_rate": 1.0098302055406614e-06,
      "loss": 0.088,
      "step": 113
    },
    {
      "epoch": 0.030571198712791632,
      "grad_norm": 0.9720597320902342,
      "learning_rate": 1.0187667560321716e-06,
      "loss": 0.0847,
      "step": 114
    },
    {
      "epoch": 0.030839367122552964,
      "grad_norm": 1.2788184490897927,
      "learning_rate": 1.027703306523682e-06,
      "loss": 0.082,
      "step": 115
    },
    {
      "epoch": 0.031107535532314292,
      "grad_norm": 1.5430443172288189,
      "learning_rate": 1.0366398570151922e-06,
      "loss": 0.102,
      "step": 116
    },
    {
      "epoch": 0.031375703942075624,
      "grad_norm": 1.473097911474061,
      "learning_rate": 1.0455764075067026e-06,
      "loss": 0.1184,
      "step": 117
    },
    {
      "epoch": 0.03164387235183695,
      "grad_norm": 1.2654661888914576,
      "learning_rate": 1.0545129579982128e-06,
      "loss": 0.079,
      "step": 118
    },
    {
      "epoch": 0.03191204076159829,
      "grad_norm": 1.1052864700478846,
      "learning_rate": 1.0634495084897232e-06,
      "loss": 0.0851,
      "step": 119
    },
    {
      "epoch": 0.032180209171359615,
      "grad_norm": 1.0968028766223177,
      "learning_rate": 1.0723860589812334e-06,
      "loss": 0.0802,
      "step": 120
    },
    {
      "epoch": 0.032448377581120944,
      "grad_norm": 2.119394522864081,
      "learning_rate": 1.0813226094727436e-06,
      "loss": 0.119,
      "step": 121
    },
    {
      "epoch": 0.03271654599088227,
      "grad_norm": 1.3754365860570354,
      "learning_rate": 1.0902591599642538e-06,
      "loss": 0.1094,
      "step": 122
    },
    {
      "epoch": 0.03298471440064361,
      "grad_norm": 1.0817566918879613,
      "learning_rate": 1.0991957104557642e-06,
      "loss": 0.0807,
      "step": 123
    },
    {
      "epoch": 0.033252882810404935,
      "grad_norm": 1.4660078891560506,
      "learning_rate": 1.1081322609472744e-06,
      "loss": 0.0955,
      "step": 124
    },
    {
      "epoch": 0.03352105122016626,
      "grad_norm": 1.322438350545607,
      "learning_rate": 1.1170688114387848e-06,
      "loss": 0.0868,
      "step": 125
    },
    {
      "epoch": 0.03378921962992759,
      "grad_norm": 1.6453669902239199,
      "learning_rate": 1.126005361930295e-06,
      "loss": 0.0896,
      "step": 126
    },
    {
      "epoch": 0.03405738803968893,
      "grad_norm": 1.3169906526374648,
      "learning_rate": 1.1349419124218054e-06,
      "loss": 0.1015,
      "step": 127
    },
    {
      "epoch": 0.034325556449450255,
      "grad_norm": 1.6763197171817381,
      "learning_rate": 1.1438784629133156e-06,
      "loss": 0.1004,
      "step": 128
    },
    {
      "epoch": 0.03459372485921158,
      "grad_norm": 1.3386567444919648,
      "learning_rate": 1.1528150134048258e-06,
      "loss": 0.1044,
      "step": 129
    },
    {
      "epoch": 0.03486189326897292,
      "grad_norm": 1.099399950219615,
      "learning_rate": 1.1617515638963362e-06,
      "loss": 0.1041,
      "step": 130
    },
    {
      "epoch": 0.035130061678734247,
      "grad_norm": 1.193225836465599,
      "learning_rate": 1.1706881143878464e-06,
      "loss": 0.087,
      "step": 131
    },
    {
      "epoch": 0.035398230088495575,
      "grad_norm": 1.198084681075304,
      "learning_rate": 1.1796246648793566e-06,
      "loss": 0.1094,
      "step": 132
    },
    {
      "epoch": 0.0356663984982569,
      "grad_norm": 1.6231020969071568,
      "learning_rate": 1.188561215370867e-06,
      "loss": 0.1303,
      "step": 133
    },
    {
      "epoch": 0.03593456690801824,
      "grad_norm": 1.1228555329730097,
      "learning_rate": 1.1974977658623772e-06,
      "loss": 0.0633,
      "step": 134
    },
    {
      "epoch": 0.036202735317779566,
      "grad_norm": 1.6275135486500187,
      "learning_rate": 1.2064343163538874e-06,
      "loss": 0.0975,
      "step": 135
    },
    {
      "epoch": 0.036470903727540895,
      "grad_norm": 1.0685846990212808,
      "learning_rate": 1.2153708668453978e-06,
      "loss": 0.0901,
      "step": 136
    },
    {
      "epoch": 0.03673907213730222,
      "grad_norm": 1.1412152534648226,
      "learning_rate": 1.2243074173369082e-06,
      "loss": 0.0677,
      "step": 137
    },
    {
      "epoch": 0.03700724054706356,
      "grad_norm": 1.2740147615387647,
      "learning_rate": 1.2332439678284184e-06,
      "loss": 0.1037,
      "step": 138
    },
    {
      "epoch": 0.037275408956824886,
      "grad_norm": 1.4248639957231923,
      "learning_rate": 1.2421805183199286e-06,
      "loss": 0.1157,
      "step": 139
    },
    {
      "epoch": 0.037543577366586214,
      "grad_norm": 1.4089851788721515,
      "learning_rate": 1.2511170688114388e-06,
      "loss": 0.0801,
      "step": 140
    },
    {
      "epoch": 0.03781174577634755,
      "grad_norm": 1.5692303894639283,
      "learning_rate": 1.2600536193029492e-06,
      "loss": 0.0938,
      "step": 141
    },
    {
      "epoch": 0.03807991418610888,
      "grad_norm": 1.0578937241746678,
      "learning_rate": 1.2689901697944596e-06,
      "loss": 0.0941,
      "step": 142
    },
    {
      "epoch": 0.038348082595870206,
      "grad_norm": 1.6745613942744022,
      "learning_rate": 1.2779267202859696e-06,
      "loss": 0.1175,
      "step": 143
    },
    {
      "epoch": 0.038616251005631534,
      "grad_norm": 1.3226994059594668,
      "learning_rate": 1.28686327077748e-06,
      "loss": 0.1159,
      "step": 144
    },
    {
      "epoch": 0.03888441941539287,
      "grad_norm": 1.9491993546160369,
      "learning_rate": 1.2957998212689904e-06,
      "loss": 0.1018,
      "step": 145
    },
    {
      "epoch": 0.0391525878251542,
      "grad_norm": 1.0454887522951368,
      "learning_rate": 1.3047363717605006e-06,
      "loss": 0.0888,
      "step": 146
    },
    {
      "epoch": 0.039420756234915526,
      "grad_norm": 1.1397291273859587,
      "learning_rate": 1.3136729222520108e-06,
      "loss": 0.1033,
      "step": 147
    },
    {
      "epoch": 0.039688924644676854,
      "grad_norm": 1.2775298474847148,
      "learning_rate": 1.322609472743521e-06,
      "loss": 0.0654,
      "step": 148
    },
    {
      "epoch": 0.03995709305443819,
      "grad_norm": 1.493806290236156,
      "learning_rate": 1.3315460232350314e-06,
      "loss": 0.0975,
      "step": 149
    },
    {
      "epoch": 0.04022526146419952,
      "grad_norm": 1.1898609187066904,
      "learning_rate": 1.3404825737265418e-06,
      "loss": 0.0814,
      "step": 150
    },
    {
      "epoch": 0.040493429873960846,
      "grad_norm": 2.3125461981097715,
      "learning_rate": 1.3494191242180518e-06,
      "loss": 0.1062,
      "step": 151
    },
    {
      "epoch": 0.04076159828372218,
      "grad_norm": 1.7080973010518823,
      "learning_rate": 1.3583556747095622e-06,
      "loss": 0.1411,
      "step": 152
    },
    {
      "epoch": 0.04102976669348351,
      "grad_norm": 1.4174747561664844,
      "learning_rate": 1.3672922252010726e-06,
      "loss": 0.0893,
      "step": 153
    },
    {
      "epoch": 0.04129793510324484,
      "grad_norm": 1.0231755913661127,
      "learning_rate": 1.3762287756925828e-06,
      "loss": 0.103,
      "step": 154
    },
    {
      "epoch": 0.041566103513006165,
      "grad_norm": 1.249424641538587,
      "learning_rate": 1.385165326184093e-06,
      "loss": 0.0987,
      "step": 155
    },
    {
      "epoch": 0.0418342719227675,
      "grad_norm": 1.4209494718223319,
      "learning_rate": 1.3941018766756034e-06,
      "loss": 0.097,
      "step": 156
    },
    {
      "epoch": 0.04210244033252883,
      "grad_norm": 1.0032072565441066,
      "learning_rate": 1.4030384271671136e-06,
      "loss": 0.0677,
      "step": 157
    },
    {
      "epoch": 0.04237060874229016,
      "grad_norm": 1.1314671181491947,
      "learning_rate": 1.411974977658624e-06,
      "loss": 0.0851,
      "step": 158
    },
    {
      "epoch": 0.042638777152051485,
      "grad_norm": 1.2415077124489695,
      "learning_rate": 1.4209115281501342e-06,
      "loss": 0.1071,
      "step": 159
    },
    {
      "epoch": 0.04290694556181282,
      "grad_norm": 1.3721044667011975,
      "learning_rate": 1.4298480786416444e-06,
      "loss": 0.0967,
      "step": 160
    },
    {
      "epoch": 0.04317511397157415,
      "grad_norm": 1.2972095895372564,
      "learning_rate": 1.4387846291331548e-06,
      "loss": 0.0826,
      "step": 161
    },
    {
      "epoch": 0.04344328238133548,
      "grad_norm": 1.3859470079002227,
      "learning_rate": 1.4477211796246652e-06,
      "loss": 0.0998,
      "step": 162
    },
    {
      "epoch": 0.04371145079109681,
      "grad_norm": 1.4393563901050674,
      "learning_rate": 1.4566577301161752e-06,
      "loss": 0.1162,
      "step": 163
    },
    {
      "epoch": 0.04397961920085814,
      "grad_norm": 1.188824644398946,
      "learning_rate": 1.4655942806076856e-06,
      "loss": 0.0964,
      "step": 164
    },
    {
      "epoch": 0.04424778761061947,
      "grad_norm": 1.2382403823881156,
      "learning_rate": 1.4745308310991958e-06,
      "loss": 0.1065,
      "step": 165
    },
    {
      "epoch": 0.0445159560203808,
      "grad_norm": 0.9862500206144987,
      "learning_rate": 1.4834673815907062e-06,
      "loss": 0.1199,
      "step": 166
    },
    {
      "epoch": 0.04478412443014213,
      "grad_norm": 1.4656954794170536,
      "learning_rate": 1.4924039320822164e-06,
      "loss": 0.1021,
      "step": 167
    },
    {
      "epoch": 0.04505229283990346,
      "grad_norm": 0.813784566619627,
      "learning_rate": 1.5013404825737266e-06,
      "loss": 0.061,
      "step": 168
    },
    {
      "epoch": 0.04532046124966479,
      "grad_norm": 1.0919041252310842,
      "learning_rate": 1.510277033065237e-06,
      "loss": 0.0804,
      "step": 169
    },
    {
      "epoch": 0.045588629659426116,
      "grad_norm": 1.090801767575039,
      "learning_rate": 1.5192135835567474e-06,
      "loss": 0.0658,
      "step": 170
    },
    {
      "epoch": 0.04585679806918745,
      "grad_norm": 1.7751565185330098,
      "learning_rate": 1.5281501340482574e-06,
      "loss": 0.12,
      "step": 171
    },
    {
      "epoch": 0.04612496647894878,
      "grad_norm": 1.3939004891517237,
      "learning_rate": 1.5370866845397678e-06,
      "loss": 0.0997,
      "step": 172
    },
    {
      "epoch": 0.04639313488871011,
      "grad_norm": 1.1556505204332022,
      "learning_rate": 1.5460232350312782e-06,
      "loss": 0.0853,
      "step": 173
    },
    {
      "epoch": 0.04666130329847144,
      "grad_norm": 1.8067880076074672,
      "learning_rate": 1.5549597855227882e-06,
      "loss": 0.1104,
      "step": 174
    },
    {
      "epoch": 0.04692947170823277,
      "grad_norm": 1.4818208702632059,
      "learning_rate": 1.5638963360142986e-06,
      "loss": 0.0993,
      "step": 175
    },
    {
      "epoch": 0.0471976401179941,
      "grad_norm": 1.060650239550061,
      "learning_rate": 1.572832886505809e-06,
      "loss": 0.0847,
      "step": 176
    },
    {
      "epoch": 0.04746580852775543,
      "grad_norm": 1.2781630339731105,
      "learning_rate": 1.5817694369973192e-06,
      "loss": 0.0995,
      "step": 177
    },
    {
      "epoch": 0.04773397693751676,
      "grad_norm": 1.2122131064061366,
      "learning_rate": 1.5907059874888296e-06,
      "loss": 0.0861,
      "step": 178
    },
    {
      "epoch": 0.04800214534727809,
      "grad_norm": 1.002858431869078,
      "learning_rate": 1.5996425379803396e-06,
      "loss": 0.0874,
      "step": 179
    },
    {
      "epoch": 0.04827031375703942,
      "grad_norm": 1.5415661235806746,
      "learning_rate": 1.60857908847185e-06,
      "loss": 0.1106,
      "step": 180
    },
    {
      "epoch": 0.04853848216680075,
      "grad_norm": 1.6895963052964655,
      "learning_rate": 1.6175156389633604e-06,
      "loss": 0.0772,
      "step": 181
    },
    {
      "epoch": 0.04880665057656208,
      "grad_norm": 1.906945789365188,
      "learning_rate": 1.6264521894548704e-06,
      "loss": 0.0942,
      "step": 182
    },
    {
      "epoch": 0.04907481898632341,
      "grad_norm": 1.3593115252223362,
      "learning_rate": 1.6353887399463808e-06,
      "loss": 0.1009,
      "step": 183
    },
    {
      "epoch": 0.04934298739608474,
      "grad_norm": 1.2707177701223717,
      "learning_rate": 1.6443252904378912e-06,
      "loss": 0.0809,
      "step": 184
    },
    {
      "epoch": 0.049611155805846074,
      "grad_norm": 1.0345907315663958,
      "learning_rate": 1.6532618409294014e-06,
      "loss": 0.0913,
      "step": 185
    },
    {
      "epoch": 0.0498793242156074,
      "grad_norm": 1.1327901025470064,
      "learning_rate": 1.6621983914209116e-06,
      "loss": 0.0717,
      "step": 186
    },
    {
      "epoch": 0.05014749262536873,
      "grad_norm": 1.617990122545335,
      "learning_rate": 1.671134941912422e-06,
      "loss": 0.1044,
      "step": 187
    },
    {
      "epoch": 0.05041566103513006,
      "grad_norm": 1.275293844844148,
      "learning_rate": 1.6800714924039322e-06,
      "loss": 0.1282,
      "step": 188
    },
    {
      "epoch": 0.050683829444891394,
      "grad_norm": 0.7745382686882211,
      "learning_rate": 1.6890080428954426e-06,
      "loss": 0.0549,
      "step": 189
    },
    {
      "epoch": 0.05095199785465272,
      "grad_norm": 0.9088949475908857,
      "learning_rate": 1.697944593386953e-06,
      "loss": 0.0783,
      "step": 190
    },
    {
      "epoch": 0.05122016626441405,
      "grad_norm": 1.1097357241011565,
      "learning_rate": 1.706881143878463e-06,
      "loss": 0.065,
      "step": 191
    },
    {
      "epoch": 0.05148833467417538,
      "grad_norm": 1.2961857805427983,
      "learning_rate": 1.7158176943699734e-06,
      "loss": 0.0784,
      "step": 192
    },
    {
      "epoch": 0.051756503083936714,
      "grad_norm": 1.6996578847145942,
      "learning_rate": 1.7247542448614838e-06,
      "loss": 0.0859,
      "step": 193
    },
    {
      "epoch": 0.05202467149369804,
      "grad_norm": 1.967996884334931,
      "learning_rate": 1.7336907953529938e-06,
      "loss": 0.0709,
      "step": 194
    },
    {
      "epoch": 0.05229283990345937,
      "grad_norm": 1.3595184169562722,
      "learning_rate": 1.7426273458445042e-06,
      "loss": 0.0859,
      "step": 195
    },
    {
      "epoch": 0.052561008313220706,
      "grad_norm": 1.135911814953798,
      "learning_rate": 1.7515638963360144e-06,
      "loss": 0.089,
      "step": 196
    },
    {
      "epoch": 0.052829176722982034,
      "grad_norm": 0.9422063898824449,
      "learning_rate": 1.7605004468275248e-06,
      "loss": 0.0729,
      "step": 197
    },
    {
      "epoch": 0.05309734513274336,
      "grad_norm": 0.9875862591713065,
      "learning_rate": 1.769436997319035e-06,
      "loss": 0.0754,
      "step": 198
    },
    {
      "epoch": 0.05336551354250469,
      "grad_norm": 1.3623897132632274,
      "learning_rate": 1.7783735478105452e-06,
      "loss": 0.091,
      "step": 199
    },
    {
      "epoch": 0.053633681952266025,
      "grad_norm": 0.9798723111197029,
      "learning_rate": 1.7873100983020556e-06,
      "loss": 0.0635,
      "step": 200
    },
    {
      "epoch": 0.053901850362027354,
      "grad_norm": 1.5508263393183908,
      "learning_rate": 1.796246648793566e-06,
      "loss": 0.0698,
      "step": 201
    },
    {
      "epoch": 0.05417001877178868,
      "grad_norm": 1.0606035220148973,
      "learning_rate": 1.805183199285076e-06,
      "loss": 0.0691,
      "step": 202
    },
    {
      "epoch": 0.05443818718155001,
      "grad_norm": 1.1805117048051683,
      "learning_rate": 1.8141197497765864e-06,
      "loss": 0.0949,
      "step": 203
    },
    {
      "epoch": 0.054706355591311345,
      "grad_norm": 1.1891551679992307,
      "learning_rate": 1.8230563002680968e-06,
      "loss": 0.0824,
      "step": 204
    },
    {
      "epoch": 0.054974524001072674,
      "grad_norm": 1.0004818694010413,
      "learning_rate": 1.831992850759607e-06,
      "loss": 0.0834,
      "step": 205
    },
    {
      "epoch": 0.055242692410834,
      "grad_norm": 1.7031573973935137,
      "learning_rate": 1.8409294012511172e-06,
      "loss": 0.0772,
      "step": 206
    },
    {
      "epoch": 0.05551086082059534,
      "grad_norm": 1.1022836761180141,
      "learning_rate": 1.8498659517426276e-06,
      "loss": 0.0727,
      "step": 207
    },
    {
      "epoch": 0.055779029230356665,
      "grad_norm": 0.97464431723532,
      "learning_rate": 1.8588025022341378e-06,
      "loss": 0.0799,
      "step": 208
    },
    {
      "epoch": 0.05604719764011799,
      "grad_norm": 1.1196614965663036,
      "learning_rate": 1.8677390527256482e-06,
      "loss": 0.0758,
      "step": 209
    },
    {
      "epoch": 0.05631536604987932,
      "grad_norm": 1.0214083368219748,
      "learning_rate": 1.8766756032171582e-06,
      "loss": 0.0804,
      "step": 210
    },
    {
      "epoch": 0.05658353445964066,
      "grad_norm": 1.7073239247139622,
      "learning_rate": 1.8856121537086686e-06,
      "loss": 0.0834,
      "step": 211
    },
    {
      "epoch": 0.056851702869401985,
      "grad_norm": 1.6994048032035973,
      "learning_rate": 1.894548704200179e-06,
      "loss": 0.1011,
      "step": 212
    },
    {
      "epoch": 0.05711987127916331,
      "grad_norm": 0.9450318226641397,
      "learning_rate": 1.903485254691689e-06,
      "loss": 0.0696,
      "step": 213
    },
    {
      "epoch": 0.05738803968892464,
      "grad_norm": 1.302625736882656,
      "learning_rate": 1.9124218051831996e-06,
      "loss": 0.0766,
      "step": 214
    },
    {
      "epoch": 0.057656208098685977,
      "grad_norm": 1.0940203942956976,
      "learning_rate": 1.9213583556747098e-06,
      "loss": 0.0813,
      "step": 215
    },
    {
      "epoch": 0.057924376508447305,
      "grad_norm": 1.0980544705207405,
      "learning_rate": 1.93029490616622e-06,
      "loss": 0.0975,
      "step": 216
    },
    {
      "epoch": 0.05819254491820863,
      "grad_norm": 1.003720133760256,
      "learning_rate": 1.93923145665773e-06,
      "loss": 0.0847,
      "step": 217
    },
    {
      "epoch": 0.05846071332796997,
      "grad_norm": 0.992449677103783,
      "learning_rate": 1.9481680071492408e-06,
      "loss": 0.0719,
      "step": 218
    },
    {
      "epoch": 0.058728881737731296,
      "grad_norm": 1.0841219777969402,
      "learning_rate": 1.9571045576407505e-06,
      "loss": 0.0828,
      "step": 219
    },
    {
      "epoch": 0.058997050147492625,
      "grad_norm": 1.1513634602915603,
      "learning_rate": 1.966041108132261e-06,
      "loss": 0.0795,
      "step": 220
    },
    {
      "epoch": 0.05926521855725395,
      "grad_norm": 1.0694947066944582,
      "learning_rate": 1.9749776586237714e-06,
      "loss": 0.0701,
      "step": 221
    },
    {
      "epoch": 0.05953338696701529,
      "grad_norm": 1.0703881862769697,
      "learning_rate": 1.9839142091152816e-06,
      "loss": 0.0688,
      "step": 222
    },
    {
      "epoch": 0.059801555376776616,
      "grad_norm": 1.1841394468171016,
      "learning_rate": 1.992850759606792e-06,
      "loss": 0.0657,
      "step": 223
    },
    {
      "epoch": 0.060069723786537944,
      "grad_norm": 1.1142825478669494,
      "learning_rate": 2.0017873100983024e-06,
      "loss": 0.0951,
      "step": 224
    },
    {
      "epoch": 0.06033789219629927,
      "grad_norm": 0.9169747298049504,
      "learning_rate": 2.0107238605898126e-06,
      "loss": 0.0838,
      "step": 225
    },
    {
      "epoch": 0.06060606060606061,
      "grad_norm": 1.0149479432400015,
      "learning_rate": 2.0196604110813228e-06,
      "loss": 0.0795,
      "step": 226
    },
    {
      "epoch": 0.060874229015821936,
      "grad_norm": 0.9592934458994421,
      "learning_rate": 2.028596961572833e-06,
      "loss": 0.0621,
      "step": 227
    },
    {
      "epoch": 0.061142397425583264,
      "grad_norm": 0.8364743216317536,
      "learning_rate": 2.037533512064343e-06,
      "loss": 0.0552,
      "step": 228
    },
    {
      "epoch": 0.0614105658353446,
      "grad_norm": 1.458384943068544,
      "learning_rate": 2.0464700625558538e-06,
      "loss": 0.0674,
      "step": 229
    },
    {
      "epoch": 0.06167873424510593,
      "grad_norm": 1.1258437427885113,
      "learning_rate": 2.055406613047364e-06,
      "loss": 0.0885,
      "step": 230
    },
    {
      "epoch": 0.061946902654867256,
      "grad_norm": 1.2821790247045186,
      "learning_rate": 2.064343163538874e-06,
      "loss": 0.0884,
      "step": 231
    },
    {
      "epoch": 0.062215071064628584,
      "grad_norm": 1.1703389731992098,
      "learning_rate": 2.0732797140303844e-06,
      "loss": 0.0953,
      "step": 232
    },
    {
      "epoch": 0.06248323947438992,
      "grad_norm": 1.1379252307096102,
      "learning_rate": 2.0822162645218946e-06,
      "loss": 0.0646,
      "step": 233
    },
    {
      "epoch": 0.06275140788415125,
      "grad_norm": 1.3492186294879762,
      "learning_rate": 2.091152815013405e-06,
      "loss": 0.0542,
      "step": 234
    },
    {
      "epoch": 0.06301957629391258,
      "grad_norm": 1.1813225197161794,
      "learning_rate": 2.1000893655049154e-06,
      "loss": 0.0794,
      "step": 235
    },
    {
      "epoch": 0.0632877447036739,
      "grad_norm": 1.3245581725610034,
      "learning_rate": 2.1090259159964256e-06,
      "loss": 0.1079,
      "step": 236
    },
    {
      "epoch": 0.06355591311343524,
      "grad_norm": 1.0003102403199535,
      "learning_rate": 2.1179624664879358e-06,
      "loss": 0.0865,
      "step": 237
    },
    {
      "epoch": 0.06382408152319657,
      "grad_norm": 1.2509886633142286,
      "learning_rate": 2.1268990169794464e-06,
      "loss": 0.0952,
      "step": 238
    },
    {
      "epoch": 0.0640922499329579,
      "grad_norm": 1.028817385796426,
      "learning_rate": 2.135835567470956e-06,
      "loss": 0.0844,
      "step": 239
    },
    {
      "epoch": 0.06436041834271923,
      "grad_norm": 1.0010745672393349,
      "learning_rate": 2.1447721179624668e-06,
      "loss": 0.0636,
      "step": 240
    },
    {
      "epoch": 0.06462858675248055,
      "grad_norm": 1.4693215373072745,
      "learning_rate": 2.153708668453977e-06,
      "loss": 0.0929,
      "step": 241
    },
    {
      "epoch": 0.06489675516224189,
      "grad_norm": 1.1922131321695308,
      "learning_rate": 2.162645218945487e-06,
      "loss": 0.0992,
      "step": 242
    },
    {
      "epoch": 0.06516492357200322,
      "grad_norm": 0.8317363838035656,
      "learning_rate": 2.1715817694369974e-06,
      "loss": 0.0686,
      "step": 243
    },
    {
      "epoch": 0.06543309198176454,
      "grad_norm": 0.9917164531210007,
      "learning_rate": 2.1805183199285075e-06,
      "loss": 0.0721,
      "step": 244
    },
    {
      "epoch": 0.06570126039152588,
      "grad_norm": 0.8756151194293332,
      "learning_rate": 2.189454870420018e-06,
      "loss": 0.0918,
      "step": 245
    },
    {
      "epoch": 0.06596942880128721,
      "grad_norm": 0.8521832570277006,
      "learning_rate": 2.1983914209115284e-06,
      "loss": 0.0624,
      "step": 246
    },
    {
      "epoch": 0.06623759721104854,
      "grad_norm": 1.1242564816638847,
      "learning_rate": 2.2073279714030386e-06,
      "loss": 0.0936,
      "step": 247
    },
    {
      "epoch": 0.06650576562080987,
      "grad_norm": 1.2103044921972406,
      "learning_rate": 2.2162645218945488e-06,
      "loss": 0.0618,
      "step": 248
    },
    {
      "epoch": 0.0667739340305712,
      "grad_norm": 0.9597814099444822,
      "learning_rate": 2.2252010723860594e-06,
      "loss": 0.089,
      "step": 249
    },
    {
      "epoch": 0.06704210244033253,
      "grad_norm": 2.5217819195007936,
      "learning_rate": 2.2341376228775696e-06,
      "loss": 0.089,
      "step": 250
    },
    {
      "epoch": 0.06731027085009386,
      "grad_norm": 1.1028408365990845,
      "learning_rate": 2.2430741733690798e-06,
      "loss": 0.0781,
      "step": 251
    },
    {
      "epoch": 0.06757843925985518,
      "grad_norm": 1.084549270598852,
      "learning_rate": 2.25201072386059e-06,
      "loss": 0.0782,
      "step": 252
    },
    {
      "epoch": 0.06784660766961652,
      "grad_norm": 1.1984779264748933,
      "learning_rate": 2.2609472743521e-06,
      "loss": 0.0795,
      "step": 253
    },
    {
      "epoch": 0.06811477607937785,
      "grad_norm": 1.0664352457383879,
      "learning_rate": 2.2698838248436108e-06,
      "loss": 0.0656,
      "step": 254
    },
    {
      "epoch": 0.06838294448913917,
      "grad_norm": 1.2451774925418897,
      "learning_rate": 2.278820375335121e-06,
      "loss": 0.0824,
      "step": 255
    },
    {
      "epoch": 0.06865111289890051,
      "grad_norm": 1.1133994485426701,
      "learning_rate": 2.287756925826631e-06,
      "loss": 0.0771,
      "step": 256
    },
    {
      "epoch": 0.06891928130866184,
      "grad_norm": 0.9239024095356093,
      "learning_rate": 2.2966934763181414e-06,
      "loss": 0.0782,
      "step": 257
    },
    {
      "epoch": 0.06918744971842317,
      "grad_norm": 1.0055580512747866,
      "learning_rate": 2.3056300268096516e-06,
      "loss": 0.0803,
      "step": 258
    },
    {
      "epoch": 0.0694556181281845,
      "grad_norm": 1.1608691448290145,
      "learning_rate": 2.3145665773011617e-06,
      "loss": 0.0586,
      "step": 259
    },
    {
      "epoch": 0.06972378653794584,
      "grad_norm": 1.2539259150831619,
      "learning_rate": 2.3235031277926724e-06,
      "loss": 0.0929,
      "step": 260
    },
    {
      "epoch": 0.06999195494770716,
      "grad_norm": 1.0651457458640197,
      "learning_rate": 2.3324396782841826e-06,
      "loss": 0.0864,
      "step": 261
    },
    {
      "epoch": 0.07026012335746849,
      "grad_norm": 0.6704416714824581,
      "learning_rate": 2.3413762287756928e-06,
      "loss": 0.0652,
      "step": 262
    },
    {
      "epoch": 0.07052829176722981,
      "grad_norm": 0.7900206228074174,
      "learning_rate": 2.350312779267203e-06,
      "loss": 0.0554,
      "step": 263
    },
    {
      "epoch": 0.07079646017699115,
      "grad_norm": 0.9597113694947739,
      "learning_rate": 2.359249329758713e-06,
      "loss": 0.0884,
      "step": 264
    },
    {
      "epoch": 0.07106462858675248,
      "grad_norm": 0.8488229827769775,
      "learning_rate": 2.3681858802502238e-06,
      "loss": 0.0663,
      "step": 265
    },
    {
      "epoch": 0.0713327969965138,
      "grad_norm": 0.7119387481319007,
      "learning_rate": 2.377122430741734e-06,
      "loss": 0.0637,
      "step": 266
    },
    {
      "epoch": 0.07160096540627514,
      "grad_norm": 1.6074978329230172,
      "learning_rate": 2.386058981233244e-06,
      "loss": 0.092,
      "step": 267
    },
    {
      "epoch": 0.07186913381603648,
      "grad_norm": 0.7059400238368266,
      "learning_rate": 2.3949955317247544e-06,
      "loss": 0.0553,
      "step": 268
    },
    {
      "epoch": 0.0721373022257978,
      "grad_norm": 0.8065024141029649,
      "learning_rate": 2.403932082216265e-06,
      "loss": 0.0661,
      "step": 269
    },
    {
      "epoch": 0.07240547063555913,
      "grad_norm": 1.627514462340302,
      "learning_rate": 2.4128686327077747e-06,
      "loss": 0.0793,
      "step": 270
    },
    {
      "epoch": 0.07267363904532047,
      "grad_norm": 0.8530652841457095,
      "learning_rate": 2.4218051831992854e-06,
      "loss": 0.0533,
      "step": 271
    },
    {
      "epoch": 0.07294180745508179,
      "grad_norm": 1.7280767664486572,
      "learning_rate": 2.4307417336907956e-06,
      "loss": 0.1086,
      "step": 272
    },
    {
      "epoch": 0.07320997586484312,
      "grad_norm": 1.1797278482687943,
      "learning_rate": 2.4396782841823058e-06,
      "loss": 0.0888,
      "step": 273
    },
    {
      "epoch": 0.07347814427460445,
      "grad_norm": 1.1738974422848678,
      "learning_rate": 2.4486148346738164e-06,
      "loss": 0.0805,
      "step": 274
    },
    {
      "epoch": 0.07374631268436578,
      "grad_norm": 1.2072795316038485,
      "learning_rate": 2.457551385165326e-06,
      "loss": 0.0908,
      "step": 275
    },
    {
      "epoch": 0.07401448109412712,
      "grad_norm": 1.4159990356977101,
      "learning_rate": 2.4664879356568368e-06,
      "loss": 0.0761,
      "step": 276
    },
    {
      "epoch": 0.07428264950388844,
      "grad_norm": 1.0466463023915136,
      "learning_rate": 2.475424486148347e-06,
      "loss": 0.1091,
      "step": 277
    },
    {
      "epoch": 0.07455081791364977,
      "grad_norm": 1.2665158322129189,
      "learning_rate": 2.484361036639857e-06,
      "loss": 0.0846,
      "step": 278
    },
    {
      "epoch": 0.07481898632341111,
      "grad_norm": 1.0631858461709345,
      "learning_rate": 2.4932975871313673e-06,
      "loss": 0.0821,
      "step": 279
    },
    {
      "epoch": 0.07508715473317243,
      "grad_norm": 1.1710981073280144,
      "learning_rate": 2.5022341376228775e-06,
      "loss": 0.1136,
      "step": 280
    },
    {
      "epoch": 0.07535532314293376,
      "grad_norm": 0.828022804384634,
      "learning_rate": 2.511170688114388e-06,
      "loss": 0.0624,
      "step": 281
    },
    {
      "epoch": 0.0756234915526951,
      "grad_norm": 1.508075132888285,
      "learning_rate": 2.5201072386058984e-06,
      "loss": 0.0821,
      "step": 282
    },
    {
      "epoch": 0.07589165996245642,
      "grad_norm": 0.9022038984688973,
      "learning_rate": 2.5290437890974086e-06,
      "loss": 0.0652,
      "step": 283
    },
    {
      "epoch": 0.07615982837221776,
      "grad_norm": 0.9393358799580087,
      "learning_rate": 2.537980339588919e-06,
      "loss": 0.0849,
      "step": 284
    },
    {
      "epoch": 0.07642799678197908,
      "grad_norm": 0.9690669281510286,
      "learning_rate": 2.5469168900804294e-06,
      "loss": 0.0882,
      "step": 285
    },
    {
      "epoch": 0.07669616519174041,
      "grad_norm": 1.003417487869948,
      "learning_rate": 2.555853440571939e-06,
      "loss": 0.0841,
      "step": 286
    },
    {
      "epoch": 0.07696433360150175,
      "grad_norm": 2.0460151772606867,
      "learning_rate": 2.5647899910634498e-06,
      "loss": 0.1027,
      "step": 287
    },
    {
      "epoch": 0.07723250201126307,
      "grad_norm": 0.7215361864674158,
      "learning_rate": 2.57372654155496e-06,
      "loss": 0.0618,
      "step": 288
    },
    {
      "epoch": 0.0775006704210244,
      "grad_norm": 0.9582439103566321,
      "learning_rate": 2.58266309204647e-06,
      "loss": 0.0747,
      "step": 289
    },
    {
      "epoch": 0.07776883883078574,
      "grad_norm": 1.1565622904192823,
      "learning_rate": 2.5915996425379808e-06,
      "loss": 0.0791,
      "step": 290
    },
    {
      "epoch": 0.07803700724054706,
      "grad_norm": 0.9954087740293288,
      "learning_rate": 2.600536193029491e-06,
      "loss": 0.077,
      "step": 291
    },
    {
      "epoch": 0.0783051756503084,
      "grad_norm": 1.1051014947767286,
      "learning_rate": 2.609472743521001e-06,
      "loss": 0.0694,
      "step": 292
    },
    {
      "epoch": 0.07857334406006973,
      "grad_norm": 1.4335081031224706,
      "learning_rate": 2.6184092940125118e-06,
      "loss": 0.0988,
      "step": 293
    },
    {
      "epoch": 0.07884151246983105,
      "grad_norm": 0.9157223238037455,
      "learning_rate": 2.6273458445040215e-06,
      "loss": 0.0687,
      "step": 294
    },
    {
      "epoch": 0.07910968087959239,
      "grad_norm": 0.9870856447241909,
      "learning_rate": 2.6362823949955317e-06,
      "loss": 0.0783,
      "step": 295
    },
    {
      "epoch": 0.07937784928935371,
      "grad_norm": 1.0884120462342197,
      "learning_rate": 2.645218945487042e-06,
      "loss": 0.0655,
      "step": 296
    },
    {
      "epoch": 0.07964601769911504,
      "grad_norm": 0.8691392738234774,
      "learning_rate": 2.6541554959785526e-06,
      "loss": 0.0605,
      "step": 297
    },
    {
      "epoch": 0.07991418610887638,
      "grad_norm": 1.0147108903119446,
      "learning_rate": 2.6630920464700628e-06,
      "loss": 0.0744,
      "step": 298
    },
    {
      "epoch": 0.0801823545186377,
      "grad_norm": 1.5883804469625813,
      "learning_rate": 2.672028596961573e-06,
      "loss": 0.068,
      "step": 299
    },
    {
      "epoch": 0.08045052292839903,
      "grad_norm": 1.2524670615500189,
      "learning_rate": 2.6809651474530836e-06,
      "loss": 0.0811,
      "step": 300
    },
    {
      "epoch": 0.08071869133816037,
      "grad_norm": 1.0479724605410745,
      "learning_rate": 2.6899016979445938e-06,
      "loss": 0.0561,
      "step": 301
    },
    {
      "epoch": 0.08098685974792169,
      "grad_norm": 1.0859122108575248,
      "learning_rate": 2.6988382484361035e-06,
      "loss": 0.079,
      "step": 302
    },
    {
      "epoch": 0.08125502815768303,
      "grad_norm": 0.9619983404050642,
      "learning_rate": 2.707774798927614e-06,
      "loss": 0.073,
      "step": 303
    },
    {
      "epoch": 0.08152319656744436,
      "grad_norm": 1.307535170062756,
      "learning_rate": 2.7167113494191243e-06,
      "loss": 0.0895,
      "step": 304
    },
    {
      "epoch": 0.08179136497720568,
      "grad_norm": 1.0701695250730898,
      "learning_rate": 2.7256478999106345e-06,
      "loss": 0.0787,
      "step": 305
    },
    {
      "epoch": 0.08205953338696702,
      "grad_norm": 1.0687106263316555,
      "learning_rate": 2.734584450402145e-06,
      "loss": 0.0626,
      "step": 306
    },
    {
      "epoch": 0.08232770179672834,
      "grad_norm": 1.002157793093474,
      "learning_rate": 2.7435210008936554e-06,
      "loss": 0.0781,
      "step": 307
    },
    {
      "epoch": 0.08259587020648967,
      "grad_norm": 0.9378859444023192,
      "learning_rate": 2.7524575513851655e-06,
      "loss": 0.056,
      "step": 308
    },
    {
      "epoch": 0.08286403861625101,
      "grad_norm": 1.685474229427092,
      "learning_rate": 2.761394101876676e-06,
      "loss": 0.0742,
      "step": 309
    },
    {
      "epoch": 0.08313220702601233,
      "grad_norm": 1.1208292505894206,
      "learning_rate": 2.770330652368186e-06,
      "loss": 0.0887,
      "step": 310
    },
    {
      "epoch": 0.08340037543577367,
      "grad_norm": 0.8097955547490224,
      "learning_rate": 2.779267202859696e-06,
      "loss": 0.0541,
      "step": 311
    },
    {
      "epoch": 0.083668543845535,
      "grad_norm": 1.0727064085873945,
      "learning_rate": 2.7882037533512068e-06,
      "loss": 0.0801,
      "step": 312
    },
    {
      "epoch": 0.08393671225529632,
      "grad_norm": 0.8981169038743971,
      "learning_rate": 2.797140303842717e-06,
      "loss": 0.0864,
      "step": 313
    },
    {
      "epoch": 0.08420488066505766,
      "grad_norm": 0.798231015805989,
      "learning_rate": 2.806076854334227e-06,
      "loss": 0.059,
      "step": 314
    },
    {
      "epoch": 0.08447304907481899,
      "grad_norm": 1.1440732517300938,
      "learning_rate": 2.8150134048257378e-06,
      "loss": 0.0903,
      "step": 315
    },
    {
      "epoch": 0.08474121748458031,
      "grad_norm": 1.3570227253510636,
      "learning_rate": 2.823949955317248e-06,
      "loss": 0.0903,
      "step": 316
    },
    {
      "epoch": 0.08500938589434165,
      "grad_norm": 1.3753091739520849,
      "learning_rate": 2.8328865058087577e-06,
      "loss": 0.0993,
      "step": 317
    },
    {
      "epoch": 0.08527755430410297,
      "grad_norm": 0.827931212541233,
      "learning_rate": 2.8418230563002683e-06,
      "loss": 0.0619,
      "step": 318
    },
    {
      "epoch": 0.0855457227138643,
      "grad_norm": 1.4903805164868298,
      "learning_rate": 2.8507596067917785e-06,
      "loss": 0.061,
      "step": 319
    },
    {
      "epoch": 0.08581389112362564,
      "grad_norm": 1.077024687984436,
      "learning_rate": 2.8596961572832887e-06,
      "loss": 0.0742,
      "step": 320
    },
    {
      "epoch": 0.08608205953338696,
      "grad_norm": 0.9191356525563837,
      "learning_rate": 2.8686327077747994e-06,
      "loss": 0.0778,
      "step": 321
    },
    {
      "epoch": 0.0863502279431483,
      "grad_norm": 1.0400915825906893,
      "learning_rate": 2.8775692582663096e-06,
      "loss": 0.0825,
      "step": 322
    },
    {
      "epoch": 0.08661839635290963,
      "grad_norm": 0.9202031895155263,
      "learning_rate": 2.8865058087578197e-06,
      "loss": 0.0686,
      "step": 323
    },
    {
      "epoch": 0.08688656476267095,
      "grad_norm": 1.0230903117879404,
      "learning_rate": 2.8954423592493304e-06,
      "loss": 0.0905,
      "step": 324
    },
    {
      "epoch": 0.08715473317243229,
      "grad_norm": 1.0574152487931676,
      "learning_rate": 2.9043789097408406e-06,
      "loss": 0.0832,
      "step": 325
    },
    {
      "epoch": 0.08742290158219362,
      "grad_norm": 0.8171027114437683,
      "learning_rate": 2.9133154602323503e-06,
      "loss": 0.0657,
      "step": 326
    },
    {
      "epoch": 0.08769106999195495,
      "grad_norm": 1.2485191655184722,
      "learning_rate": 2.9222520107238605e-06,
      "loss": 0.0769,
      "step": 327
    },
    {
      "epoch": 0.08795923840171628,
      "grad_norm": 1.3259720374834723,
      "learning_rate": 2.931188561215371e-06,
      "loss": 0.0891,
      "step": 328
    },
    {
      "epoch": 0.0882274068114776,
      "grad_norm": 2.811216326336966,
      "learning_rate": 2.9401251117068813e-06,
      "loss": 0.0842,
      "step": 329
    },
    {
      "epoch": 0.08849557522123894,
      "grad_norm": 1.1527833879102507,
      "learning_rate": 2.9490616621983915e-06,
      "loss": 0.0912,
      "step": 330
    },
    {
      "epoch": 0.08876374363100027,
      "grad_norm": 1.1658411322197932,
      "learning_rate": 2.957998212689902e-06,
      "loss": 0.0687,
      "step": 331
    },
    {
      "epoch": 0.0890319120407616,
      "grad_norm": 1.081588085132814,
      "learning_rate": 2.9669347631814124e-06,
      "loss": 0.0803,
      "step": 332
    },
    {
      "epoch": 0.08930008045052293,
      "grad_norm": 0.9446185208778421,
      "learning_rate": 2.975871313672922e-06,
      "loss": 0.0601,
      "step": 333
    },
    {
      "epoch": 0.08956824886028426,
      "grad_norm": 1.0997109244581862,
      "learning_rate": 2.9848078641644327e-06,
      "loss": 0.0751,
      "step": 334
    },
    {
      "epoch": 0.08983641727004558,
      "grad_norm": 0.8143012910624685,
      "learning_rate": 2.993744414655943e-06,
      "loss": 0.0738,
      "step": 335
    },
    {
      "epoch": 0.09010458567980692,
      "grad_norm": 1.406278919886631,
      "learning_rate": 3.002680965147453e-06,
      "loss": 0.0919,
      "step": 336
    },
    {
      "epoch": 0.09037275408956826,
      "grad_norm": 1.1249163229623687,
      "learning_rate": 3.0116175156389638e-06,
      "loss": 0.0579,
      "step": 337
    },
    {
      "epoch": 0.09064092249932958,
      "grad_norm": 0.6473890487621132,
      "learning_rate": 3.020554066130474e-06,
      "loss": 0.0499,
      "step": 338
    },
    {
      "epoch": 0.09090909090909091,
      "grad_norm": 1.2174147140056035,
      "learning_rate": 3.029490616621984e-06,
      "loss": 0.0961,
      "step": 339
    },
    {
      "epoch": 0.09117725931885223,
      "grad_norm": 1.2417750502005842,
      "learning_rate": 3.0384271671134948e-06,
      "loss": 0.0753,
      "step": 340
    },
    {
      "epoch": 0.09144542772861357,
      "grad_norm": 1.3856305404094686,
      "learning_rate": 3.0473637176050045e-06,
      "loss": 0.0747,
      "step": 341
    },
    {
      "epoch": 0.0917135961383749,
      "grad_norm": 0.8251725957137185,
      "learning_rate": 3.0563002680965147e-06,
      "loss": 0.0505,
      "step": 342
    },
    {
      "epoch": 0.09198176454813622,
      "grad_norm": 0.9690766891612745,
      "learning_rate": 3.0652368185880253e-06,
      "loss": 0.0779,
      "step": 343
    },
    {
      "epoch": 0.09224993295789756,
      "grad_norm": 0.9426468068913342,
      "learning_rate": 3.0741733690795355e-06,
      "loss": 0.0625,
      "step": 344
    },
    {
      "epoch": 0.0925181013676589,
      "grad_norm": 1.6635687048533288,
      "learning_rate": 3.0831099195710457e-06,
      "loss": 0.076,
      "step": 345
    },
    {
      "epoch": 0.09278626977742022,
      "grad_norm": 1.2042708166020768,
      "learning_rate": 3.0920464700625564e-06,
      "loss": 0.0942,
      "step": 346
    },
    {
      "epoch": 0.09305443818718155,
      "grad_norm": 0.918881700830368,
      "learning_rate": 3.1009830205540666e-06,
      "loss": 0.0557,
      "step": 347
    },
    {
      "epoch": 0.09332260659694289,
      "grad_norm": 1.1206222445934224,
      "learning_rate": 3.1099195710455763e-06,
      "loss": 0.0736,
      "step": 348
    },
    {
      "epoch": 0.09359077500670421,
      "grad_norm": 1.92039125119349,
      "learning_rate": 3.1188561215370874e-06,
      "loss": 0.107,
      "step": 349
    },
    {
      "epoch": 0.09385894341646554,
      "grad_norm": 1.4661266745846624,
      "learning_rate": 3.127792672028597e-06,
      "loss": 0.0917,
      "step": 350
    },
    {
      "epoch": 0.09412711182622686,
      "grad_norm": 1.0335237981658958,
      "learning_rate": 3.1367292225201073e-06,
      "loss": 0.0622,
      "step": 351
    },
    {
      "epoch": 0.0943952802359882,
      "grad_norm": 0.9000440806026753,
      "learning_rate": 3.145665773011618e-06,
      "loss": 0.0568,
      "step": 352
    },
    {
      "epoch": 0.09466344864574953,
      "grad_norm": 0.887741801012921,
      "learning_rate": 3.154602323503128e-06,
      "loss": 0.0502,
      "step": 353
    },
    {
      "epoch": 0.09493161705551086,
      "grad_norm": 0.903705476438776,
      "learning_rate": 3.1635388739946383e-06,
      "loss": 0.0509,
      "step": 354
    },
    {
      "epoch": 0.09519978546527219,
      "grad_norm": 1.1072642212309256,
      "learning_rate": 3.172475424486149e-06,
      "loss": 0.0734,
      "step": 355
    },
    {
      "epoch": 0.09546795387503353,
      "grad_norm": 1.1012522319067872,
      "learning_rate": 3.181411974977659e-06,
      "loss": 0.0731,
      "step": 356
    },
    {
      "epoch": 0.09573612228479485,
      "grad_norm": 0.8976382109675861,
      "learning_rate": 3.190348525469169e-06,
      "loss": 0.0844,
      "step": 357
    },
    {
      "epoch": 0.09600429069455618,
      "grad_norm": 1.6457634228322715,
      "learning_rate": 3.199285075960679e-06,
      "loss": 0.072,
      "step": 358
    },
    {
      "epoch": 0.09627245910431752,
      "grad_norm": 1.3141929485046302,
      "learning_rate": 3.2082216264521897e-06,
      "loss": 0.0797,
      "step": 359
    },
    {
      "epoch": 0.09654062751407884,
      "grad_norm": 0.6857987991249308,
      "learning_rate": 3.2171581769437e-06,
      "loss": 0.0592,
      "step": 360
    },
    {
      "epoch": 0.09680879592384017,
      "grad_norm": 0.9072021830930354,
      "learning_rate": 3.22609472743521e-06,
      "loss": 0.0716,
      "step": 361
    },
    {
      "epoch": 0.0970769643336015,
      "grad_norm": 0.9729303928487002,
      "learning_rate": 3.2350312779267208e-06,
      "loss": 0.0836,
      "step": 362
    },
    {
      "epoch": 0.09734513274336283,
      "grad_norm": 0.8535025577818581,
      "learning_rate": 3.243967828418231e-06,
      "loss": 0.0597,
      "step": 363
    },
    {
      "epoch": 0.09761330115312417,
      "grad_norm": 0.7185008385879719,
      "learning_rate": 3.2529043789097407e-06,
      "loss": 0.0567,
      "step": 364
    },
    {
      "epoch": 0.09788146956288549,
      "grad_norm": 1.4503499628007936,
      "learning_rate": 3.2618409294012513e-06,
      "loss": 0.0913,
      "step": 365
    },
    {
      "epoch": 0.09814963797264682,
      "grad_norm": 1.0550875866772251,
      "learning_rate": 3.2707774798927615e-06,
      "loss": 0.0653,
      "step": 366
    },
    {
      "epoch": 0.09841780638240816,
      "grad_norm": 1.0355247644516965,
      "learning_rate": 3.2797140303842717e-06,
      "loss": 0.091,
      "step": 367
    },
    {
      "epoch": 0.09868597479216948,
      "grad_norm": 0.869526429019359,
      "learning_rate": 3.2886505808757823e-06,
      "loss": 0.0663,
      "step": 368
    },
    {
      "epoch": 0.09895414320193081,
      "grad_norm": 1.0454422901891616,
      "learning_rate": 3.2975871313672925e-06,
      "loss": 0.0685,
      "step": 369
    },
    {
      "epoch": 0.09922231161169215,
      "grad_norm": 1.3344383792030077,
      "learning_rate": 3.3065236818588027e-06,
      "loss": 0.0837,
      "step": 370
    },
    {
      "epoch": 0.09949048002145347,
      "grad_norm": 0.8711190171305349,
      "learning_rate": 3.3154602323503134e-06,
      "loss": 0.0631,
      "step": 371
    },
    {
      "epoch": 0.0997586484312148,
      "grad_norm": 1.8034412169688117,
      "learning_rate": 3.324396782841823e-06,
      "loss": 0.0799,
      "step": 372
    },
    {
      "epoch": 0.10002681684097613,
      "grad_norm": 1.1544396747079537,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 0.0658,
      "step": 373
    },
    {
      "epoch": 0.10029498525073746,
      "grad_norm": 1.0956568837305507,
      "learning_rate": 3.342269883824844e-06,
      "loss": 0.0971,
      "step": 374
    },
    {
      "epoch": 0.1005631536604988,
      "grad_norm": 0.9152152667154431,
      "learning_rate": 3.351206434316354e-06,
      "loss": 0.0699,
      "step": 375
    },
    {
      "epoch": 0.10083132207026012,
      "grad_norm": 1.2003730717751693,
      "learning_rate": 3.3601429848078643e-06,
      "loss": 0.0728,
      "step": 376
    },
    {
      "epoch": 0.10109949048002145,
      "grad_norm": 0.7839119427197051,
      "learning_rate": 3.369079535299375e-06,
      "loss": 0.0609,
      "step": 377
    },
    {
      "epoch": 0.10136765888978279,
      "grad_norm": 0.7857265623885019,
      "learning_rate": 3.378016085790885e-06,
      "loss": 0.0712,
      "step": 378
    },
    {
      "epoch": 0.10163582729954411,
      "grad_norm": 1.1896410448477541,
      "learning_rate": 3.386952636282395e-06,
      "loss": 0.0787,
      "step": 379
    },
    {
      "epoch": 0.10190399570930544,
      "grad_norm": 0.95025320751986,
      "learning_rate": 3.395889186773906e-06,
      "loss": 0.0584,
      "step": 380
    },
    {
      "epoch": 0.10217216411906678,
      "grad_norm": 1.007263048487301,
      "learning_rate": 3.4048257372654157e-06,
      "loss": 0.0589,
      "step": 381
    },
    {
      "epoch": 0.1024403325288281,
      "grad_norm": 1.0622571583865128,
      "learning_rate": 3.413762287756926e-06,
      "loss": 0.1023,
      "step": 382
    },
    {
      "epoch": 0.10270850093858944,
      "grad_norm": 1.5801401474760421,
      "learning_rate": 3.4226988382484365e-06,
      "loss": 0.0691,
      "step": 383
    },
    {
      "epoch": 0.10297666934835076,
      "grad_norm": 0.9180005102082369,
      "learning_rate": 3.4316353887399467e-06,
      "loss": 0.059,
      "step": 384
    },
    {
      "epoch": 0.10324483775811209,
      "grad_norm": 1.3006990037120956,
      "learning_rate": 3.440571939231457e-06,
      "loss": 0.1537,
      "step": 385
    },
    {
      "epoch": 0.10351300616787343,
      "grad_norm": 3.388616644982409,
      "learning_rate": 3.4495084897229676e-06,
      "loss": 0.072,
      "step": 386
    },
    {
      "epoch": 0.10378117457763475,
      "grad_norm": 1.6250064874481471,
      "learning_rate": 3.4584450402144778e-06,
      "loss": 0.0798,
      "step": 387
    },
    {
      "epoch": 0.10404934298739608,
      "grad_norm": 2.2028328340019545,
      "learning_rate": 3.4673815907059875e-06,
      "loss": 0.1086,
      "step": 388
    },
    {
      "epoch": 0.10431751139715742,
      "grad_norm": 0.8735149416116579,
      "learning_rate": 3.4763181411974977e-06,
      "loss": 0.085,
      "step": 389
    },
    {
      "epoch": 0.10458567980691874,
      "grad_norm": 0.8826399197314917,
      "learning_rate": 3.4852546916890083e-06,
      "loss": 0.0777,
      "step": 390
    },
    {
      "epoch": 0.10485384821668008,
      "grad_norm": 1.7772903501906536,
      "learning_rate": 3.4941912421805185e-06,
      "loss": 0.067,
      "step": 391
    },
    {
      "epoch": 0.10512201662644141,
      "grad_norm": 0.6612012304390023,
      "learning_rate": 3.5031277926720287e-06,
      "loss": 0.0555,
      "step": 392
    },
    {
      "epoch": 0.10539018503620273,
      "grad_norm": 1.0992829113760316,
      "learning_rate": 3.5120643431635393e-06,
      "loss": 0.085,
      "step": 393
    },
    {
      "epoch": 0.10565835344596407,
      "grad_norm": 0.9158163317418148,
      "learning_rate": 3.5210008936550495e-06,
      "loss": 0.072,
      "step": 394
    },
    {
      "epoch": 0.10592652185572539,
      "grad_norm": 1.1235386907017653,
      "learning_rate": 3.5299374441465593e-06,
      "loss": 0.078,
      "step": 395
    },
    {
      "epoch": 0.10619469026548672,
      "grad_norm": 0.9295731162362485,
      "learning_rate": 3.53887399463807e-06,
      "loss": 0.0769,
      "step": 396
    },
    {
      "epoch": 0.10646285867524806,
      "grad_norm": 0.8189493812836391,
      "learning_rate": 3.54781054512958e-06,
      "loss": 0.0766,
      "step": 397
    },
    {
      "epoch": 0.10673102708500938,
      "grad_norm": 0.7264464833157396,
      "learning_rate": 3.5567470956210903e-06,
      "loss": 0.0625,
      "step": 398
    },
    {
      "epoch": 0.10699919549477072,
      "grad_norm": 0.9336534372125695,
      "learning_rate": 3.565683646112601e-06,
      "loss": 0.0642,
      "step": 399
    },
    {
      "epoch": 0.10726736390453205,
      "grad_norm": 0.8082667028332312,
      "learning_rate": 3.574620196604111e-06,
      "loss": 0.0762,
      "step": 400
    },
    {
      "epoch": 0.10753553231429337,
      "grad_norm": 1.1965741254330413,
      "learning_rate": 3.5835567470956213e-06,
      "loss": 0.0851,
      "step": 401
    },
    {
      "epoch": 0.10780370072405471,
      "grad_norm": 0.7280924257260993,
      "learning_rate": 3.592493297587132e-06,
      "loss": 0.0592,
      "step": 402
    },
    {
      "epoch": 0.10807186913381604,
      "grad_norm": 0.720547204956332,
      "learning_rate": 3.6014298480786417e-06,
      "loss": 0.054,
      "step": 403
    },
    {
      "epoch": 0.10834003754357736,
      "grad_norm": 1.3765699071733941,
      "learning_rate": 3.610366398570152e-06,
      "loss": 0.0634,
      "step": 404
    },
    {
      "epoch": 0.1086082059533387,
      "grad_norm": 1.1563674924199965,
      "learning_rate": 3.6193029490616625e-06,
      "loss": 0.0729,
      "step": 405
    },
    {
      "epoch": 0.10887637436310002,
      "grad_norm": 0.8671488689492501,
      "learning_rate": 3.6282394995531727e-06,
      "loss": 0.0592,
      "step": 406
    },
    {
      "epoch": 0.10914454277286136,
      "grad_norm": 0.7925270272445619,
      "learning_rate": 3.637176050044683e-06,
      "loss": 0.068,
      "step": 407
    },
    {
      "epoch": 0.10941271118262269,
      "grad_norm": 2.7992708918294036,
      "learning_rate": 3.6461126005361935e-06,
      "loss": 0.091,
      "step": 408
    },
    {
      "epoch": 0.10968087959238401,
      "grad_norm": 0.8933122734105511,
      "learning_rate": 3.6550491510277037e-06,
      "loss": 0.0639,
      "step": 409
    },
    {
      "epoch": 0.10994904800214535,
      "grad_norm": 0.9735053304309693,
      "learning_rate": 3.663985701519214e-06,
      "loss": 0.0692,
      "step": 410
    },
    {
      "epoch": 0.11021721641190668,
      "grad_norm": 1.1329451443201408,
      "learning_rate": 3.6729222520107246e-06,
      "loss": 0.0608,
      "step": 411
    },
    {
      "epoch": 0.110485384821668,
      "grad_norm": 1.0833987010340869,
      "learning_rate": 3.6818588025022343e-06,
      "loss": 0.082,
      "step": 412
    },
    {
      "epoch": 0.11075355323142934,
      "grad_norm": 0.6075048864304109,
      "learning_rate": 3.6907953529937445e-06,
      "loss": 0.0572,
      "step": 413
    },
    {
      "epoch": 0.11102172164119067,
      "grad_norm": 0.8820252585041759,
      "learning_rate": 3.699731903485255e-06,
      "loss": 0.0735,
      "step": 414
    },
    {
      "epoch": 0.111289890050952,
      "grad_norm": 0.8103064957156563,
      "learning_rate": 3.7086684539767653e-06,
      "loss": 0.0769,
      "step": 415
    },
    {
      "epoch": 0.11155805846071333,
      "grad_norm": 0.9355080913163316,
      "learning_rate": 3.7176050044682755e-06,
      "loss": 0.0916,
      "step": 416
    },
    {
      "epoch": 0.11182622687047465,
      "grad_norm": 1.1030965188188988,
      "learning_rate": 3.726541554959786e-06,
      "loss": 0.0849,
      "step": 417
    },
    {
      "epoch": 0.11209439528023599,
      "grad_norm": 0.8955547579110715,
      "learning_rate": 3.7354781054512963e-06,
      "loss": 0.07,
      "step": 418
    },
    {
      "epoch": 0.11236256368999732,
      "grad_norm": 1.2804628803781006,
      "learning_rate": 3.744414655942806e-06,
      "loss": 0.0712,
      "step": 419
    },
    {
      "epoch": 0.11263073209975864,
      "grad_norm": 0.973257227307041,
      "learning_rate": 3.7533512064343163e-06,
      "loss": 0.0651,
      "step": 420
    },
    {
      "epoch": 0.11289890050951998,
      "grad_norm": 0.9144714758455673,
      "learning_rate": 3.762287756925827e-06,
      "loss": 0.0772,
      "step": 421
    },
    {
      "epoch": 0.11316706891928131,
      "grad_norm": 0.8811434862280502,
      "learning_rate": 3.771224307417337e-06,
      "loss": 0.0708,
      "step": 422
    },
    {
      "epoch": 0.11343523732904263,
      "grad_norm": 0.9275137033365234,
      "learning_rate": 3.7801608579088473e-06,
      "loss": 0.0734,
      "step": 423
    },
    {
      "epoch": 0.11370340573880397,
      "grad_norm": 0.5946467810079347,
      "learning_rate": 3.789097408400358e-06,
      "loss": 0.0509,
      "step": 424
    },
    {
      "epoch": 0.1139715741485653,
      "grad_norm": 1.0392142809650076,
      "learning_rate": 3.798033958891868e-06,
      "loss": 0.0669,
      "step": 425
    },
    {
      "epoch": 0.11423974255832663,
      "grad_norm": 0.9945808768720859,
      "learning_rate": 3.806970509383378e-06,
      "loss": 0.0813,
      "step": 426
    },
    {
      "epoch": 0.11450791096808796,
      "grad_norm": 1.252642635051418,
      "learning_rate": 3.815907059874889e-06,
      "loss": 0.0625,
      "step": 427
    },
    {
      "epoch": 0.11477607937784928,
      "grad_norm": 0.8750667824002176,
      "learning_rate": 3.824843610366399e-06,
      "loss": 0.0613,
      "step": 428
    },
    {
      "epoch": 0.11504424778761062,
      "grad_norm": 1.1042134673035502,
      "learning_rate": 3.833780160857909e-06,
      "loss": 0.0824,
      "step": 429
    },
    {
      "epoch": 0.11531241619737195,
      "grad_norm": 1.055246599454212,
      "learning_rate": 3.8427167113494195e-06,
      "loss": 0.0685,
      "step": 430
    },
    {
      "epoch": 0.11558058460713327,
      "grad_norm": 0.9600247518426471,
      "learning_rate": 3.85165326184093e-06,
      "loss": 0.0642,
      "step": 431
    },
    {
      "epoch": 0.11584875301689461,
      "grad_norm": 1.5056101596788642,
      "learning_rate": 3.86058981233244e-06,
      "loss": 0.0826,
      "step": 432
    },
    {
      "epoch": 0.11611692142665594,
      "grad_norm": 1.021655401782829,
      "learning_rate": 3.86952636282395e-06,
      "loss": 0.0774,
      "step": 433
    },
    {
      "epoch": 0.11638508983641727,
      "grad_norm": 1.6619518501251627,
      "learning_rate": 3.87846291331546e-06,
      "loss": 0.0971,
      "step": 434
    },
    {
      "epoch": 0.1166532582461786,
      "grad_norm": 1.1813479577991888,
      "learning_rate": 3.8873994638069705e-06,
      "loss": 0.07,
      "step": 435
    },
    {
      "epoch": 0.11692142665593994,
      "grad_norm": 0.786445758749673,
      "learning_rate": 3.8963360142984816e-06,
      "loss": 0.0585,
      "step": 436
    },
    {
      "epoch": 0.11718959506570126,
      "grad_norm": 0.6088598052153829,
      "learning_rate": 3.905272564789992e-06,
      "loss": 0.0428,
      "step": 437
    },
    {
      "epoch": 0.11745776347546259,
      "grad_norm": 1.468539965074711,
      "learning_rate": 3.914209115281501e-06,
      "loss": 0.0732,
      "step": 438
    },
    {
      "epoch": 0.11772593188522391,
      "grad_norm": 2.6555324733651844,
      "learning_rate": 3.923145665773012e-06,
      "loss": 0.0964,
      "step": 439
    },
    {
      "epoch": 0.11799410029498525,
      "grad_norm": 1.3817248719686481,
      "learning_rate": 3.932082216264522e-06,
      "loss": 0.0964,
      "step": 440
    },
    {
      "epoch": 0.11826226870474658,
      "grad_norm": 1.6721245521382098,
      "learning_rate": 3.9410187667560325e-06,
      "loss": 0.0679,
      "step": 441
    },
    {
      "epoch": 0.1185304371145079,
      "grad_norm": 0.8951246377469109,
      "learning_rate": 3.949955317247543e-06,
      "loss": 0.0733,
      "step": 442
    },
    {
      "epoch": 0.11879860552426924,
      "grad_norm": 0.9157102904250533,
      "learning_rate": 3.958891867739053e-06,
      "loss": 0.0537,
      "step": 443
    },
    {
      "epoch": 0.11906677393403058,
      "grad_norm": 0.9833019540438356,
      "learning_rate": 3.967828418230563e-06,
      "loss": 0.0861,
      "step": 444
    },
    {
      "epoch": 0.1193349423437919,
      "grad_norm": 0.9813063291336864,
      "learning_rate": 3.976764968722074e-06,
      "loss": 0.0858,
      "step": 445
    },
    {
      "epoch": 0.11960311075355323,
      "grad_norm": 1.0712971937769715,
      "learning_rate": 3.985701519213584e-06,
      "loss": 0.0908,
      "step": 446
    },
    {
      "epoch": 0.11987127916331457,
      "grad_norm": 1.0460466685790129,
      "learning_rate": 3.994638069705094e-06,
      "loss": 0.0655,
      "step": 447
    },
    {
      "epoch": 0.12013944757307589,
      "grad_norm": 0.9771270953154231,
      "learning_rate": 4.003574620196605e-06,
      "loss": 0.0674,
      "step": 448
    },
    {
      "epoch": 0.12040761598283722,
      "grad_norm": 1.1754193477088417,
      "learning_rate": 4.012511170688115e-06,
      "loss": 0.0776,
      "step": 449
    },
    {
      "epoch": 0.12067578439259855,
      "grad_norm": 0.8875512805066087,
      "learning_rate": 4.021447721179625e-06,
      "loss": 0.0824,
      "step": 450
    },
    {
      "epoch": 0.12094395280235988,
      "grad_norm": 1.4439920003823963,
      "learning_rate": 4.030384271671135e-06,
      "loss": 0.0794,
      "step": 451
    },
    {
      "epoch": 0.12121212121212122,
      "grad_norm": 1.0877772359601794,
      "learning_rate": 4.0393208221626455e-06,
      "loss": 0.0893,
      "step": 452
    },
    {
      "epoch": 0.12148028962188254,
      "grad_norm": 0.8399015261876391,
      "learning_rate": 4.048257372654156e-06,
      "loss": 0.0597,
      "step": 453
    },
    {
      "epoch": 0.12174845803164387,
      "grad_norm": 0.9154618737483778,
      "learning_rate": 4.057193923145666e-06,
      "loss": 0.0811,
      "step": 454
    },
    {
      "epoch": 0.12201662644140521,
      "grad_norm": 0.7294910575903597,
      "learning_rate": 4.066130473637176e-06,
      "loss": 0.0583,
      "step": 455
    },
    {
      "epoch": 0.12228479485116653,
      "grad_norm": 0.8192999514974959,
      "learning_rate": 4.075067024128686e-06,
      "loss": 0.0711,
      "step": 456
    },
    {
      "epoch": 0.12255296326092786,
      "grad_norm": 1.0804038523900057,
      "learning_rate": 4.0840035746201965e-06,
      "loss": 0.0799,
      "step": 457
    },
    {
      "epoch": 0.1228211316706892,
      "grad_norm": 1.18231349058715,
      "learning_rate": 4.0929401251117075e-06,
      "loss": 0.0759,
      "step": 458
    },
    {
      "epoch": 0.12308930008045052,
      "grad_norm": 1.0971368765685716,
      "learning_rate": 4.101876675603218e-06,
      "loss": 0.0921,
      "step": 459
    },
    {
      "epoch": 0.12335746849021186,
      "grad_norm": 0.992674402696658,
      "learning_rate": 4.110813226094728e-06,
      "loss": 0.0683,
      "step": 460
    },
    {
      "epoch": 0.12362563689997318,
      "grad_norm": 0.9673134781586926,
      "learning_rate": 4.119749776586238e-06,
      "loss": 0.0846,
      "step": 461
    },
    {
      "epoch": 0.12389380530973451,
      "grad_norm": 0.9742752411950797,
      "learning_rate": 4.128686327077748e-06,
      "loss": 0.0733,
      "step": 462
    },
    {
      "epoch": 0.12416197371949585,
      "grad_norm": 1.0635529993593695,
      "learning_rate": 4.1376228775692585e-06,
      "loss": 0.0775,
      "step": 463
    },
    {
      "epoch": 0.12443014212925717,
      "grad_norm": 1.0650182732480298,
      "learning_rate": 4.146559428060769e-06,
      "loss": 0.0642,
      "step": 464
    },
    {
      "epoch": 0.1246983105390185,
      "grad_norm": 0.9293667582643371,
      "learning_rate": 4.155495978552279e-06,
      "loss": 0.0573,
      "step": 465
    },
    {
      "epoch": 0.12496647894877984,
      "grad_norm": 1.0126793969591357,
      "learning_rate": 4.164432529043789e-06,
      "loss": 0.074,
      "step": 466
    },
    {
      "epoch": 0.12523464735854117,
      "grad_norm": 0.8200191311995798,
      "learning_rate": 4.1733690795353e-06,
      "loss": 0.0634,
      "step": 467
    },
    {
      "epoch": 0.1255028157683025,
      "grad_norm": 1.1820205903006376,
      "learning_rate": 4.18230563002681e-06,
      "loss": 0.0814,
      "step": 468
    },
    {
      "epoch": 0.12577098417806382,
      "grad_norm": 0.8882258356911072,
      "learning_rate": 4.19124218051832e-06,
      "loss": 0.0757,
      "step": 469
    },
    {
      "epoch": 0.12603915258782517,
      "grad_norm": 0.6749977100624934,
      "learning_rate": 4.200178731009831e-06,
      "loss": 0.0554,
      "step": 470
    },
    {
      "epoch": 0.1263073209975865,
      "grad_norm": 0.9624656027661049,
      "learning_rate": 4.209115281501341e-06,
      "loss": 0.0879,
      "step": 471
    },
    {
      "epoch": 0.1265754894073478,
      "grad_norm": 1.1434353268458874,
      "learning_rate": 4.218051831992851e-06,
      "loss": 0.092,
      "step": 472
    },
    {
      "epoch": 0.12684365781710916,
      "grad_norm": 0.9015544704914815,
      "learning_rate": 4.226988382484361e-06,
      "loss": 0.0686,
      "step": 473
    },
    {
      "epoch": 0.12711182622687048,
      "grad_norm": 1.3229621969737553,
      "learning_rate": 4.2359249329758715e-06,
      "loss": 0.0768,
      "step": 474
    },
    {
      "epoch": 0.1273799946366318,
      "grad_norm": 0.914331834017034,
      "learning_rate": 4.244861483467382e-06,
      "loss": 0.0763,
      "step": 475
    },
    {
      "epoch": 0.12764816304639315,
      "grad_norm": 0.8331861325707116,
      "learning_rate": 4.253798033958893e-06,
      "loss": 0.066,
      "step": 476
    },
    {
      "epoch": 0.12791633145615447,
      "grad_norm": 1.2165336558394069,
      "learning_rate": 4.262734584450403e-06,
      "loss": 0.0828,
      "step": 477
    },
    {
      "epoch": 0.1281844998659158,
      "grad_norm": 0.9932826529136056,
      "learning_rate": 4.271671134941912e-06,
      "loss": 0.0844,
      "step": 478
    },
    {
      "epoch": 0.1284526682756771,
      "grad_norm": 1.0675943501939562,
      "learning_rate": 4.280607685433423e-06,
      "loss": 0.1115,
      "step": 479
    },
    {
      "epoch": 0.12872083668543846,
      "grad_norm": 0.8987343897708248,
      "learning_rate": 4.2895442359249335e-06,
      "loss": 0.0697,
      "step": 480
    },
    {
      "epoch": 0.12898900509519978,
      "grad_norm": 0.815139970203552,
      "learning_rate": 4.298480786416444e-06,
      "loss": 0.0727,
      "step": 481
    },
    {
      "epoch": 0.1292571735049611,
      "grad_norm": 0.910883252574242,
      "learning_rate": 4.307417336907954e-06,
      "loss": 0.0628,
      "step": 482
    },
    {
      "epoch": 0.12952534191472245,
      "grad_norm": 1.3298315588571508,
      "learning_rate": 4.316353887399464e-06,
      "loss": 0.0998,
      "step": 483
    },
    {
      "epoch": 0.12979351032448377,
      "grad_norm": 0.5751987297929497,
      "learning_rate": 4.325290437890974e-06,
      "loss": 0.0514,
      "step": 484
    },
    {
      "epoch": 0.1300616787342451,
      "grad_norm": 0.9025309733786545,
      "learning_rate": 4.3342269883824845e-06,
      "loss": 0.0441,
      "step": 485
    },
    {
      "epoch": 0.13032984714400644,
      "grad_norm": 0.9592140461702673,
      "learning_rate": 4.343163538873995e-06,
      "loss": 0.0764,
      "step": 486
    },
    {
      "epoch": 0.13059801555376777,
      "grad_norm": 0.9915824231527102,
      "learning_rate": 4.352100089365505e-06,
      "loss": 0.0838,
      "step": 487
    },
    {
      "epoch": 0.1308661839635291,
      "grad_norm": 0.8148280108622022,
      "learning_rate": 4.361036639857015e-06,
      "loss": 0.0652,
      "step": 488
    },
    {
      "epoch": 0.13113435237329044,
      "grad_norm": 0.8110114865775023,
      "learning_rate": 4.369973190348526e-06,
      "loss": 0.0731,
      "step": 489
    },
    {
      "epoch": 0.13140252078305176,
      "grad_norm": 0.9382468491421662,
      "learning_rate": 4.378909740840036e-06,
      "loss": 0.0788,
      "step": 490
    },
    {
      "epoch": 0.13167068919281308,
      "grad_norm": 0.669865933683254,
      "learning_rate": 4.3878462913315465e-06,
      "loss": 0.0508,
      "step": 491
    },
    {
      "epoch": 0.13193885760257443,
      "grad_norm": 1.0548767545832722,
      "learning_rate": 4.396782841823057e-06,
      "loss": 0.0833,
      "step": 492
    },
    {
      "epoch": 0.13220702601233575,
      "grad_norm": 0.9310081694817219,
      "learning_rate": 4.405719392314567e-06,
      "loss": 0.0563,
      "step": 493
    },
    {
      "epoch": 0.13247519442209707,
      "grad_norm": 1.4873241864686493,
      "learning_rate": 4.414655942806077e-06,
      "loss": 0.0906,
      "step": 494
    },
    {
      "epoch": 0.13274336283185842,
      "grad_norm": 0.7466982007412089,
      "learning_rate": 4.423592493297587e-06,
      "loss": 0.066,
      "step": 495
    },
    {
      "epoch": 0.13301153124161974,
      "grad_norm": 6.549921233618365,
      "learning_rate": 4.4325290437890975e-06,
      "loss": 0.0849,
      "step": 496
    },
    {
      "epoch": 0.13327969965138106,
      "grad_norm": 1.1850852344843947,
      "learning_rate": 4.441465594280608e-06,
      "loss": 0.1006,
      "step": 497
    },
    {
      "epoch": 0.1335478680611424,
      "grad_norm": 0.9599501906156346,
      "learning_rate": 4.450402144772119e-06,
      "loss": 0.071,
      "step": 498
    },
    {
      "epoch": 0.13381603647090373,
      "grad_norm": 0.6613122372374796,
      "learning_rate": 4.459338695263629e-06,
      "loss": 0.0511,
      "step": 499
    },
    {
      "epoch": 0.13408420488066505,
      "grad_norm": 0.7274558616569075,
      "learning_rate": 4.468275245755139e-06,
      "loss": 0.0569,
      "step": 500
    },
    {
      "epoch": 0.13435237329042637,
      "grad_norm": 0.6553914130475342,
      "learning_rate": 4.477211796246649e-06,
      "loss": 0.0444,
      "step": 501
    },
    {
      "epoch": 0.13462054170018772,
      "grad_norm": 0.9214035952956035,
      "learning_rate": 4.4861483467381595e-06,
      "loss": 0.0952,
      "step": 502
    },
    {
      "epoch": 0.13488871010994904,
      "grad_norm": 0.6800480855024846,
      "learning_rate": 4.49508489722967e-06,
      "loss": 0.0586,
      "step": 503
    },
    {
      "epoch": 0.13515687851971037,
      "grad_norm": 0.9687928078495497,
      "learning_rate": 4.50402144772118e-06,
      "loss": 0.0818,
      "step": 504
    },
    {
      "epoch": 0.13542504692947172,
      "grad_norm": 1.0448640684308557,
      "learning_rate": 4.51295799821269e-06,
      "loss": 0.0924,
      "step": 505
    },
    {
      "epoch": 0.13569321533923304,
      "grad_norm": 0.858131542510142,
      "learning_rate": 4.5218945487042e-06,
      "loss": 0.0663,
      "step": 506
    },
    {
      "epoch": 0.13596138374899436,
      "grad_norm": 0.8436086147946624,
      "learning_rate": 4.530831099195711e-06,
      "loss": 0.0667,
      "step": 507
    },
    {
      "epoch": 0.1362295521587557,
      "grad_norm": 2.1066669524317843,
      "learning_rate": 4.5397676496872215e-06,
      "loss": 0.1133,
      "step": 508
    },
    {
      "epoch": 0.13649772056851703,
      "grad_norm": 0.9183759144866963,
      "learning_rate": 4.548704200178731e-06,
      "loss": 0.075,
      "step": 509
    },
    {
      "epoch": 0.13676588897827835,
      "grad_norm": 0.8226875570281615,
      "learning_rate": 4.557640750670242e-06,
      "loss": 0.0745,
      "step": 510
    },
    {
      "epoch": 0.1370340573880397,
      "grad_norm": 0.8715530458420914,
      "learning_rate": 4.566577301161752e-06,
      "loss": 0.0669,
      "step": 511
    },
    {
      "epoch": 0.13730222579780102,
      "grad_norm": 0.9925259115116932,
      "learning_rate": 4.575513851653262e-06,
      "loss": 0.0657,
      "step": 512
    },
    {
      "epoch": 0.13757039420756234,
      "grad_norm": 0.6546855538953195,
      "learning_rate": 4.5844504021447725e-06,
      "loss": 0.0513,
      "step": 513
    },
    {
      "epoch": 0.1378385626173237,
      "grad_norm": 1.350478732524976,
      "learning_rate": 4.593386952636283e-06,
      "loss": 0.075,
      "step": 514
    },
    {
      "epoch": 0.138106731027085,
      "grad_norm": 0.8933342582644004,
      "learning_rate": 4.602323503127793e-06,
      "loss": 0.0675,
      "step": 515
    },
    {
      "epoch": 0.13837489943684633,
      "grad_norm": 1.0202490110975357,
      "learning_rate": 4.611260053619303e-06,
      "loss": 0.0746,
      "step": 516
    },
    {
      "epoch": 0.13864306784660768,
      "grad_norm": 0.7005508485761252,
      "learning_rate": 4.620196604110813e-06,
      "loss": 0.0618,
      "step": 517
    },
    {
      "epoch": 0.138911236256369,
      "grad_norm": 0.7109384080894464,
      "learning_rate": 4.6291331546023235e-06,
      "loss": 0.0605,
      "step": 518
    },
    {
      "epoch": 0.13917940466613032,
      "grad_norm": 0.9563738280846782,
      "learning_rate": 4.638069705093834e-06,
      "loss": 0.0783,
      "step": 519
    },
    {
      "epoch": 0.13944757307589167,
      "grad_norm": 1.3636375020852676,
      "learning_rate": 4.647006255585345e-06,
      "loss": 0.0881,
      "step": 520
    },
    {
      "epoch": 0.139715741485653,
      "grad_norm": 0.7255393317443555,
      "learning_rate": 4.655942806076855e-06,
      "loss": 0.0593,
      "step": 521
    },
    {
      "epoch": 0.13998390989541432,
      "grad_norm": 0.6918089214278179,
      "learning_rate": 4.664879356568365e-06,
      "loss": 0.0733,
      "step": 522
    },
    {
      "epoch": 0.14025207830517564,
      "grad_norm": 0.8351970419627537,
      "learning_rate": 4.673815907059875e-06,
      "loss": 0.0842,
      "step": 523
    },
    {
      "epoch": 0.14052024671493699,
      "grad_norm": 1.0642601386011878,
      "learning_rate": 4.6827524575513855e-06,
      "loss": 0.0822,
      "step": 524
    },
    {
      "epoch": 0.1407884151246983,
      "grad_norm": 0.8818098211601748,
      "learning_rate": 4.691689008042896e-06,
      "loss": 0.0704,
      "step": 525
    },
    {
      "epoch": 0.14105658353445963,
      "grad_norm": 0.7943586043249317,
      "learning_rate": 4.700625558534406e-06,
      "loss": 0.0695,
      "step": 526
    },
    {
      "epoch": 0.14132475194422098,
      "grad_norm": 0.6536497460148595,
      "learning_rate": 4.709562109025916e-06,
      "loss": 0.056,
      "step": 527
    },
    {
      "epoch": 0.1415929203539823,
      "grad_norm": 1.5667849774609017,
      "learning_rate": 4.718498659517426e-06,
      "loss": 0.0924,
      "step": 528
    },
    {
      "epoch": 0.14186108876374362,
      "grad_norm": 0.7565027568868655,
      "learning_rate": 4.727435210008937e-06,
      "loss": 0.0701,
      "step": 529
    },
    {
      "epoch": 0.14212925717350497,
      "grad_norm": 1.0077783748022413,
      "learning_rate": 4.7363717605004475e-06,
      "loss": 0.0751,
      "step": 530
    },
    {
      "epoch": 0.1423974255832663,
      "grad_norm": 0.7372614052460006,
      "learning_rate": 4.745308310991958e-06,
      "loss": 0.0651,
      "step": 531
    },
    {
      "epoch": 0.1426655939930276,
      "grad_norm": 0.8179854580683265,
      "learning_rate": 4.754244861483468e-06,
      "loss": 0.0624,
      "step": 532
    },
    {
      "epoch": 0.14293376240278896,
      "grad_norm": 0.7921806357668186,
      "learning_rate": 4.763181411974978e-06,
      "loss": 0.0654,
      "step": 533
    },
    {
      "epoch": 0.14320193081255028,
      "grad_norm": 1.0094717121856216,
      "learning_rate": 4.772117962466488e-06,
      "loss": 0.0778,
      "step": 534
    },
    {
      "epoch": 0.1434700992223116,
      "grad_norm": 1.1648539047305473,
      "learning_rate": 4.7810545129579985e-06,
      "loss": 0.0986,
      "step": 535
    },
    {
      "epoch": 0.14373826763207295,
      "grad_norm": 0.6223026277966669,
      "learning_rate": 4.789991063449509e-06,
      "loss": 0.0519,
      "step": 536
    },
    {
      "epoch": 0.14400643604183427,
      "grad_norm": 0.6586799327475868,
      "learning_rate": 4.798927613941019e-06,
      "loss": 0.0488,
      "step": 537
    },
    {
      "epoch": 0.1442746044515956,
      "grad_norm": 0.709980287852735,
      "learning_rate": 4.80786416443253e-06,
      "loss": 0.0461,
      "step": 538
    },
    {
      "epoch": 0.14454277286135694,
      "grad_norm": 1.4837721203925192,
      "learning_rate": 4.81680071492404e-06,
      "loss": 0.0874,
      "step": 539
    },
    {
      "epoch": 0.14481094127111827,
      "grad_norm": 0.7866693482783739,
      "learning_rate": 4.8257372654155495e-06,
      "loss": 0.0575,
      "step": 540
    },
    {
      "epoch": 0.1450791096808796,
      "grad_norm": 0.7167748661541813,
      "learning_rate": 4.8346738159070605e-06,
      "loss": 0.0674,
      "step": 541
    },
    {
      "epoch": 0.14534727809064094,
      "grad_norm": 0.8462443866200521,
      "learning_rate": 4.843610366398571e-06,
      "loss": 0.0589,
      "step": 542
    },
    {
      "epoch": 0.14561544650040226,
      "grad_norm": 0.6643656764470212,
      "learning_rate": 4.852546916890081e-06,
      "loss": 0.056,
      "step": 543
    },
    {
      "epoch": 0.14588361491016358,
      "grad_norm": 0.8552434165874294,
      "learning_rate": 4.861483467381591e-06,
      "loss": 0.0823,
      "step": 544
    },
    {
      "epoch": 0.1461517833199249,
      "grad_norm": 0.8903376388832913,
      "learning_rate": 4.870420017873101e-06,
      "loss": 0.07,
      "step": 545
    },
    {
      "epoch": 0.14641995172968625,
      "grad_norm": 0.8353595510018919,
      "learning_rate": 4.8793565683646115e-06,
      "loss": 0.0658,
      "step": 546
    },
    {
      "epoch": 0.14668812013944757,
      "grad_norm": 0.8215287733778167,
      "learning_rate": 4.888293118856122e-06,
      "loss": 0.0593,
      "step": 547
    },
    {
      "epoch": 0.1469562885492089,
      "grad_norm": 1.000115285057022,
      "learning_rate": 4.897229669347633e-06,
      "loss": 0.0765,
      "step": 548
    },
    {
      "epoch": 0.14722445695897024,
      "grad_norm": 0.7405585700170938,
      "learning_rate": 4.906166219839142e-06,
      "loss": 0.0501,
      "step": 549
    },
    {
      "epoch": 0.14749262536873156,
      "grad_norm": 0.9142307742236662,
      "learning_rate": 4.915102770330652e-06,
      "loss": 0.0592,
      "step": 550
    },
    {
      "epoch": 0.14776079377849288,
      "grad_norm": 0.6160805294173842,
      "learning_rate": 4.924039320822163e-06,
      "loss": 0.0491,
      "step": 551
    },
    {
      "epoch": 0.14802896218825423,
      "grad_norm": 1.1391281691417314,
      "learning_rate": 4.9329758713136735e-06,
      "loss": 0.0614,
      "step": 552
    },
    {
      "epoch": 0.14829713059801555,
      "grad_norm": 0.616929305519688,
      "learning_rate": 4.941912421805184e-06,
      "loss": 0.0499,
      "step": 553
    },
    {
      "epoch": 0.14856529900777687,
      "grad_norm": 0.8509743989261174,
      "learning_rate": 4.950848972296694e-06,
      "loss": 0.0733,
      "step": 554
    },
    {
      "epoch": 0.14883346741753822,
      "grad_norm": 0.7293632113916422,
      "learning_rate": 4.959785522788204e-06,
      "loss": 0.0756,
      "step": 555
    },
    {
      "epoch": 0.14910163582729954,
      "grad_norm": 0.8819254466979063,
      "learning_rate": 4.968722073279714e-06,
      "loss": 0.0848,
      "step": 556
    },
    {
      "epoch": 0.14936980423706087,
      "grad_norm": 0.8080837803424797,
      "learning_rate": 4.9776586237712245e-06,
      "loss": 0.0812,
      "step": 557
    },
    {
      "epoch": 0.14963797264682221,
      "grad_norm": 0.7911275831913056,
      "learning_rate": 4.986595174262735e-06,
      "loss": 0.0644,
      "step": 558
    },
    {
      "epoch": 0.14990614105658354,
      "grad_norm": 0.47038585896765805,
      "learning_rate": 4.995531724754245e-06,
      "loss": 0.0395,
      "step": 559
    },
    {
      "epoch": 0.15017430946634486,
      "grad_norm": 0.8369747807835302,
      "learning_rate": 5.004468275245755e-06,
      "loss": 0.079,
      "step": 560
    },
    {
      "epoch": 0.1504424778761062,
      "grad_norm": 1.0874992943514181,
      "learning_rate": 5.013404825737266e-06,
      "loss": 0.0849,
      "step": 561
    },
    {
      "epoch": 0.15071064628586753,
      "grad_norm": 0.733063815926613,
      "learning_rate": 5.022341376228776e-06,
      "loss": 0.0822,
      "step": 562
    },
    {
      "epoch": 0.15097881469562885,
      "grad_norm": 0.7077567305270279,
      "learning_rate": 5.031277926720286e-06,
      "loss": 0.0612,
      "step": 563
    },
    {
      "epoch": 0.1512469831053902,
      "grad_norm": 1.2505369073509383,
      "learning_rate": 5.040214477211797e-06,
      "loss": 0.0811,
      "step": 564
    },
    {
      "epoch": 0.15151515151515152,
      "grad_norm": 0.9800800107631583,
      "learning_rate": 5.049151027703308e-06,
      "loss": 0.057,
      "step": 565
    },
    {
      "epoch": 0.15178331992491284,
      "grad_norm": 0.7504456722153466,
      "learning_rate": 5.058087578194817e-06,
      "loss": 0.0525,
      "step": 566
    },
    {
      "epoch": 0.15205148833467416,
      "grad_norm": 0.6270440486099853,
      "learning_rate": 5.067024128686327e-06,
      "loss": 0.0506,
      "step": 567
    },
    {
      "epoch": 0.1523196567444355,
      "grad_norm": 1.1128116960948156,
      "learning_rate": 5.075960679177838e-06,
      "loss": 0.0722,
      "step": 568
    },
    {
      "epoch": 0.15258782515419683,
      "grad_norm": 0.9071547221861076,
      "learning_rate": 5.084897229669348e-06,
      "loss": 0.0797,
      "step": 569
    },
    {
      "epoch": 0.15285599356395815,
      "grad_norm": 0.6287612256262585,
      "learning_rate": 5.093833780160859e-06,
      "loss": 0.0406,
      "step": 570
    },
    {
      "epoch": 0.1531241619737195,
      "grad_norm": 0.7376754946842811,
      "learning_rate": 5.102770330652369e-06,
      "loss": 0.0737,
      "step": 571
    },
    {
      "epoch": 0.15339233038348082,
      "grad_norm": 1.4511842740734502,
      "learning_rate": 5.111706881143878e-06,
      "loss": 0.05,
      "step": 572
    },
    {
      "epoch": 0.15366049879324215,
      "grad_norm": 1.9345293173757914,
      "learning_rate": 5.120643431635389e-06,
      "loss": 0.0573,
      "step": 573
    },
    {
      "epoch": 0.1539286672030035,
      "grad_norm": 0.8099357012864555,
      "learning_rate": 5.1295799821268995e-06,
      "loss": 0.0746,
      "step": 574
    },
    {
      "epoch": 0.15419683561276482,
      "grad_norm": 1.084405035264434,
      "learning_rate": 5.13851653261841e-06,
      "loss": 0.0801,
      "step": 575
    },
    {
      "epoch": 0.15446500402252614,
      "grad_norm": 0.875224172009727,
      "learning_rate": 5.14745308310992e-06,
      "loss": 0.0793,
      "step": 576
    },
    {
      "epoch": 0.15473317243228749,
      "grad_norm": 0.6845971696845544,
      "learning_rate": 5.156389633601431e-06,
      "loss": 0.0501,
      "step": 577
    },
    {
      "epoch": 0.1550013408420488,
      "grad_norm": 0.7015040398761847,
      "learning_rate": 5.16532618409294e-06,
      "loss": 0.0561,
      "step": 578
    },
    {
      "epoch": 0.15526950925181013,
      "grad_norm": 0.8101119698797669,
      "learning_rate": 5.174262734584451e-06,
      "loss": 0.0638,
      "step": 579
    },
    {
      "epoch": 0.15553767766157148,
      "grad_norm": 0.902501554293187,
      "learning_rate": 5.1831992850759615e-06,
      "loss": 0.0809,
      "step": 580
    },
    {
      "epoch": 0.1558058460713328,
      "grad_norm": 0.747158066441556,
      "learning_rate": 5.192135835567471e-06,
      "loss": 0.0438,
      "step": 581
    },
    {
      "epoch": 0.15607401448109412,
      "grad_norm": 0.644553316811843,
      "learning_rate": 5.201072386058982e-06,
      "loss": 0.0685,
      "step": 582
    },
    {
      "epoch": 0.15634218289085547,
      "grad_norm": 0.8887122120977885,
      "learning_rate": 5.210008936550492e-06,
      "loss": 0.0702,
      "step": 583
    },
    {
      "epoch": 0.1566103513006168,
      "grad_norm": 0.7918242279058249,
      "learning_rate": 5.218945487042002e-06,
      "loss": 0.0552,
      "step": 584
    },
    {
      "epoch": 0.1568785197103781,
      "grad_norm": 0.7173083180734886,
      "learning_rate": 5.2278820375335125e-06,
      "loss": 0.0592,
      "step": 585
    },
    {
      "epoch": 0.15714668812013946,
      "grad_norm": 0.7722190178674668,
      "learning_rate": 5.2368185880250235e-06,
      "loss": 0.0556,
      "step": 586
    },
    {
      "epoch": 0.15741485652990078,
      "grad_norm": 0.8365821825053458,
      "learning_rate": 5.245755138516533e-06,
      "loss": 0.0848,
      "step": 587
    },
    {
      "epoch": 0.1576830249396621,
      "grad_norm": 0.7757694062814907,
      "learning_rate": 5.254691689008043e-06,
      "loss": 0.0509,
      "step": 588
    },
    {
      "epoch": 0.15795119334942342,
      "grad_norm": 1.0987816124799554,
      "learning_rate": 5.263628239499554e-06,
      "loss": 0.0766,
      "step": 589
    },
    {
      "epoch": 0.15821936175918477,
      "grad_norm": 0.8239668683990621,
      "learning_rate": 5.2725647899910635e-06,
      "loss": 0.0701,
      "step": 590
    },
    {
      "epoch": 0.1584875301689461,
      "grad_norm": 1.0241210739973214,
      "learning_rate": 5.2815013404825745e-06,
      "loss": 0.0764,
      "step": 591
    },
    {
      "epoch": 0.15875569857870742,
      "grad_norm": 0.7479314447965554,
      "learning_rate": 5.290437890974084e-06,
      "loss": 0.0507,
      "step": 592
    },
    {
      "epoch": 0.15902386698846877,
      "grad_norm": 0.73417080951846,
      "learning_rate": 5.299374441465595e-06,
      "loss": 0.0582,
      "step": 593
    },
    {
      "epoch": 0.1592920353982301,
      "grad_norm": 0.7483484584095388,
      "learning_rate": 5.308310991957105e-06,
      "loss": 0.0686,
      "step": 594
    },
    {
      "epoch": 0.1595602038079914,
      "grad_norm": 1.206962169332511,
      "learning_rate": 5.3172475424486145e-06,
      "loss": 0.0887,
      "step": 595
    },
    {
      "epoch": 0.15982837221775276,
      "grad_norm": 0.7954937171928648,
      "learning_rate": 5.3261840929401255e-06,
      "loss": 0.0522,
      "step": 596
    },
    {
      "epoch": 0.16009654062751408,
      "grad_norm": 0.6674926580235067,
      "learning_rate": 5.335120643431636e-06,
      "loss": 0.0492,
      "step": 597
    },
    {
      "epoch": 0.1603647090372754,
      "grad_norm": 0.8918142975031488,
      "learning_rate": 5.344057193923146e-06,
      "loss": 0.0734,
      "step": 598
    },
    {
      "epoch": 0.16063287744703675,
      "grad_norm": 1.554050689585358,
      "learning_rate": 5.352993744414656e-06,
      "loss": 0.0704,
      "step": 599
    },
    {
      "epoch": 0.16090104585679807,
      "grad_norm": 0.6165248410471955,
      "learning_rate": 5.361930294906167e-06,
      "loss": 0.0717,
      "step": 600
    },
    {
      "epoch": 0.1611692142665594,
      "grad_norm": 1.2799547351650304,
      "learning_rate": 5.3708668453976765e-06,
      "loss": 0.0759,
      "step": 601
    },
    {
      "epoch": 0.16143738267632074,
      "grad_norm": 1.3544895398760768,
      "learning_rate": 5.3798033958891875e-06,
      "loss": 0.0675,
      "step": 602
    },
    {
      "epoch": 0.16170555108608206,
      "grad_norm": 0.7554679017626319,
      "learning_rate": 5.388739946380698e-06,
      "loss": 0.0626,
      "step": 603
    },
    {
      "epoch": 0.16197371949584338,
      "grad_norm": 0.6859086960854943,
      "learning_rate": 5.397676496872207e-06,
      "loss": 0.054,
      "step": 604
    },
    {
      "epoch": 0.16224188790560473,
      "grad_norm": 0.8028854363715127,
      "learning_rate": 5.406613047363718e-06,
      "loss": 0.0713,
      "step": 605
    },
    {
      "epoch": 0.16251005631536605,
      "grad_norm": 0.9924929945204021,
      "learning_rate": 5.415549597855228e-06,
      "loss": 0.08,
      "step": 606
    },
    {
      "epoch": 0.16277822472512737,
      "grad_norm": 0.9439529533222822,
      "learning_rate": 5.4244861483467385e-06,
      "loss": 0.0876,
      "step": 607
    },
    {
      "epoch": 0.16304639313488872,
      "grad_norm": 0.7868861895706132,
      "learning_rate": 5.433422698838249e-06,
      "loss": 0.0803,
      "step": 608
    },
    {
      "epoch": 0.16331456154465004,
      "grad_norm": 0.8302736765526366,
      "learning_rate": 5.44235924932976e-06,
      "loss": 0.0674,
      "step": 609
    },
    {
      "epoch": 0.16358272995441137,
      "grad_norm": 0.6945923024705025,
      "learning_rate": 5.451295799821269e-06,
      "loss": 0.0545,
      "step": 610
    },
    {
      "epoch": 0.1638508983641727,
      "grad_norm": 0.692563574807357,
      "learning_rate": 5.460232350312779e-06,
      "loss": 0.0506,
      "step": 611
    },
    {
      "epoch": 0.16411906677393404,
      "grad_norm": 0.5964248691787993,
      "learning_rate": 5.46916890080429e-06,
      "loss": 0.0494,
      "step": 612
    },
    {
      "epoch": 0.16438723518369536,
      "grad_norm": 0.6159780108053444,
      "learning_rate": 5.4781054512958e-06,
      "loss": 0.0594,
      "step": 613
    },
    {
      "epoch": 0.16465540359345668,
      "grad_norm": 0.6806279251009878,
      "learning_rate": 5.487042001787311e-06,
      "loss": 0.0525,
      "step": 614
    },
    {
      "epoch": 0.16492357200321803,
      "grad_norm": 0.8489523265868572,
      "learning_rate": 5.495978552278821e-06,
      "loss": 0.0746,
      "step": 615
    },
    {
      "epoch": 0.16519174041297935,
      "grad_norm": 0.9277273860675815,
      "learning_rate": 5.504915102770331e-06,
      "loss": 0.0762,
      "step": 616
    },
    {
      "epoch": 0.16545990882274067,
      "grad_norm": 0.7561222996460426,
      "learning_rate": 5.513851653261841e-06,
      "loss": 0.0504,
      "step": 617
    },
    {
      "epoch": 0.16572807723250202,
      "grad_norm": 0.8105741312570891,
      "learning_rate": 5.522788203753352e-06,
      "loss": 0.0742,
      "step": 618
    },
    {
      "epoch": 0.16599624564226334,
      "grad_norm": 0.986618312108674,
      "learning_rate": 5.531724754244862e-06,
      "loss": 0.0636,
      "step": 619
    },
    {
      "epoch": 0.16626441405202466,
      "grad_norm": 0.5174299889183016,
      "learning_rate": 5.540661304736372e-06,
      "loss": 0.0478,
      "step": 620
    },
    {
      "epoch": 0.166532582461786,
      "grad_norm": 0.7972667418516348,
      "learning_rate": 5.549597855227883e-06,
      "loss": 0.0599,
      "step": 621
    },
    {
      "epoch": 0.16680075087154733,
      "grad_norm": 0.7521932679692337,
      "learning_rate": 5.558534405719392e-06,
      "loss": 0.0636,
      "step": 622
    },
    {
      "epoch": 0.16706891928130865,
      "grad_norm": 0.7596575425805953,
      "learning_rate": 5.567470956210903e-06,
      "loss": 0.0626,
      "step": 623
    },
    {
      "epoch": 0.16733708769107,
      "grad_norm": 1.0278251859880065,
      "learning_rate": 5.5764075067024135e-06,
      "loss": 0.0842,
      "step": 624
    },
    {
      "epoch": 0.16760525610083132,
      "grad_norm": 0.9691655242783864,
      "learning_rate": 5.585344057193923e-06,
      "loss": 0.0736,
      "step": 625
    },
    {
      "epoch": 0.16787342451059264,
      "grad_norm": 0.7920761320230346,
      "learning_rate": 5.594280607685434e-06,
      "loss": 0.0645,
      "step": 626
    },
    {
      "epoch": 0.168141592920354,
      "grad_norm": 0.669562952121449,
      "learning_rate": 5.603217158176945e-06,
      "loss": 0.0528,
      "step": 627
    },
    {
      "epoch": 0.16840976133011532,
      "grad_norm": 0.8605030640451825,
      "learning_rate": 5.612153708668454e-06,
      "loss": 0.0572,
      "step": 628
    },
    {
      "epoch": 0.16867792973987664,
      "grad_norm": 0.7266176227579504,
      "learning_rate": 5.6210902591599645e-06,
      "loss": 0.0687,
      "step": 629
    },
    {
      "epoch": 0.16894609814963799,
      "grad_norm": 0.543948406582875,
      "learning_rate": 5.6300268096514755e-06,
      "loss": 0.0554,
      "step": 630
    },
    {
      "epoch": 0.1692142665593993,
      "grad_norm": 0.7049075660316708,
      "learning_rate": 5.638963360142985e-06,
      "loss": 0.0685,
      "step": 631
    },
    {
      "epoch": 0.16948243496916063,
      "grad_norm": 0.7040457753482281,
      "learning_rate": 5.647899910634496e-06,
      "loss": 0.0575,
      "step": 632
    },
    {
      "epoch": 0.16975060337892195,
      "grad_norm": 0.9803494851741242,
      "learning_rate": 5.656836461126006e-06,
      "loss": 0.0803,
      "step": 633
    },
    {
      "epoch": 0.1700187717886833,
      "grad_norm": 0.7701798894334154,
      "learning_rate": 5.6657730116175155e-06,
      "loss": 0.0591,
      "step": 634
    },
    {
      "epoch": 0.17028694019844462,
      "grad_norm": 1.0212495968892932,
      "learning_rate": 5.6747095621090265e-06,
      "loss": 0.0663,
      "step": 635
    },
    {
      "epoch": 0.17055510860820594,
      "grad_norm": 0.9483956706703315,
      "learning_rate": 5.683646112600537e-06,
      "loss": 0.061,
      "step": 636
    },
    {
      "epoch": 0.1708232770179673,
      "grad_norm": 0.6614724647899509,
      "learning_rate": 5.692582663092047e-06,
      "loss": 0.0582,
      "step": 637
    },
    {
      "epoch": 0.1710914454277286,
      "grad_norm": 0.9309092810402265,
      "learning_rate": 5.701519213583557e-06,
      "loss": 0.0678,
      "step": 638
    },
    {
      "epoch": 0.17135961383748993,
      "grad_norm": 0.6903336159620008,
      "learning_rate": 5.710455764075068e-06,
      "loss": 0.0763,
      "step": 639
    },
    {
      "epoch": 0.17162778224725128,
      "grad_norm": 0.7321010050308411,
      "learning_rate": 5.7193923145665775e-06,
      "loss": 0.0696,
      "step": 640
    },
    {
      "epoch": 0.1718959506570126,
      "grad_norm": 1.2930160472906331,
      "learning_rate": 5.7283288650580885e-06,
      "loss": 0.0878,
      "step": 641
    },
    {
      "epoch": 0.17216411906677392,
      "grad_norm": 0.9092015676328017,
      "learning_rate": 5.737265415549599e-06,
      "loss": 0.083,
      "step": 642
    },
    {
      "epoch": 0.17243228747653527,
      "grad_norm": 0.6640708112231667,
      "learning_rate": 5.746201966041108e-06,
      "loss": 0.0445,
      "step": 643
    },
    {
      "epoch": 0.1727004558862966,
      "grad_norm": 0.7783794510147659,
      "learning_rate": 5.755138516532619e-06,
      "loss": 0.0522,
      "step": 644
    },
    {
      "epoch": 0.17296862429605792,
      "grad_norm": 0.8009510229888075,
      "learning_rate": 5.764075067024129e-06,
      "loss": 0.0605,
      "step": 645
    },
    {
      "epoch": 0.17323679270581926,
      "grad_norm": 1.219838757150327,
      "learning_rate": 5.7730116175156395e-06,
      "loss": 0.0697,
      "step": 646
    },
    {
      "epoch": 0.17350496111558059,
      "grad_norm": 0.7651132017369591,
      "learning_rate": 5.78194816800715e-06,
      "loss": 0.067,
      "step": 647
    },
    {
      "epoch": 0.1737731295253419,
      "grad_norm": 0.6797202399923142,
      "learning_rate": 5.790884718498661e-06,
      "loss": 0.0548,
      "step": 648
    },
    {
      "epoch": 0.17404129793510326,
      "grad_norm": 0.743548123014249,
      "learning_rate": 5.79982126899017e-06,
      "loss": 0.0632,
      "step": 649
    },
    {
      "epoch": 0.17430946634486458,
      "grad_norm": 0.6220641344644409,
      "learning_rate": 5.808757819481681e-06,
      "loss": 0.0593,
      "step": 650
    },
    {
      "epoch": 0.1745776347546259,
      "grad_norm": 0.5767877867894173,
      "learning_rate": 5.817694369973191e-06,
      "loss": 0.0491,
      "step": 651
    },
    {
      "epoch": 0.17484580316438725,
      "grad_norm": 0.9321535693181042,
      "learning_rate": 5.826630920464701e-06,
      "loss": 0.0969,
      "step": 652
    },
    {
      "epoch": 0.17511397157414857,
      "grad_norm": 0.7302847182071687,
      "learning_rate": 5.835567470956212e-06,
      "loss": 0.0526,
      "step": 653
    },
    {
      "epoch": 0.1753821399839099,
      "grad_norm": 0.9198630289552819,
      "learning_rate": 5.844504021447721e-06,
      "loss": 0.0809,
      "step": 654
    },
    {
      "epoch": 0.1756503083936712,
      "grad_norm": 0.8641497488576019,
      "learning_rate": 5.853440571939232e-06,
      "loss": 0.0572,
      "step": 655
    },
    {
      "epoch": 0.17591847680343256,
      "grad_norm": 0.9951660257601543,
      "learning_rate": 5.862377122430742e-06,
      "loss": 0.0751,
      "step": 656
    },
    {
      "epoch": 0.17618664521319388,
      "grad_norm": 0.9342921859576737,
      "learning_rate": 5.871313672922252e-06,
      "loss": 0.0696,
      "step": 657
    },
    {
      "epoch": 0.1764548136229552,
      "grad_norm": 1.0568270426819426,
      "learning_rate": 5.880250223413763e-06,
      "loss": 0.0938,
      "step": 658
    },
    {
      "epoch": 0.17672298203271655,
      "grad_norm": 1.0442793889830606,
      "learning_rate": 5.889186773905273e-06,
      "loss": 0.0534,
      "step": 659
    },
    {
      "epoch": 0.17699115044247787,
      "grad_norm": 0.846547023328437,
      "learning_rate": 5.898123324396783e-06,
      "loss": 0.0589,
      "step": 660
    },
    {
      "epoch": 0.1772593188522392,
      "grad_norm": 0.9473986989715096,
      "learning_rate": 5.907059874888293e-06,
      "loss": 0.0622,
      "step": 661
    },
    {
      "epoch": 0.17752748726200054,
      "grad_norm": 0.6554661375896103,
      "learning_rate": 5.915996425379804e-06,
      "loss": 0.0488,
      "step": 662
    },
    {
      "epoch": 0.17779565567176187,
      "grad_norm": 1.6004788984951979,
      "learning_rate": 5.924932975871314e-06,
      "loss": 0.0898,
      "step": 663
    },
    {
      "epoch": 0.1780638240815232,
      "grad_norm": 0.6196855926162149,
      "learning_rate": 5.933869526362825e-06,
      "loss": 0.0427,
      "step": 664
    },
    {
      "epoch": 0.17833199249128454,
      "grad_norm": 2.1065193315449484,
      "learning_rate": 5.942806076854335e-06,
      "loss": 0.0796,
      "step": 665
    },
    {
      "epoch": 0.17860016090104586,
      "grad_norm": 0.7052395430019095,
      "learning_rate": 5.951742627345844e-06,
      "loss": 0.0526,
      "step": 666
    },
    {
      "epoch": 0.17886832931080718,
      "grad_norm": 1.6387407274136412,
      "learning_rate": 5.960679177837355e-06,
      "loss": 0.0875,
      "step": 667
    },
    {
      "epoch": 0.17913649772056853,
      "grad_norm": 0.8190896983147427,
      "learning_rate": 5.9696157283288655e-06,
      "loss": 0.0597,
      "step": 668
    },
    {
      "epoch": 0.17940466613032985,
      "grad_norm": 0.9239746877888685,
      "learning_rate": 5.978552278820376e-06,
      "loss": 0.0589,
      "step": 669
    },
    {
      "epoch": 0.17967283454009117,
      "grad_norm": 0.7258910140441152,
      "learning_rate": 5.987488829311886e-06,
      "loss": 0.0584,
      "step": 670
    },
    {
      "epoch": 0.17994100294985252,
      "grad_norm": 0.6944481489693888,
      "learning_rate": 5.996425379803397e-06,
      "loss": 0.0619,
      "step": 671
    },
    {
      "epoch": 0.18020917135961384,
      "grad_norm": 0.5876336060985152,
      "learning_rate": 6.005361930294906e-06,
      "loss": 0.0541,
      "step": 672
    },
    {
      "epoch": 0.18047733976937516,
      "grad_norm": 1.7155744326653621,
      "learning_rate": 6.0142984807864165e-06,
      "loss": 0.083,
      "step": 673
    },
    {
      "epoch": 0.1807455081791365,
      "grad_norm": 0.798476351806146,
      "learning_rate": 6.0232350312779275e-06,
      "loss": 0.062,
      "step": 674
    },
    {
      "epoch": 0.18101367658889783,
      "grad_norm": 0.7091761098575183,
      "learning_rate": 6.032171581769437e-06,
      "loss": 0.0582,
      "step": 675
    },
    {
      "epoch": 0.18128184499865915,
      "grad_norm": 1.2051898320378824,
      "learning_rate": 6.041108132260948e-06,
      "loss": 0.0861,
      "step": 676
    },
    {
      "epoch": 0.18155001340842047,
      "grad_norm": 0.6781611369231361,
      "learning_rate": 6.050044682752458e-06,
      "loss": 0.0487,
      "step": 677
    },
    {
      "epoch": 0.18181818181818182,
      "grad_norm": 0.6771110157218314,
      "learning_rate": 6.058981233243968e-06,
      "loss": 0.054,
      "step": 678
    },
    {
      "epoch": 0.18208635022794314,
      "grad_norm": 0.7362147440205621,
      "learning_rate": 6.0679177837354785e-06,
      "loss": 0.0501,
      "step": 679
    },
    {
      "epoch": 0.18235451863770447,
      "grad_norm": 0.7620786157266373,
      "learning_rate": 6.0768543342269895e-06,
      "loss": 0.0563,
      "step": 680
    },
    {
      "epoch": 0.18262268704746581,
      "grad_norm": 1.0385514676432708,
      "learning_rate": 6.085790884718499e-06,
      "loss": 0.083,
      "step": 681
    },
    {
      "epoch": 0.18289085545722714,
      "grad_norm": 0.6984213217944104,
      "learning_rate": 6.094727435210009e-06,
      "loss": 0.0655,
      "step": 682
    },
    {
      "epoch": 0.18315902386698846,
      "grad_norm": 0.7142706630671537,
      "learning_rate": 6.10366398570152e-06,
      "loss": 0.0698,
      "step": 683
    },
    {
      "epoch": 0.1834271922767498,
      "grad_norm": 0.7154365286170392,
      "learning_rate": 6.1126005361930295e-06,
      "loss": 0.0629,
      "step": 684
    },
    {
      "epoch": 0.18369536068651113,
      "grad_norm": 0.647465816127068,
      "learning_rate": 6.1215370866845405e-06,
      "loss": 0.0572,
      "step": 685
    },
    {
      "epoch": 0.18396352909627245,
      "grad_norm": 0.7181705009097983,
      "learning_rate": 6.130473637176051e-06,
      "loss": 0.0735,
      "step": 686
    },
    {
      "epoch": 0.1842316975060338,
      "grad_norm": 0.6845166979442948,
      "learning_rate": 6.139410187667561e-06,
      "loss": 0.0492,
      "step": 687
    },
    {
      "epoch": 0.18449986591579512,
      "grad_norm": 0.6513732441259296,
      "learning_rate": 6.148346738159071e-06,
      "loss": 0.0659,
      "step": 688
    },
    {
      "epoch": 0.18476803432555644,
      "grad_norm": 0.683061812339965,
      "learning_rate": 6.157283288650582e-06,
      "loss": 0.0562,
      "step": 689
    },
    {
      "epoch": 0.1850362027353178,
      "grad_norm": 0.7399919598847905,
      "learning_rate": 6.1662198391420915e-06,
      "loss": 0.0455,
      "step": 690
    },
    {
      "epoch": 0.1853043711450791,
      "grad_norm": 0.5870580379027058,
      "learning_rate": 6.175156389633602e-06,
      "loss": 0.0597,
      "step": 691
    },
    {
      "epoch": 0.18557253955484043,
      "grad_norm": 0.7438342058561855,
      "learning_rate": 6.184092940125113e-06,
      "loss": 0.0638,
      "step": 692
    },
    {
      "epoch": 0.18584070796460178,
      "grad_norm": 0.6849801703471039,
      "learning_rate": 6.193029490616622e-06,
      "loss": 0.0651,
      "step": 693
    },
    {
      "epoch": 0.1861088763743631,
      "grad_norm": 0.6160922392252735,
      "learning_rate": 6.201966041108133e-06,
      "loss": 0.0463,
      "step": 694
    },
    {
      "epoch": 0.18637704478412442,
      "grad_norm": 0.5765825815633243,
      "learning_rate": 6.210902591599643e-06,
      "loss": 0.0436,
      "step": 695
    },
    {
      "epoch": 0.18664521319388577,
      "grad_norm": 0.5969321503429132,
      "learning_rate": 6.219839142091153e-06,
      "loss": 0.0693,
      "step": 696
    },
    {
      "epoch": 0.1869133816036471,
      "grad_norm": 0.643916489809957,
      "learning_rate": 6.228775692582664e-06,
      "loss": 0.0814,
      "step": 697
    },
    {
      "epoch": 0.18718155001340842,
      "grad_norm": 0.7132738026241066,
      "learning_rate": 6.237712243074175e-06,
      "loss": 0.0628,
      "step": 698
    },
    {
      "epoch": 0.18744971842316976,
      "grad_norm": 0.6426709438117871,
      "learning_rate": 6.246648793565684e-06,
      "loss": 0.0568,
      "step": 699
    },
    {
      "epoch": 0.18771788683293109,
      "grad_norm": 0.5918446386191234,
      "learning_rate": 6.255585344057194e-06,
      "loss": 0.0484,
      "step": 700
    },
    {
      "epoch": 0.1879860552426924,
      "grad_norm": 0.5462704738042106,
      "learning_rate": 6.264521894548705e-06,
      "loss": 0.044,
      "step": 701
    },
    {
      "epoch": 0.18825422365245373,
      "grad_norm": 0.7314206212373209,
      "learning_rate": 6.273458445040215e-06,
      "loss": 0.0727,
      "step": 702
    },
    {
      "epoch": 0.18852239206221508,
      "grad_norm": 0.6659283042547272,
      "learning_rate": 6.282394995531726e-06,
      "loss": 0.0589,
      "step": 703
    },
    {
      "epoch": 0.1887905604719764,
      "grad_norm": 0.6684930273676993,
      "learning_rate": 6.291331546023236e-06,
      "loss": 0.0661,
      "step": 704
    },
    {
      "epoch": 0.18905872888173772,
      "grad_norm": 0.8904769710559233,
      "learning_rate": 6.300268096514745e-06,
      "loss": 0.0632,
      "step": 705
    },
    {
      "epoch": 0.18932689729149907,
      "grad_norm": 0.8789493229124882,
      "learning_rate": 6.309204647006256e-06,
      "loss": 0.0676,
      "step": 706
    },
    {
      "epoch": 0.1895950657012604,
      "grad_norm": 0.6501276177777463,
      "learning_rate": 6.3181411974977665e-06,
      "loss": 0.0545,
      "step": 707
    },
    {
      "epoch": 0.1898632341110217,
      "grad_norm": 1.0843691126438044,
      "learning_rate": 6.327077747989277e-06,
      "loss": 0.0551,
      "step": 708
    },
    {
      "epoch": 0.19013140252078306,
      "grad_norm": 1.0010513130523677,
      "learning_rate": 6.336014298480787e-06,
      "loss": 0.0754,
      "step": 709
    },
    {
      "epoch": 0.19039957093054438,
      "grad_norm": 1.049178684107961,
      "learning_rate": 6.344950848972298e-06,
      "loss": 0.088,
      "step": 710
    },
    {
      "epoch": 0.1906677393403057,
      "grad_norm": 0.7670234048871027,
      "learning_rate": 6.353887399463807e-06,
      "loss": 0.063,
      "step": 711
    },
    {
      "epoch": 0.19093590775006705,
      "grad_norm": 0.71542928973344,
      "learning_rate": 6.362823949955318e-06,
      "loss": 0.047,
      "step": 712
    },
    {
      "epoch": 0.19120407615982837,
      "grad_norm": 1.4447203202553363,
      "learning_rate": 6.3717605004468285e-06,
      "loss": 0.0461,
      "step": 713
    },
    {
      "epoch": 0.1914722445695897,
      "grad_norm": 0.7640657463427055,
      "learning_rate": 6.380697050938338e-06,
      "loss": 0.0546,
      "step": 714
    },
    {
      "epoch": 0.19174041297935104,
      "grad_norm": 0.9553917219481026,
      "learning_rate": 6.389633601429849e-06,
      "loss": 0.0729,
      "step": 715
    },
    {
      "epoch": 0.19200858138911236,
      "grad_norm": 0.5881144124533128,
      "learning_rate": 6.398570151921358e-06,
      "loss": 0.0454,
      "step": 716
    },
    {
      "epoch": 0.1922767497988737,
      "grad_norm": 0.7518000587182011,
      "learning_rate": 6.407506702412869e-06,
      "loss": 0.0704,
      "step": 717
    },
    {
      "epoch": 0.19254491820863504,
      "grad_norm": 0.743107153030265,
      "learning_rate": 6.4164432529043795e-06,
      "loss": 0.0679,
      "step": 718
    },
    {
      "epoch": 0.19281308661839636,
      "grad_norm": 2.0013645612435336,
      "learning_rate": 6.425379803395889e-06,
      "loss": 0.0759,
      "step": 719
    },
    {
      "epoch": 0.19308125502815768,
      "grad_norm": 0.6417405077349332,
      "learning_rate": 6.4343163538874e-06,
      "loss": 0.0649,
      "step": 720
    },
    {
      "epoch": 0.19334942343791903,
      "grad_norm": 0.6050146092617343,
      "learning_rate": 6.44325290437891e-06,
      "loss": 0.0438,
      "step": 721
    },
    {
      "epoch": 0.19361759184768035,
      "grad_norm": 0.6985273578197806,
      "learning_rate": 6.45218945487042e-06,
      "loss": 0.0472,
      "step": 722
    },
    {
      "epoch": 0.19388576025744167,
      "grad_norm": 0.7744006084837489,
      "learning_rate": 6.4611260053619305e-06,
      "loss": 0.0696,
      "step": 723
    },
    {
      "epoch": 0.194153928667203,
      "grad_norm": 1.0599279825449373,
      "learning_rate": 6.4700625558534415e-06,
      "loss": 0.066,
      "step": 724
    },
    {
      "epoch": 0.19442209707696434,
      "grad_norm": 0.6193995176176766,
      "learning_rate": 6.478999106344951e-06,
      "loss": 0.0626,
      "step": 725
    },
    {
      "epoch": 0.19469026548672566,
      "grad_norm": 0.680651661916281,
      "learning_rate": 6.487935656836462e-06,
      "loss": 0.0609,
      "step": 726
    },
    {
      "epoch": 0.19495843389648698,
      "grad_norm": 0.7131379200257753,
      "learning_rate": 6.496872207327972e-06,
      "loss": 0.0582,
      "step": 727
    },
    {
      "epoch": 0.19522660230624833,
      "grad_norm": 0.9252839433347055,
      "learning_rate": 6.5058087578194814e-06,
      "loss": 0.0648,
      "step": 728
    },
    {
      "epoch": 0.19549477071600965,
      "grad_norm": 0.6412540232830793,
      "learning_rate": 6.5147453083109925e-06,
      "loss": 0.0605,
      "step": 729
    },
    {
      "epoch": 0.19576293912577097,
      "grad_norm": 1.0658277131055782,
      "learning_rate": 6.523681858802503e-06,
      "loss": 0.0836,
      "step": 730
    },
    {
      "epoch": 0.19603110753553232,
      "grad_norm": 1.922141508857633,
      "learning_rate": 6.532618409294013e-06,
      "loss": 0.0904,
      "step": 731
    },
    {
      "epoch": 0.19629927594529364,
      "grad_norm": 0.5708354262791332,
      "learning_rate": 6.541554959785523e-06,
      "loss": 0.0552,
      "step": 732
    },
    {
      "epoch": 0.19656744435505497,
      "grad_norm": 0.6681504053636753,
      "learning_rate": 6.550491510277034e-06,
      "loss": 0.0491,
      "step": 733
    },
    {
      "epoch": 0.19683561276481631,
      "grad_norm": 0.588696101953411,
      "learning_rate": 6.5594280607685435e-06,
      "loss": 0.0457,
      "step": 734
    },
    {
      "epoch": 0.19710378117457764,
      "grad_norm": 0.7627814762672769,
      "learning_rate": 6.5683646112600545e-06,
      "loss": 0.0627,
      "step": 735
    },
    {
      "epoch": 0.19737194958433896,
      "grad_norm": 0.8569070314146487,
      "learning_rate": 6.577301161751565e-06,
      "loss": 0.0685,
      "step": 736
    },
    {
      "epoch": 0.1976401179941003,
      "grad_norm": 0.8085062704934447,
      "learning_rate": 6.586237712243074e-06,
      "loss": 0.0781,
      "step": 737
    },
    {
      "epoch": 0.19790828640386163,
      "grad_norm": 0.9934832587248366,
      "learning_rate": 6.595174262734585e-06,
      "loss": 0.063,
      "step": 738
    },
    {
      "epoch": 0.19817645481362295,
      "grad_norm": 0.7144092411098115,
      "learning_rate": 6.604110813226095e-06,
      "loss": 0.0632,
      "step": 739
    },
    {
      "epoch": 0.1984446232233843,
      "grad_norm": 0.45612866487212417,
      "learning_rate": 6.6130473637176055e-06,
      "loss": 0.0432,
      "step": 740
    },
    {
      "epoch": 0.19871279163314562,
      "grad_norm": 0.6463781410835909,
      "learning_rate": 6.621983914209116e-06,
      "loss": 0.0688,
      "step": 741
    },
    {
      "epoch": 0.19898096004290694,
      "grad_norm": 0.6339875479418067,
      "learning_rate": 6.630920464700627e-06,
      "loss": 0.0554,
      "step": 742
    },
    {
      "epoch": 0.1992491284526683,
      "grad_norm": 0.6039134839528961,
      "learning_rate": 6.639857015192136e-06,
      "loss": 0.0509,
      "step": 743
    },
    {
      "epoch": 0.1995172968624296,
      "grad_norm": 0.9204030128839711,
      "learning_rate": 6.648793565683646e-06,
      "loss": 0.0691,
      "step": 744
    },
    {
      "epoch": 0.19978546527219093,
      "grad_norm": 0.6527283096508315,
      "learning_rate": 6.657730116175157e-06,
      "loss": 0.0672,
      "step": 745
    },
    {
      "epoch": 0.20005363368195225,
      "grad_norm": 0.5656081531362551,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.0493,
      "step": 746
    },
    {
      "epoch": 0.2003218020917136,
      "grad_norm": 0.5388498726659108,
      "learning_rate": 6.675603217158178e-06,
      "loss": 0.0505,
      "step": 747
    },
    {
      "epoch": 0.20058997050147492,
      "grad_norm": 0.634389081117935,
      "learning_rate": 6.684539767649688e-06,
      "loss": 0.0598,
      "step": 748
    },
    {
      "epoch": 0.20085813891123624,
      "grad_norm": 0.7855810387645783,
      "learning_rate": 6.693476318141198e-06,
      "loss": 0.0577,
      "step": 749
    },
    {
      "epoch": 0.2011263073209976,
      "grad_norm": 0.9079828407382363,
      "learning_rate": 6.702412868632708e-06,
      "loss": 0.0572,
      "step": 750
    },
    {
      "epoch": 0.20139447573075892,
      "grad_norm": 0.670347649066555,
      "learning_rate": 6.711349419124219e-06,
      "loss": 0.0832,
      "step": 751
    },
    {
      "epoch": 0.20166264414052024,
      "grad_norm": 0.6314476472333589,
      "learning_rate": 6.720285969615729e-06,
      "loss": 0.064,
      "step": 752
    },
    {
      "epoch": 0.20193081255028159,
      "grad_norm": 0.6554007399542807,
      "learning_rate": 6.729222520107239e-06,
      "loss": 0.0595,
      "step": 753
    },
    {
      "epoch": 0.2021989809600429,
      "grad_norm": 0.9287464566943074,
      "learning_rate": 6.73815907059875e-06,
      "loss": 0.0673,
      "step": 754
    },
    {
      "epoch": 0.20246714936980423,
      "grad_norm": 1.0299474180112156,
      "learning_rate": 6.747095621090259e-06,
      "loss": 0.0768,
      "step": 755
    },
    {
      "epoch": 0.20273531777956558,
      "grad_norm": 0.6450382377010188,
      "learning_rate": 6.75603217158177e-06,
      "loss": 0.0467,
      "step": 756
    },
    {
      "epoch": 0.2030034861893269,
      "grad_norm": 0.9463199608458214,
      "learning_rate": 6.7649687220732805e-06,
      "loss": 0.0572,
      "step": 757
    },
    {
      "epoch": 0.20327165459908822,
      "grad_norm": 0.7266263781091684,
      "learning_rate": 6.77390527256479e-06,
      "loss": 0.0542,
      "step": 758
    },
    {
      "epoch": 0.20353982300884957,
      "grad_norm": 0.8903936647426972,
      "learning_rate": 6.782841823056301e-06,
      "loss": 0.073,
      "step": 759
    },
    {
      "epoch": 0.2038079914186109,
      "grad_norm": 1.032084666218354,
      "learning_rate": 6.791778373547812e-06,
      "loss": 0.0625,
      "step": 760
    },
    {
      "epoch": 0.2040761598283722,
      "grad_norm": 0.8230260287218641,
      "learning_rate": 6.800714924039321e-06,
      "loss": 0.08,
      "step": 761
    },
    {
      "epoch": 0.20434432823813356,
      "grad_norm": 0.7691609736623329,
      "learning_rate": 6.8096514745308315e-06,
      "loss": 0.0795,
      "step": 762
    },
    {
      "epoch": 0.20461249664789488,
      "grad_norm": 0.9735415188732415,
      "learning_rate": 6.8185880250223425e-06,
      "loss": 0.0563,
      "step": 763
    },
    {
      "epoch": 0.2048806650576562,
      "grad_norm": 0.6016653128156609,
      "learning_rate": 6.827524575513852e-06,
      "loss": 0.0581,
      "step": 764
    },
    {
      "epoch": 0.20514883346741755,
      "grad_norm": 1.0826231146782523,
      "learning_rate": 6.836461126005363e-06,
      "loss": 0.0632,
      "step": 765
    },
    {
      "epoch": 0.20541700187717887,
      "grad_norm": 0.9275693654000192,
      "learning_rate": 6.845397676496873e-06,
      "loss": 0.0575,
      "step": 766
    },
    {
      "epoch": 0.2056851702869402,
      "grad_norm": 0.8153801558858359,
      "learning_rate": 6.8543342269883824e-06,
      "loss": 0.0649,
      "step": 767
    },
    {
      "epoch": 0.20595333869670152,
      "grad_norm": 0.6880063379564347,
      "learning_rate": 6.8632707774798935e-06,
      "loss": 0.0616,
      "step": 768
    },
    {
      "epoch": 0.20622150710646286,
      "grad_norm": 0.9700171857880753,
      "learning_rate": 6.8722073279714045e-06,
      "loss": 0.0756,
      "step": 769
    },
    {
      "epoch": 0.20648967551622419,
      "grad_norm": 1.0846412565864332,
      "learning_rate": 6.881143878462914e-06,
      "loss": 0.0633,
      "step": 770
    },
    {
      "epoch": 0.2067578439259855,
      "grad_norm": 0.6348597667757498,
      "learning_rate": 6.890080428954424e-06,
      "loss": 0.0661,
      "step": 771
    },
    {
      "epoch": 0.20702601233574686,
      "grad_norm": 0.797333089257687,
      "learning_rate": 6.899016979445935e-06,
      "loss": 0.0672,
      "step": 772
    },
    {
      "epoch": 0.20729418074550818,
      "grad_norm": 0.5463356833367039,
      "learning_rate": 6.9079535299374445e-06,
      "loss": 0.0447,
      "step": 773
    },
    {
      "epoch": 0.2075623491552695,
      "grad_norm": 0.6191221910365377,
      "learning_rate": 6.9168900804289555e-06,
      "loss": 0.0571,
      "step": 774
    },
    {
      "epoch": 0.20783051756503085,
      "grad_norm": 0.5603309360612991,
      "learning_rate": 6.925826630920466e-06,
      "loss": 0.0518,
      "step": 775
    },
    {
      "epoch": 0.20809868597479217,
      "grad_norm": 0.9648687674676005,
      "learning_rate": 6.934763181411975e-06,
      "loss": 0.0646,
      "step": 776
    },
    {
      "epoch": 0.2083668543845535,
      "grad_norm": 0.45845159925062956,
      "learning_rate": 6.943699731903486e-06,
      "loss": 0.0405,
      "step": 777
    },
    {
      "epoch": 0.20863502279431484,
      "grad_norm": 0.4927102770396721,
      "learning_rate": 6.9526362823949954e-06,
      "loss": 0.0505,
      "step": 778
    },
    {
      "epoch": 0.20890319120407616,
      "grad_norm": 0.552962488159094,
      "learning_rate": 6.9615728328865065e-06,
      "loss": 0.0584,
      "step": 779
    },
    {
      "epoch": 0.20917135961383748,
      "grad_norm": 0.6149816687747405,
      "learning_rate": 6.970509383378017e-06,
      "loss": 0.0596,
      "step": 780
    },
    {
      "epoch": 0.20943952802359883,
      "grad_norm": 0.7737296116789633,
      "learning_rate": 6.979445933869526e-06,
      "loss": 0.0571,
      "step": 781
    },
    {
      "epoch": 0.20970769643336015,
      "grad_norm": 0.6835090103999594,
      "learning_rate": 6.988382484361037e-06,
      "loss": 0.065,
      "step": 782
    },
    {
      "epoch": 0.20997586484312147,
      "grad_norm": 0.7096155170503166,
      "learning_rate": 6.997319034852548e-06,
      "loss": 0.0529,
      "step": 783
    },
    {
      "epoch": 0.21024403325288282,
      "grad_norm": 0.6805209391388485,
      "learning_rate": 7.0062555853440575e-06,
      "loss": 0.056,
      "step": 784
    },
    {
      "epoch": 0.21051220166264414,
      "grad_norm": 0.7119773150022219,
      "learning_rate": 7.015192135835568e-06,
      "loss": 0.0707,
      "step": 785
    },
    {
      "epoch": 0.21078037007240547,
      "grad_norm": 0.554659736143505,
      "learning_rate": 7.024128686327079e-06,
      "loss": 0.048,
      "step": 786
    },
    {
      "epoch": 0.21104853848216681,
      "grad_norm": 0.6601577051403399,
      "learning_rate": 7.033065236818588e-06,
      "loss": 0.0462,
      "step": 787
    },
    {
      "epoch": 0.21131670689192814,
      "grad_norm": 0.9211809589193016,
      "learning_rate": 7.042001787310099e-06,
      "loss": 0.0677,
      "step": 788
    },
    {
      "epoch": 0.21158487530168946,
      "grad_norm": 0.6163791453478012,
      "learning_rate": 7.050938337801609e-06,
      "loss": 0.0521,
      "step": 789
    },
    {
      "epoch": 0.21185304371145078,
      "grad_norm": 0.40380838402768837,
      "learning_rate": 7.059874888293119e-06,
      "loss": 0.0384,
      "step": 790
    },
    {
      "epoch": 0.21212121212121213,
      "grad_norm": 0.833909629797656,
      "learning_rate": 7.06881143878463e-06,
      "loss": 0.0537,
      "step": 791
    },
    {
      "epoch": 0.21238938053097345,
      "grad_norm": 1.1530861121014904,
      "learning_rate": 7.07774798927614e-06,
      "loss": 0.0951,
      "step": 792
    },
    {
      "epoch": 0.21265754894073477,
      "grad_norm": 0.8188636801174013,
      "learning_rate": 7.08668453976765e-06,
      "loss": 0.0697,
      "step": 793
    },
    {
      "epoch": 0.21292571735049612,
      "grad_norm": 0.7751355457348119,
      "learning_rate": 7.09562109025916e-06,
      "loss": 0.0716,
      "step": 794
    },
    {
      "epoch": 0.21319388576025744,
      "grad_norm": 0.7196255635276604,
      "learning_rate": 7.104557640750671e-06,
      "loss": 0.0646,
      "step": 795
    },
    {
      "epoch": 0.21346205417001876,
      "grad_norm": 0.7736053275119991,
      "learning_rate": 7.113494191242181e-06,
      "loss": 0.0627,
      "step": 796
    },
    {
      "epoch": 0.2137302225797801,
      "grad_norm": 0.8514683615561239,
      "learning_rate": 7.122430741733692e-06,
      "loss": 0.0685,
      "step": 797
    },
    {
      "epoch": 0.21399839098954143,
      "grad_norm": 0.5222887146153904,
      "learning_rate": 7.131367292225202e-06,
      "loss": 0.0544,
      "step": 798
    },
    {
      "epoch": 0.21426655939930275,
      "grad_norm": 0.6663872399483806,
      "learning_rate": 7.140303842716711e-06,
      "loss": 0.0552,
      "step": 799
    },
    {
      "epoch": 0.2145347278090641,
      "grad_norm": 0.7370556435111065,
      "learning_rate": 7.149240393208222e-06,
      "loss": 0.0586,
      "step": 800
    },
    {
      "epoch": 0.21480289621882542,
      "grad_norm": 0.6546419429478776,
      "learning_rate": 7.1581769436997325e-06,
      "loss": 0.0605,
      "step": 801
    },
    {
      "epoch": 0.21507106462858674,
      "grad_norm": 0.8946085794405229,
      "learning_rate": 7.167113494191243e-06,
      "loss": 0.0752,
      "step": 802
    },
    {
      "epoch": 0.2153392330383481,
      "grad_norm": 0.541265191858049,
      "learning_rate": 7.176050044682753e-06,
      "loss": 0.0596,
      "step": 803
    },
    {
      "epoch": 0.21560740144810941,
      "grad_norm": 0.9304484664641347,
      "learning_rate": 7.184986595174264e-06,
      "loss": 0.0722,
      "step": 804
    },
    {
      "epoch": 0.21587556985787074,
      "grad_norm": 0.6079871798137215,
      "learning_rate": 7.193923145665773e-06,
      "loss": 0.0531,
      "step": 805
    },
    {
      "epoch": 0.21614373826763209,
      "grad_norm": 0.675408167444496,
      "learning_rate": 7.2028596961572834e-06,
      "loss": 0.0514,
      "step": 806
    },
    {
      "epoch": 0.2164119066773934,
      "grad_norm": 0.6361953625181196,
      "learning_rate": 7.2117962466487945e-06,
      "loss": 0.055,
      "step": 807
    },
    {
      "epoch": 0.21668007508715473,
      "grad_norm": 0.700218172568977,
      "learning_rate": 7.220732797140304e-06,
      "loss": 0.0734,
      "step": 808
    },
    {
      "epoch": 0.21694824349691608,
      "grad_norm": 0.7624824922129924,
      "learning_rate": 7.229669347631815e-06,
      "loss": 0.0519,
      "step": 809
    },
    {
      "epoch": 0.2172164119066774,
      "grad_norm": 0.7079601223869098,
      "learning_rate": 7.238605898123325e-06,
      "loss": 0.0584,
      "step": 810
    },
    {
      "epoch": 0.21748458031643872,
      "grad_norm": 0.5521740348722388,
      "learning_rate": 7.247542448614835e-06,
      "loss": 0.0448,
      "step": 811
    },
    {
      "epoch": 0.21775274872620004,
      "grad_norm": 0.778954048716299,
      "learning_rate": 7.2564789991063455e-06,
      "loss": 0.0553,
      "step": 812
    },
    {
      "epoch": 0.2180209171359614,
      "grad_norm": 0.7355032627907264,
      "learning_rate": 7.2654155495978565e-06,
      "loss": 0.061,
      "step": 813
    },
    {
      "epoch": 0.2182890855457227,
      "grad_norm": 0.7668956497469606,
      "learning_rate": 7.274352100089366e-06,
      "loss": 0.0505,
      "step": 814
    },
    {
      "epoch": 0.21855725395548403,
      "grad_norm": 0.4678948657869099,
      "learning_rate": 7.283288650580876e-06,
      "loss": 0.0388,
      "step": 815
    },
    {
      "epoch": 0.21882542236524538,
      "grad_norm": 0.6300577022663109,
      "learning_rate": 7.292225201072387e-06,
      "loss": 0.0629,
      "step": 816
    },
    {
      "epoch": 0.2190935907750067,
      "grad_norm": 0.8240611502389711,
      "learning_rate": 7.3011617515638964e-06,
      "loss": 0.0811,
      "step": 817
    },
    {
      "epoch": 0.21936175918476802,
      "grad_norm": 0.5334912625091875,
      "learning_rate": 7.3100983020554075e-06,
      "loss": 0.0597,
      "step": 818
    },
    {
      "epoch": 0.21962992759452937,
      "grad_norm": 0.9344188331640078,
      "learning_rate": 7.319034852546918e-06,
      "loss": 0.0584,
      "step": 819
    },
    {
      "epoch": 0.2198980960042907,
      "grad_norm": 0.6195094826631371,
      "learning_rate": 7.327971403038428e-06,
      "loss": 0.093,
      "step": 820
    },
    {
      "epoch": 0.22016626441405202,
      "grad_norm": 0.7350945497655491,
      "learning_rate": 7.336907953529938e-06,
      "loss": 0.056,
      "step": 821
    },
    {
      "epoch": 0.22043443282381336,
      "grad_norm": 0.7389468481777717,
      "learning_rate": 7.345844504021449e-06,
      "loss": 0.0609,
      "step": 822
    },
    {
      "epoch": 0.22070260123357469,
      "grad_norm": 0.5480749027899575,
      "learning_rate": 7.3547810545129585e-06,
      "loss": 0.0607,
      "step": 823
    },
    {
      "epoch": 0.220970769643336,
      "grad_norm": 0.6358348155909911,
      "learning_rate": 7.363717605004469e-06,
      "loss": 0.0579,
      "step": 824
    },
    {
      "epoch": 0.22123893805309736,
      "grad_norm": 0.9408099167799561,
      "learning_rate": 7.37265415549598e-06,
      "loss": 0.0679,
      "step": 825
    },
    {
      "epoch": 0.22150710646285868,
      "grad_norm": 0.5620699642124926,
      "learning_rate": 7.381590705987489e-06,
      "loss": 0.0449,
      "step": 826
    },
    {
      "epoch": 0.22177527487262,
      "grad_norm": 0.5964240925617175,
      "learning_rate": 7.390527256479e-06,
      "loss": 0.0489,
      "step": 827
    },
    {
      "epoch": 0.22204344328238135,
      "grad_norm": 0.6139468901504379,
      "learning_rate": 7.39946380697051e-06,
      "loss": 0.0548,
      "step": 828
    },
    {
      "epoch": 0.22231161169214267,
      "grad_norm": 0.6963174669383083,
      "learning_rate": 7.40840035746202e-06,
      "loss": 0.0489,
      "step": 829
    },
    {
      "epoch": 0.222579780101904,
      "grad_norm": 0.7612530912502669,
      "learning_rate": 7.417336907953531e-06,
      "loss": 0.0595,
      "step": 830
    },
    {
      "epoch": 0.22284794851166534,
      "grad_norm": 0.7539832432683371,
      "learning_rate": 7.426273458445042e-06,
      "loss": 0.0565,
      "step": 831
    },
    {
      "epoch": 0.22311611692142666,
      "grad_norm": 0.5610360101535783,
      "learning_rate": 7.435210008936551e-06,
      "loss": 0.0481,
      "step": 832
    },
    {
      "epoch": 0.22338428533118798,
      "grad_norm": 0.6708051272039379,
      "learning_rate": 7.444146559428061e-06,
      "loss": 0.0536,
      "step": 833
    },
    {
      "epoch": 0.2236524537409493,
      "grad_norm": 0.7948128165152702,
      "learning_rate": 7.453083109919572e-06,
      "loss": 0.0571,
      "step": 834
    },
    {
      "epoch": 0.22392062215071065,
      "grad_norm": 0.8119772632790323,
      "learning_rate": 7.462019660411082e-06,
      "loss": 0.0607,
      "step": 835
    },
    {
      "epoch": 0.22418879056047197,
      "grad_norm": 0.5938019231714236,
      "learning_rate": 7.470956210902593e-06,
      "loss": 0.0514,
      "step": 836
    },
    {
      "epoch": 0.2244569589702333,
      "grad_norm": 0.480339720088699,
      "learning_rate": 7.479892761394103e-06,
      "loss": 0.0572,
      "step": 837
    },
    {
      "epoch": 0.22472512737999464,
      "grad_norm": 0.7321419282499156,
      "learning_rate": 7.488829311885612e-06,
      "loss": 0.0614,
      "step": 838
    },
    {
      "epoch": 0.22499329578975596,
      "grad_norm": 0.7862174304581754,
      "learning_rate": 7.497765862377123e-06,
      "loss": 0.0597,
      "step": 839
    },
    {
      "epoch": 0.2252614641995173,
      "grad_norm": 0.6837691085703087,
      "learning_rate": 7.506702412868633e-06,
      "loss": 0.0571,
      "step": 840
    },
    {
      "epoch": 0.22552963260927864,
      "grad_norm": 0.5766328720356244,
      "learning_rate": 7.515638963360144e-06,
      "loss": 0.0612,
      "step": 841
    },
    {
      "epoch": 0.22579780101903996,
      "grad_norm": 0.8518640741431467,
      "learning_rate": 7.524575513851654e-06,
      "loss": 0.0652,
      "step": 842
    },
    {
      "epoch": 0.22606596942880128,
      "grad_norm": 0.6266510418612837,
      "learning_rate": 7.533512064343163e-06,
      "loss": 0.076,
      "step": 843
    },
    {
      "epoch": 0.22633413783856263,
      "grad_norm": 0.6071164995865543,
      "learning_rate": 7.542448614834674e-06,
      "loss": 0.0522,
      "step": 844
    },
    {
      "epoch": 0.22660230624832395,
      "grad_norm": 0.6121507959380281,
      "learning_rate": 7.551385165326185e-06,
      "loss": 0.0518,
      "step": 845
    },
    {
      "epoch": 0.22687047465808527,
      "grad_norm": 0.5731850553581667,
      "learning_rate": 7.560321715817695e-06,
      "loss": 0.0452,
      "step": 846
    },
    {
      "epoch": 0.22713864306784662,
      "grad_norm": 0.8535266943982462,
      "learning_rate": 7.569258266309205e-06,
      "loss": 0.0718,
      "step": 847
    },
    {
      "epoch": 0.22740681147760794,
      "grad_norm": 0.5806179329893834,
      "learning_rate": 7.578194816800716e-06,
      "loss": 0.0647,
      "step": 848
    },
    {
      "epoch": 0.22767497988736926,
      "grad_norm": 0.6007673294586532,
      "learning_rate": 7.587131367292225e-06,
      "loss": 0.0422,
      "step": 849
    },
    {
      "epoch": 0.2279431482971306,
      "grad_norm": 0.7635953689482369,
      "learning_rate": 7.596067917783736e-06,
      "loss": 0.0711,
      "step": 850
    },
    {
      "epoch": 0.22821131670689193,
      "grad_norm": 1.1178338935923162,
      "learning_rate": 7.6050044682752465e-06,
      "loss": 0.0692,
      "step": 851
    },
    {
      "epoch": 0.22847948511665325,
      "grad_norm": 0.5318873712770549,
      "learning_rate": 7.613941018766756e-06,
      "loss": 0.0521,
      "step": 852
    },
    {
      "epoch": 0.2287476535264146,
      "grad_norm": 0.8985212543648291,
      "learning_rate": 7.622877569258267e-06,
      "loss": 0.06,
      "step": 853
    },
    {
      "epoch": 0.22901582193617592,
      "grad_norm": 1.1797356361978766,
      "learning_rate": 7.631814119749778e-06,
      "loss": 0.0695,
      "step": 854
    },
    {
      "epoch": 0.22928399034593724,
      "grad_norm": 0.8184713676267756,
      "learning_rate": 7.640750670241287e-06,
      "loss": 0.0704,
      "step": 855
    },
    {
      "epoch": 0.22955215875569857,
      "grad_norm": 1.0384965914146926,
      "learning_rate": 7.649687220732798e-06,
      "loss": 0.0862,
      "step": 856
    },
    {
      "epoch": 0.22982032716545991,
      "grad_norm": 0.7610325618010567,
      "learning_rate": 7.658623771224308e-06,
      "loss": 0.0585,
      "step": 857
    },
    {
      "epoch": 0.23008849557522124,
      "grad_norm": 0.7263880463799376,
      "learning_rate": 7.667560321715819e-06,
      "loss": 0.0623,
      "step": 858
    },
    {
      "epoch": 0.23035666398498256,
      "grad_norm": 0.4807807023999299,
      "learning_rate": 7.676496872207328e-06,
      "loss": 0.0392,
      "step": 859
    },
    {
      "epoch": 0.2306248323947439,
      "grad_norm": 0.915275660737601,
      "learning_rate": 7.685433422698839e-06,
      "loss": 0.0793,
      "step": 860
    },
    {
      "epoch": 0.23089300080450523,
      "grad_norm": 0.6927968400855913,
      "learning_rate": 7.694369973190348e-06,
      "loss": 0.0564,
      "step": 861
    },
    {
      "epoch": 0.23116116921426655,
      "grad_norm": 0.6241870950750825,
      "learning_rate": 7.70330652368186e-06,
      "loss": 0.0479,
      "step": 862
    },
    {
      "epoch": 0.2314293376240279,
      "grad_norm": 0.7556368603029495,
      "learning_rate": 7.71224307417337e-06,
      "loss": 0.0559,
      "step": 863
    },
    {
      "epoch": 0.23169750603378922,
      "grad_norm": 0.7097977952793527,
      "learning_rate": 7.72117962466488e-06,
      "loss": 0.0659,
      "step": 864
    },
    {
      "epoch": 0.23196567444355054,
      "grad_norm": 0.5177822919638604,
      "learning_rate": 7.730116175156391e-06,
      "loss": 0.043,
      "step": 865
    },
    {
      "epoch": 0.2322338428533119,
      "grad_norm": 0.5847983833993663,
      "learning_rate": 7.7390527256479e-06,
      "loss": 0.0534,
      "step": 866
    },
    {
      "epoch": 0.2325020112630732,
      "grad_norm": 0.823173932051227,
      "learning_rate": 7.74798927613941e-06,
      "loss": 0.0745,
      "step": 867
    },
    {
      "epoch": 0.23277017967283453,
      "grad_norm": 0.5009681755231303,
      "learning_rate": 7.75692582663092e-06,
      "loss": 0.0528,
      "step": 868
    },
    {
      "epoch": 0.23303834808259588,
      "grad_norm": 1.0199982189387549,
      "learning_rate": 7.765862377122432e-06,
      "loss": 0.073,
      "step": 869
    },
    {
      "epoch": 0.2333065164923572,
      "grad_norm": 0.6737003312932051,
      "learning_rate": 7.774798927613941e-06,
      "loss": 0.0673,
      "step": 870
    },
    {
      "epoch": 0.23357468490211852,
      "grad_norm": 0.6752496192867327,
      "learning_rate": 7.783735478105452e-06,
      "loss": 0.041,
      "step": 871
    },
    {
      "epoch": 0.23384285331187987,
      "grad_norm": 0.5037915271983077,
      "learning_rate": 7.792672028596963e-06,
      "loss": 0.0401,
      "step": 872
    },
    {
      "epoch": 0.2341110217216412,
      "grad_norm": 0.8169239919037079,
      "learning_rate": 7.801608579088472e-06,
      "loss": 0.0685,
      "step": 873
    },
    {
      "epoch": 0.23437919013140252,
      "grad_norm": 0.8391500186631106,
      "learning_rate": 7.810545129579983e-06,
      "loss": 0.0545,
      "step": 874
    },
    {
      "epoch": 0.23464735854116386,
      "grad_norm": 0.8150667820952092,
      "learning_rate": 7.819481680071493e-06,
      "loss": 0.0652,
      "step": 875
    },
    {
      "epoch": 0.23491552695092519,
      "grad_norm": 0.6264426857860458,
      "learning_rate": 7.828418230563002e-06,
      "loss": 0.0547,
      "step": 876
    },
    {
      "epoch": 0.2351836953606865,
      "grad_norm": 0.8134112683694246,
      "learning_rate": 7.837354781054513e-06,
      "loss": 0.057,
      "step": 877
    },
    {
      "epoch": 0.23545186377044783,
      "grad_norm": 0.6083904370377538,
      "learning_rate": 7.846291331546024e-06,
      "loss": 0.0789,
      "step": 878
    },
    {
      "epoch": 0.23572003218020918,
      "grad_norm": 0.6001353311499135,
      "learning_rate": 7.855227882037534e-06,
      "loss": 0.0482,
      "step": 879
    },
    {
      "epoch": 0.2359882005899705,
      "grad_norm": 0.6620588703489341,
      "learning_rate": 7.864164432529045e-06,
      "loss": 0.0643,
      "step": 880
    },
    {
      "epoch": 0.23625636899973182,
      "grad_norm": 0.789291932526228,
      "learning_rate": 7.873100983020556e-06,
      "loss": 0.0633,
      "step": 881
    },
    {
      "epoch": 0.23652453740949317,
      "grad_norm": 0.5284387863787268,
      "learning_rate": 7.882037533512065e-06,
      "loss": 0.0502,
      "step": 882
    },
    {
      "epoch": 0.2367927058192545,
      "grad_norm": 0.6725341413669234,
      "learning_rate": 7.890974084003576e-06,
      "loss": 0.0453,
      "step": 883
    },
    {
      "epoch": 0.2370608742290158,
      "grad_norm": 0.524813557038549,
      "learning_rate": 7.899910634495085e-06,
      "loss": 0.0462,
      "step": 884
    },
    {
      "epoch": 0.23732904263877716,
      "grad_norm": 1.1488010368727157,
      "learning_rate": 7.908847184986595e-06,
      "loss": 0.0847,
      "step": 885
    },
    {
      "epoch": 0.23759721104853848,
      "grad_norm": 0.5537651629181534,
      "learning_rate": 7.917783735478106e-06,
      "loss": 0.0467,
      "step": 886
    },
    {
      "epoch": 0.2378653794582998,
      "grad_norm": 0.7749166298656239,
      "learning_rate": 7.926720285969617e-06,
      "loss": 0.0718,
      "step": 887
    },
    {
      "epoch": 0.23813354786806115,
      "grad_norm": 0.7645899977632732,
      "learning_rate": 7.935656836461126e-06,
      "loss": 0.069,
      "step": 888
    },
    {
      "epoch": 0.23840171627782247,
      "grad_norm": 0.6075901641135533,
      "learning_rate": 7.944593386952637e-06,
      "loss": 0.0655,
      "step": 889
    },
    {
      "epoch": 0.2386698846875838,
      "grad_norm": 0.8423462141079968,
      "learning_rate": 7.953529937444148e-06,
      "loss": 0.0801,
      "step": 890
    },
    {
      "epoch": 0.23893805309734514,
      "grad_norm": 0.6073224284984499,
      "learning_rate": 7.962466487935658e-06,
      "loss": 0.06,
      "step": 891
    },
    {
      "epoch": 0.23920622150710646,
      "grad_norm": 1.1648594726131332,
      "learning_rate": 7.971403038427169e-06,
      "loss": 0.0619,
      "step": 892
    },
    {
      "epoch": 0.23947438991686779,
      "grad_norm": 0.8909215026738083,
      "learning_rate": 7.980339588918678e-06,
      "loss": 0.0604,
      "step": 893
    },
    {
      "epoch": 0.23974255832662913,
      "grad_norm": 0.8066015283724232,
      "learning_rate": 7.989276139410187e-06,
      "loss": 0.0616,
      "step": 894
    },
    {
      "epoch": 0.24001072673639046,
      "grad_norm": 0.604755819715382,
      "learning_rate": 7.998212689901698e-06,
      "loss": 0.0557,
      "step": 895
    },
    {
      "epoch": 0.24027889514615178,
      "grad_norm": 0.7222841586906538,
      "learning_rate": 8.00714924039321e-06,
      "loss": 0.0585,
      "step": 896
    },
    {
      "epoch": 0.24054706355591313,
      "grad_norm": 0.7352634596043768,
      "learning_rate": 8.016085790884719e-06,
      "loss": 0.0494,
      "step": 897
    },
    {
      "epoch": 0.24081523196567445,
      "grad_norm": 0.7017740197602509,
      "learning_rate": 8.02502234137623e-06,
      "loss": 0.0526,
      "step": 898
    },
    {
      "epoch": 0.24108340037543577,
      "grad_norm": 0.6464193318091813,
      "learning_rate": 8.033958891867741e-06,
      "loss": 0.0443,
      "step": 899
    },
    {
      "epoch": 0.2413515687851971,
      "grad_norm": 0.7951271709068436,
      "learning_rate": 8.04289544235925e-06,
      "loss": 0.0574,
      "step": 900
    },
    {
      "epoch": 0.24161973719495844,
      "grad_norm": 0.6137734565500936,
      "learning_rate": 8.05183199285076e-06,
      "loss": 0.0438,
      "step": 901
    },
    {
      "epoch": 0.24188790560471976,
      "grad_norm": 0.7820226516609139,
      "learning_rate": 8.06076854334227e-06,
      "loss": 0.0545,
      "step": 902
    },
    {
      "epoch": 0.24215607401448108,
      "grad_norm": 0.563780104756091,
      "learning_rate": 8.06970509383378e-06,
      "loss": 0.0524,
      "step": 903
    },
    {
      "epoch": 0.24242424242424243,
      "grad_norm": 0.6931644539860238,
      "learning_rate": 8.078641644325291e-06,
      "loss": 0.0592,
      "step": 904
    },
    {
      "epoch": 0.24269241083400375,
      "grad_norm": 0.6859909114052662,
      "learning_rate": 8.0875781948168e-06,
      "loss": 0.0568,
      "step": 905
    },
    {
      "epoch": 0.24296057924376507,
      "grad_norm": 0.928319627995896,
      "learning_rate": 8.096514745308311e-06,
      "loss": 0.0869,
      "step": 906
    },
    {
      "epoch": 0.24322874765352642,
      "grad_norm": 0.691708928671463,
      "learning_rate": 8.105451295799822e-06,
      "loss": 0.0601,
      "step": 907
    },
    {
      "epoch": 0.24349691606328774,
      "grad_norm": 0.6665719493309162,
      "learning_rate": 8.114387846291332e-06,
      "loss": 0.0421,
      "step": 908
    },
    {
      "epoch": 0.24376508447304907,
      "grad_norm": 0.7438422072025176,
      "learning_rate": 8.123324396782843e-06,
      "loss": 0.0711,
      "step": 909
    },
    {
      "epoch": 0.24403325288281041,
      "grad_norm": 0.6446036821278024,
      "learning_rate": 8.132260947274352e-06,
      "loss": 0.0508,
      "step": 910
    },
    {
      "epoch": 0.24430142129257174,
      "grad_norm": 0.6005692800639328,
      "learning_rate": 8.141197497765863e-06,
      "loss": 0.0661,
      "step": 911
    },
    {
      "epoch": 0.24456958970233306,
      "grad_norm": 0.8938961327249468,
      "learning_rate": 8.150134048257373e-06,
      "loss": 0.0564,
      "step": 912
    },
    {
      "epoch": 0.2448377581120944,
      "grad_norm": 0.5968895734141487,
      "learning_rate": 8.159070598748884e-06,
      "loss": 0.0367,
      "step": 913
    },
    {
      "epoch": 0.24510592652185573,
      "grad_norm": 0.7871740531855175,
      "learning_rate": 8.168007149240393e-06,
      "loss": 0.0559,
      "step": 914
    },
    {
      "epoch": 0.24537409493161705,
      "grad_norm": 0.637551375796371,
      "learning_rate": 8.176943699731904e-06,
      "loss": 0.0501,
      "step": 915
    },
    {
      "epoch": 0.2456422633413784,
      "grad_norm": 0.8154387546175497,
      "learning_rate": 8.185880250223415e-06,
      "loss": 0.0687,
      "step": 916
    },
    {
      "epoch": 0.24591043175113972,
      "grad_norm": 0.586194381551852,
      "learning_rate": 8.194816800714924e-06,
      "loss": 0.0493,
      "step": 917
    },
    {
      "epoch": 0.24617860016090104,
      "grad_norm": 0.8807508292400504,
      "learning_rate": 8.203753351206435e-06,
      "loss": 0.0597,
      "step": 918
    },
    {
      "epoch": 0.2464467685706624,
      "grad_norm": 0.7654807499221806,
      "learning_rate": 8.212689901697945e-06,
      "loss": 0.0578,
      "step": 919
    },
    {
      "epoch": 0.2467149369804237,
      "grad_norm": 1.8894647306254642,
      "learning_rate": 8.221626452189456e-06,
      "loss": 0.0732,
      "step": 920
    },
    {
      "epoch": 0.24698310539018503,
      "grad_norm": 0.8042343108910344,
      "learning_rate": 8.230563002680965e-06,
      "loss": 0.0603,
      "step": 921
    },
    {
      "epoch": 0.24725127379994635,
      "grad_norm": 0.5974569485535882,
      "learning_rate": 8.239499553172476e-06,
      "loss": 0.0555,
      "step": 922
    },
    {
      "epoch": 0.2475194422097077,
      "grad_norm": 0.6081532476159442,
      "learning_rate": 8.248436103663986e-06,
      "loss": 0.0415,
      "step": 923
    },
    {
      "epoch": 0.24778761061946902,
      "grad_norm": 0.672627802864003,
      "learning_rate": 8.257372654155497e-06,
      "loss": 0.0601,
      "step": 924
    },
    {
      "epoch": 0.24805577902923034,
      "grad_norm": 0.7018590878081592,
      "learning_rate": 8.266309204647008e-06,
      "loss": 0.058,
      "step": 925
    },
    {
      "epoch": 0.2483239474389917,
      "grad_norm": 0.6022603429913677,
      "learning_rate": 8.275245755138517e-06,
      "loss": 0.0574,
      "step": 926
    },
    {
      "epoch": 0.24859211584875301,
      "grad_norm": 0.7660595830126783,
      "learning_rate": 8.284182305630028e-06,
      "loss": 0.0543,
      "step": 927
    },
    {
      "epoch": 0.24886028425851434,
      "grad_norm": 0.8697793367321248,
      "learning_rate": 8.293118856121537e-06,
      "loss": 0.0647,
      "step": 928
    },
    {
      "epoch": 0.24912845266827569,
      "grad_norm": 0.7249985205243299,
      "learning_rate": 8.302055406613048e-06,
      "loss": 0.0578,
      "step": 929
    },
    {
      "epoch": 0.249396621078037,
      "grad_norm": 0.790442464060151,
      "learning_rate": 8.310991957104558e-06,
      "loss": 0.0514,
      "step": 930
    },
    {
      "epoch": 0.24966478948779833,
      "grad_norm": 0.6139450930385338,
      "learning_rate": 8.319928507596069e-06,
      "loss": 0.0462,
      "step": 931
    },
    {
      "epoch": 0.24993295789755968,
      "grad_norm": 0.6896407577807269,
      "learning_rate": 8.328865058087578e-06,
      "loss": 0.0586,
      "step": 932
    },
    {
      "epoch": 0.250201126307321,
      "grad_norm": 0.7650698727407607,
      "learning_rate": 8.33780160857909e-06,
      "loss": 0.0634,
      "step": 933
    },
    {
      "epoch": 0.25046929471708235,
      "grad_norm": 0.4670828956333003,
      "learning_rate": 8.3467381590706e-06,
      "loss": 0.0362,
      "step": 934
    },
    {
      "epoch": 0.25073746312684364,
      "grad_norm": 0.6958426960844856,
      "learning_rate": 8.35567470956211e-06,
      "loss": 0.0573,
      "step": 935
    },
    {
      "epoch": 0.251005631536605,
      "grad_norm": 0.8764140587163858,
      "learning_rate": 8.36461126005362e-06,
      "loss": 0.0453,
      "step": 936
    },
    {
      "epoch": 0.25127379994636634,
      "grad_norm": 0.7673356156776654,
      "learning_rate": 8.37354781054513e-06,
      "loss": 0.0681,
      "step": 937
    },
    {
      "epoch": 0.25154196835612763,
      "grad_norm": 0.4478473814274472,
      "learning_rate": 8.38248436103664e-06,
      "loss": 0.0404,
      "step": 938
    },
    {
      "epoch": 0.251810136765889,
      "grad_norm": 0.668969094249436,
      "learning_rate": 8.39142091152815e-06,
      "loss": 0.0652,
      "step": 939
    },
    {
      "epoch": 0.25207830517565033,
      "grad_norm": 0.7538091479871053,
      "learning_rate": 8.400357462019661e-06,
      "loss": 0.0676,
      "step": 940
    },
    {
      "epoch": 0.2523464735854116,
      "grad_norm": 0.4333366253309145,
      "learning_rate": 8.40929401251117e-06,
      "loss": 0.0456,
      "step": 941
    },
    {
      "epoch": 0.252614641995173,
      "grad_norm": 0.7700579996556718,
      "learning_rate": 8.418230563002682e-06,
      "loss": 0.0641,
      "step": 942
    },
    {
      "epoch": 0.2528828104049343,
      "grad_norm": 0.5276077661127693,
      "learning_rate": 8.427167113494193e-06,
      "loss": 0.0492,
      "step": 943
    },
    {
      "epoch": 0.2531509788146956,
      "grad_norm": 0.6869835500427105,
      "learning_rate": 8.436103663985702e-06,
      "loss": 0.0503,
      "step": 944
    },
    {
      "epoch": 0.25341914722445696,
      "grad_norm": 1.2973263013390004,
      "learning_rate": 8.445040214477213e-06,
      "loss": 0.0574,
      "step": 945
    },
    {
      "epoch": 0.2536873156342183,
      "grad_norm": 0.6292400183413236,
      "learning_rate": 8.453976764968723e-06,
      "loss": 0.0494,
      "step": 946
    },
    {
      "epoch": 0.2539554840439796,
      "grad_norm": 0.8170201856506285,
      "learning_rate": 8.462913315460232e-06,
      "loss": 0.069,
      "step": 947
    },
    {
      "epoch": 0.25422365245374096,
      "grad_norm": 0.8883212693708068,
      "learning_rate": 8.471849865951743e-06,
      "loss": 0.0633,
      "step": 948
    },
    {
      "epoch": 0.2544918208635023,
      "grad_norm": 0.7630928207909858,
      "learning_rate": 8.480786416443254e-06,
      "loss": 0.0582,
      "step": 949
    },
    {
      "epoch": 0.2547599892732636,
      "grad_norm": 0.6126004445222877,
      "learning_rate": 8.489722966934763e-06,
      "loss": 0.0546,
      "step": 950
    },
    {
      "epoch": 0.25502815768302495,
      "grad_norm": 1.5122336589806007,
      "learning_rate": 8.498659517426274e-06,
      "loss": 0.0673,
      "step": 951
    },
    {
      "epoch": 0.2552963260927863,
      "grad_norm": 0.6899019681036999,
      "learning_rate": 8.507596067917786e-06,
      "loss": 0.0538,
      "step": 952
    },
    {
      "epoch": 0.2555644945025476,
      "grad_norm": 0.5130942519524846,
      "learning_rate": 8.516532618409295e-06,
      "loss": 0.0465,
      "step": 953
    },
    {
      "epoch": 0.25583266291230894,
      "grad_norm": 1.0530481008505155,
      "learning_rate": 8.525469168900806e-06,
      "loss": 0.0605,
      "step": 954
    },
    {
      "epoch": 0.25610083132207023,
      "grad_norm": 0.518302535721877,
      "learning_rate": 8.534405719392315e-06,
      "loss": 0.0421,
      "step": 955
    },
    {
      "epoch": 0.2563689997318316,
      "grad_norm": 0.7573322925643349,
      "learning_rate": 8.543342269883825e-06,
      "loss": 0.0699,
      "step": 956
    },
    {
      "epoch": 0.25663716814159293,
      "grad_norm": 1.2476866653178869,
      "learning_rate": 8.552278820375336e-06,
      "loss": 0.0683,
      "step": 957
    },
    {
      "epoch": 0.2569053365513542,
      "grad_norm": 0.4914396775913551,
      "learning_rate": 8.561215370866847e-06,
      "loss": 0.0543,
      "step": 958
    },
    {
      "epoch": 0.2571735049611156,
      "grad_norm": 0.8061720447165363,
      "learning_rate": 8.570151921358356e-06,
      "loss": 0.0685,
      "step": 959
    },
    {
      "epoch": 0.2574416733708769,
      "grad_norm": 0.5857652117623684,
      "learning_rate": 8.579088471849867e-06,
      "loss": 0.045,
      "step": 960
    },
    {
      "epoch": 0.2577098417806382,
      "grad_norm": 0.5909047883637759,
      "learning_rate": 8.588025022341378e-06,
      "loss": 0.0612,
      "step": 961
    },
    {
      "epoch": 0.25797801019039956,
      "grad_norm": 1.0393365434663968,
      "learning_rate": 8.596961572832887e-06,
      "loss": 0.0721,
      "step": 962
    },
    {
      "epoch": 0.2582461786001609,
      "grad_norm": 1.051358034922522,
      "learning_rate": 8.605898123324398e-06,
      "loss": 0.0649,
      "step": 963
    },
    {
      "epoch": 0.2585143470099222,
      "grad_norm": 0.9210418580735691,
      "learning_rate": 8.614834673815908e-06,
      "loss": 0.0687,
      "step": 964
    },
    {
      "epoch": 0.25878251541968356,
      "grad_norm": 0.7566778738587868,
      "learning_rate": 8.623771224307417e-06,
      "loss": 0.0476,
      "step": 965
    },
    {
      "epoch": 0.2590506838294449,
      "grad_norm": 0.721868408774214,
      "learning_rate": 8.632707774798928e-06,
      "loss": 0.0529,
      "step": 966
    },
    {
      "epoch": 0.2593188522392062,
      "grad_norm": 0.5634817468783463,
      "learning_rate": 8.641644325290438e-06,
      "loss": 0.0468,
      "step": 967
    },
    {
      "epoch": 0.25958702064896755,
      "grad_norm": 0.6019583463099332,
      "learning_rate": 8.650580875781949e-06,
      "loss": 0.0392,
      "step": 968
    },
    {
      "epoch": 0.2598551890587289,
      "grad_norm": 0.652480640490992,
      "learning_rate": 8.65951742627346e-06,
      "loss": 0.0551,
      "step": 969
    },
    {
      "epoch": 0.2601233574684902,
      "grad_norm": 0.6721375058000493,
      "learning_rate": 8.668453976764969e-06,
      "loss": 0.0465,
      "step": 970
    },
    {
      "epoch": 0.26039152587825154,
      "grad_norm": 0.9328464996839846,
      "learning_rate": 8.67739052725648e-06,
      "loss": 0.0801,
      "step": 971
    },
    {
      "epoch": 0.2606596942880129,
      "grad_norm": 0.7857595012352139,
      "learning_rate": 8.68632707774799e-06,
      "loss": 0.0739,
      "step": 972
    },
    {
      "epoch": 0.2609278626977742,
      "grad_norm": 0.6447579404961109,
      "learning_rate": 8.6952636282395e-06,
      "loss": 0.0555,
      "step": 973
    },
    {
      "epoch": 0.26119603110753553,
      "grad_norm": 0.5577057236734498,
      "learning_rate": 8.70420017873101e-06,
      "loss": 0.0451,
      "step": 974
    },
    {
      "epoch": 0.2614641995172969,
      "grad_norm": 1.2135674538071934,
      "learning_rate": 8.71313672922252e-06,
      "loss": 0.062,
      "step": 975
    },
    {
      "epoch": 0.2617323679270582,
      "grad_norm": 0.8310745766948165,
      "learning_rate": 8.72207327971403e-06,
      "loss": 0.0938,
      "step": 976
    },
    {
      "epoch": 0.2620005363368195,
      "grad_norm": 0.6158366022939997,
      "learning_rate": 8.731009830205541e-06,
      "loss": 0.0592,
      "step": 977
    },
    {
      "epoch": 0.26226870474658087,
      "grad_norm": 0.6359645763327832,
      "learning_rate": 8.739946380697052e-06,
      "loss": 0.0494,
      "step": 978
    },
    {
      "epoch": 0.26253687315634217,
      "grad_norm": 1.0502437345759257,
      "learning_rate": 8.748882931188562e-06,
      "loss": 0.0911,
      "step": 979
    },
    {
      "epoch": 0.2628050415661035,
      "grad_norm": 0.6175171276587017,
      "learning_rate": 8.757819481680073e-06,
      "loss": 0.0539,
      "step": 980
    },
    {
      "epoch": 0.26307320997586486,
      "grad_norm": 0.6749708959577044,
      "learning_rate": 8.766756032171582e-06,
      "loss": 0.0634,
      "step": 981
    },
    {
      "epoch": 0.26334137838562616,
      "grad_norm": 0.49555002188692787,
      "learning_rate": 8.775692582663093e-06,
      "loss": 0.0587,
      "step": 982
    },
    {
      "epoch": 0.2636095467953875,
      "grad_norm": 0.8687916311240435,
      "learning_rate": 8.784629133154602e-06,
      "loss": 0.0552,
      "step": 983
    },
    {
      "epoch": 0.26387771520514886,
      "grad_norm": 0.5543685744511956,
      "learning_rate": 8.793565683646113e-06,
      "loss": 0.0503,
      "step": 984
    },
    {
      "epoch": 0.26414588361491015,
      "grad_norm": 0.9107328064719851,
      "learning_rate": 8.802502234137623e-06,
      "loss": 0.0768,
      "step": 985
    },
    {
      "epoch": 0.2644140520246715,
      "grad_norm": 0.9027915874226776,
      "learning_rate": 8.811438784629134e-06,
      "loss": 0.0595,
      "step": 986
    },
    {
      "epoch": 0.26468222043443285,
      "grad_norm": 0.6522561716767531,
      "learning_rate": 8.820375335120645e-06,
      "loss": 0.0417,
      "step": 987
    },
    {
      "epoch": 0.26495038884419414,
      "grad_norm": 0.4962857610158182,
      "learning_rate": 8.829311885612154e-06,
      "loss": 0.0415,
      "step": 988
    },
    {
      "epoch": 0.2652185572539555,
      "grad_norm": 0.6728417980795768,
      "learning_rate": 8.838248436103665e-06,
      "loss": 0.0582,
      "step": 989
    },
    {
      "epoch": 0.26548672566371684,
      "grad_norm": 0.7842330537946006,
      "learning_rate": 8.847184986595175e-06,
      "loss": 0.064,
      "step": 990
    },
    {
      "epoch": 0.26575489407347813,
      "grad_norm": 0.6175498477543568,
      "learning_rate": 8.856121537086686e-06,
      "loss": 0.0411,
      "step": 991
    },
    {
      "epoch": 0.2660230624832395,
      "grad_norm": 0.658647518155521,
      "learning_rate": 8.865058087578195e-06,
      "loss": 0.0657,
      "step": 992
    },
    {
      "epoch": 0.26629123089300083,
      "grad_norm": 0.6864937394665676,
      "learning_rate": 8.873994638069706e-06,
      "loss": 0.0478,
      "step": 993
    },
    {
      "epoch": 0.2665593993027621,
      "grad_norm": 0.6185926792241573,
      "learning_rate": 8.882931188561215e-06,
      "loss": 0.0502,
      "step": 994
    },
    {
      "epoch": 0.2668275677125235,
      "grad_norm": 0.6393543357525548,
      "learning_rate": 8.891867739052726e-06,
      "loss": 0.0651,
      "step": 995
    },
    {
      "epoch": 0.2670957361222848,
      "grad_norm": 0.6194030557164245,
      "learning_rate": 8.900804289544237e-06,
      "loss": 0.0533,
      "step": 996
    },
    {
      "epoch": 0.2673639045320461,
      "grad_norm": 0.8608883766157296,
      "learning_rate": 8.909740840035747e-06,
      "loss": 0.0701,
      "step": 997
    },
    {
      "epoch": 0.26763207294180746,
      "grad_norm": 0.6945023804738137,
      "learning_rate": 8.918677390527258e-06,
      "loss": 0.0741,
      "step": 998
    },
    {
      "epoch": 0.26790024135156876,
      "grad_norm": 0.5843338370911121,
      "learning_rate": 8.927613941018767e-06,
      "loss": 0.0495,
      "step": 999
    },
    {
      "epoch": 0.2681684097613301,
      "grad_norm": 0.6206738894298448,
      "learning_rate": 8.936550491510278e-06,
      "loss": 0.061,
      "step": 1000
    },
    {
      "epoch": 0.26843657817109146,
      "grad_norm": 0.7964771863924608,
      "learning_rate": 8.945487042001788e-06,
      "loss": 0.0706,
      "step": 1001
    },
    {
      "epoch": 0.26870474658085275,
      "grad_norm": 0.8235197971875386,
      "learning_rate": 8.954423592493299e-06,
      "loss": 0.0656,
      "step": 1002
    },
    {
      "epoch": 0.2689729149906141,
      "grad_norm": 0.5037490913995571,
      "learning_rate": 8.963360142984808e-06,
      "loss": 0.0418,
      "step": 1003
    },
    {
      "epoch": 0.26924108340037545,
      "grad_norm": 1.048784315575701,
      "learning_rate": 8.972296693476319e-06,
      "loss": 0.0741,
      "step": 1004
    },
    {
      "epoch": 0.26950925181013674,
      "grad_norm": 0.5853159150447301,
      "learning_rate": 8.98123324396783e-06,
      "loss": 0.0531,
      "step": 1005
    },
    {
      "epoch": 0.2697774202198981,
      "grad_norm": 0.7691364112473073,
      "learning_rate": 8.99016979445934e-06,
      "loss": 0.048,
      "step": 1006
    },
    {
      "epoch": 0.27004558862965944,
      "grad_norm": 0.6306331783167517,
      "learning_rate": 8.99910634495085e-06,
      "loss": 0.0725,
      "step": 1007
    },
    {
      "epoch": 0.27031375703942073,
      "grad_norm": 0.5081907019308503,
      "learning_rate": 9.00804289544236e-06,
      "loss": 0.048,
      "step": 1008
    },
    {
      "epoch": 0.2705819254491821,
      "grad_norm": 0.7247969216222044,
      "learning_rate": 9.01697944593387e-06,
      "loss": 0.0626,
      "step": 1009
    },
    {
      "epoch": 0.27085009385894343,
      "grad_norm": 0.5182847795356663,
      "learning_rate": 9.02591599642538e-06,
      "loss": 0.0453,
      "step": 1010
    },
    {
      "epoch": 0.2711182622687047,
      "grad_norm": 0.621615755971977,
      "learning_rate": 9.034852546916891e-06,
      "loss": 0.0415,
      "step": 1011
    },
    {
      "epoch": 0.2713864306784661,
      "grad_norm": 0.6607938541868332,
      "learning_rate": 9.0437890974084e-06,
      "loss": 0.0448,
      "step": 1012
    },
    {
      "epoch": 0.2716545990882274,
      "grad_norm": 0.5557200661827993,
      "learning_rate": 9.052725647899912e-06,
      "loss": 0.0565,
      "step": 1013
    },
    {
      "epoch": 0.2719227674979887,
      "grad_norm": 0.5558571489023096,
      "learning_rate": 9.061662198391423e-06,
      "loss": 0.0483,
      "step": 1014
    },
    {
      "epoch": 0.27219093590775006,
      "grad_norm": 0.48628039931342637,
      "learning_rate": 9.070598748882932e-06,
      "loss": 0.0479,
      "step": 1015
    },
    {
      "epoch": 0.2724591043175114,
      "grad_norm": 0.4817534033592412,
      "learning_rate": 9.079535299374443e-06,
      "loss": 0.0536,
      "step": 1016
    },
    {
      "epoch": 0.2727272727272727,
      "grad_norm": 0.6634588409570141,
      "learning_rate": 9.088471849865952e-06,
      "loss": 0.0511,
      "step": 1017
    },
    {
      "epoch": 0.27299544113703406,
      "grad_norm": 0.5068500419797295,
      "learning_rate": 9.097408400357462e-06,
      "loss": 0.0526,
      "step": 1018
    },
    {
      "epoch": 0.2732636095467954,
      "grad_norm": 0.7080402718833271,
      "learning_rate": 9.106344950848973e-06,
      "loss": 0.0773,
      "step": 1019
    },
    {
      "epoch": 0.2735317779565567,
      "grad_norm": 0.7437439716392791,
      "learning_rate": 9.115281501340484e-06,
      "loss": 0.0772,
      "step": 1020
    },
    {
      "epoch": 0.27379994636631805,
      "grad_norm": 0.42275638927373943,
      "learning_rate": 9.124218051831993e-06,
      "loss": 0.0501,
      "step": 1021
    },
    {
      "epoch": 0.2740681147760794,
      "grad_norm": 0.4980086095522623,
      "learning_rate": 9.133154602323504e-06,
      "loss": 0.0508,
      "step": 1022
    },
    {
      "epoch": 0.2743362831858407,
      "grad_norm": 0.8111532213277497,
      "learning_rate": 9.142091152815015e-06,
      "loss": 0.0666,
      "step": 1023
    },
    {
      "epoch": 0.27460445159560204,
      "grad_norm": 0.6977891408528311,
      "learning_rate": 9.151027703306525e-06,
      "loss": 0.0462,
      "step": 1024
    },
    {
      "epoch": 0.2748726200053634,
      "grad_norm": 0.5020957373738434,
      "learning_rate": 9.159964253798036e-06,
      "loss": 0.0472,
      "step": 1025
    },
    {
      "epoch": 0.2751407884151247,
      "grad_norm": 0.7721213385558459,
      "learning_rate": 9.168900804289545e-06,
      "loss": 0.0758,
      "step": 1026
    },
    {
      "epoch": 0.27540895682488603,
      "grad_norm": 0.48250191303558965,
      "learning_rate": 9.177837354781054e-06,
      "loss": 0.0362,
      "step": 1027
    },
    {
      "epoch": 0.2756771252346474,
      "grad_norm": 0.9170517547245687,
      "learning_rate": 9.186773905272565e-06,
      "loss": 0.0562,
      "step": 1028
    },
    {
      "epoch": 0.2759452936444087,
      "grad_norm": 0.7780528461956794,
      "learning_rate": 9.195710455764075e-06,
      "loss": 0.0512,
      "step": 1029
    },
    {
      "epoch": 0.27621346205417,
      "grad_norm": 0.8568728483540402,
      "learning_rate": 9.204647006255586e-06,
      "loss": 0.065,
      "step": 1030
    },
    {
      "epoch": 0.27648163046393137,
      "grad_norm": 0.7231529764681266,
      "learning_rate": 9.213583556747097e-06,
      "loss": 0.0504,
      "step": 1031
    },
    {
      "epoch": 0.27674979887369267,
      "grad_norm": 1.4510960016837429,
      "learning_rate": 9.222520107238606e-06,
      "loss": 0.0678,
      "step": 1032
    },
    {
      "epoch": 0.277017967283454,
      "grad_norm": 0.653576965736248,
      "learning_rate": 9.231456657730117e-06,
      "loss": 0.0469,
      "step": 1033
    },
    {
      "epoch": 0.27728613569321536,
      "grad_norm": 1.0757416177249457,
      "learning_rate": 9.240393208221627e-06,
      "loss": 0.0756,
      "step": 1034
    },
    {
      "epoch": 0.27755430410297666,
      "grad_norm": 0.5098378843646046,
      "learning_rate": 9.249329758713138e-06,
      "loss": 0.0411,
      "step": 1035
    },
    {
      "epoch": 0.277822472512738,
      "grad_norm": 0.9646397874228843,
      "learning_rate": 9.258266309204647e-06,
      "loss": 0.0829,
      "step": 1036
    },
    {
      "epoch": 0.27809064092249935,
      "grad_norm": 0.649432006245567,
      "learning_rate": 9.267202859696158e-06,
      "loss": 0.047,
      "step": 1037
    },
    {
      "epoch": 0.27835880933226065,
      "grad_norm": 0.639605357977057,
      "learning_rate": 9.276139410187667e-06,
      "loss": 0.0719,
      "step": 1038
    },
    {
      "epoch": 0.278626977742022,
      "grad_norm": 0.7921143678019575,
      "learning_rate": 9.285075960679178e-06,
      "loss": 0.089,
      "step": 1039
    },
    {
      "epoch": 0.27889514615178335,
      "grad_norm": 0.5956386826367281,
      "learning_rate": 9.29401251117069e-06,
      "loss": 0.052,
      "step": 1040
    },
    {
      "epoch": 0.27916331456154464,
      "grad_norm": 0.9116127392221476,
      "learning_rate": 9.302949061662199e-06,
      "loss": 0.0648,
      "step": 1041
    },
    {
      "epoch": 0.279431482971306,
      "grad_norm": 0.7520329878990218,
      "learning_rate": 9.31188561215371e-06,
      "loss": 0.0464,
      "step": 1042
    },
    {
      "epoch": 0.27969965138106734,
      "grad_norm": 0.6022048551338887,
      "learning_rate": 9.32082216264522e-06,
      "loss": 0.0621,
      "step": 1043
    },
    {
      "epoch": 0.27996781979082863,
      "grad_norm": 1.393949141929977,
      "learning_rate": 9.32975871313673e-06,
      "loss": 0.0803,
      "step": 1044
    },
    {
      "epoch": 0.28023598820059,
      "grad_norm": 0.6495763689052004,
      "learning_rate": 9.33869526362824e-06,
      "loss": 0.0553,
      "step": 1045
    },
    {
      "epoch": 0.2805041566103513,
      "grad_norm": 0.6746904093844857,
      "learning_rate": 9.34763181411975e-06,
      "loss": 0.0553,
      "step": 1046
    },
    {
      "epoch": 0.2807723250201126,
      "grad_norm": 0.552107745573398,
      "learning_rate": 9.35656836461126e-06,
      "loss": 0.0529,
      "step": 1047
    },
    {
      "epoch": 0.28104049342987397,
      "grad_norm": 0.6497245971155569,
      "learning_rate": 9.365504915102771e-06,
      "loss": 0.0499,
      "step": 1048
    },
    {
      "epoch": 0.28130866183963527,
      "grad_norm": 0.7272730621428664,
      "learning_rate": 9.374441465594282e-06,
      "loss": 0.0448,
      "step": 1049
    },
    {
      "epoch": 0.2815768302493966,
      "grad_norm": 0.6484358015409448,
      "learning_rate": 9.383378016085791e-06,
      "loss": 0.0604,
      "step": 1050
    },
    {
      "epoch": 0.28184499865915796,
      "grad_norm": 0.7510489590503887,
      "learning_rate": 9.392314566577302e-06,
      "loss": 0.0514,
      "step": 1051
    },
    {
      "epoch": 0.28211316706891926,
      "grad_norm": 0.5371802431393131,
      "learning_rate": 9.401251117068812e-06,
      "loss": 0.0435,
      "step": 1052
    },
    {
      "epoch": 0.2823813354786806,
      "grad_norm": 0.47874966320861945,
      "learning_rate": 9.410187667560323e-06,
      "loss": 0.0542,
      "step": 1053
    },
    {
      "epoch": 0.28264950388844196,
      "grad_norm": 0.9621708383271833,
      "learning_rate": 9.419124218051832e-06,
      "loss": 0.0564,
      "step": 1054
    },
    {
      "epoch": 0.28291767229820325,
      "grad_norm": 0.6323614411408063,
      "learning_rate": 9.428060768543343e-06,
      "loss": 0.0453,
      "step": 1055
    },
    {
      "epoch": 0.2831858407079646,
      "grad_norm": 0.5484857293081461,
      "learning_rate": 9.436997319034853e-06,
      "loss": 0.0383,
      "step": 1056
    },
    {
      "epoch": 0.28345400911772595,
      "grad_norm": 0.5372153442667412,
      "learning_rate": 9.445933869526364e-06,
      "loss": 0.0519,
      "step": 1057
    },
    {
      "epoch": 0.28372217752748724,
      "grad_norm": 0.7419231869336939,
      "learning_rate": 9.454870420017875e-06,
      "loss": 0.0434,
      "step": 1058
    },
    {
      "epoch": 0.2839903459372486,
      "grad_norm": 0.59402491157313,
      "learning_rate": 9.463806970509384e-06,
      "loss": 0.0504,
      "step": 1059
    },
    {
      "epoch": 0.28425851434700994,
      "grad_norm": 0.930869987585662,
      "learning_rate": 9.472743521000895e-06,
      "loss": 0.0656,
      "step": 1060
    },
    {
      "epoch": 0.28452668275677123,
      "grad_norm": 0.6486607995316331,
      "learning_rate": 9.481680071492404e-06,
      "loss": 0.0564,
      "step": 1061
    },
    {
      "epoch": 0.2847948511665326,
      "grad_norm": 0.6712834828663943,
      "learning_rate": 9.490616621983915e-06,
      "loss": 0.0618,
      "step": 1062
    },
    {
      "epoch": 0.28506301957629393,
      "grad_norm": 1.3713518521465855,
      "learning_rate": 9.499553172475425e-06,
      "loss": 0.0797,
      "step": 1063
    },
    {
      "epoch": 0.2853311879860552,
      "grad_norm": 0.603819232680413,
      "learning_rate": 9.508489722966936e-06,
      "loss": 0.0531,
      "step": 1064
    },
    {
      "epoch": 0.2855993563958166,
      "grad_norm": 0.5956016266402101,
      "learning_rate": 9.517426273458445e-06,
      "loss": 0.0614,
      "step": 1065
    },
    {
      "epoch": 0.2858675248055779,
      "grad_norm": 0.46252523601941165,
      "learning_rate": 9.526362823949956e-06,
      "loss": 0.0421,
      "step": 1066
    },
    {
      "epoch": 0.2861356932153392,
      "grad_norm": 0.731060264461064,
      "learning_rate": 9.535299374441467e-06,
      "loss": 0.0626,
      "step": 1067
    },
    {
      "epoch": 0.28640386162510056,
      "grad_norm": 0.5824600238051167,
      "learning_rate": 9.544235924932977e-06,
      "loss": 0.0435,
      "step": 1068
    },
    {
      "epoch": 0.2866720300348619,
      "grad_norm": 0.5954843312453503,
      "learning_rate": 9.553172475424488e-06,
      "loss": 0.0489,
      "step": 1069
    },
    {
      "epoch": 0.2869401984446232,
      "grad_norm": 1.176109853012195,
      "learning_rate": 9.562109025915997e-06,
      "loss": 0.065,
      "step": 1070
    },
    {
      "epoch": 0.28720836685438456,
      "grad_norm": 0.5634419748716052,
      "learning_rate": 9.571045576407506e-06,
      "loss": 0.0649,
      "step": 1071
    },
    {
      "epoch": 0.2874765352641459,
      "grad_norm": 0.5857694096213143,
      "learning_rate": 9.579982126899017e-06,
      "loss": 0.0503,
      "step": 1072
    },
    {
      "epoch": 0.2877447036739072,
      "grad_norm": 0.6608186945724066,
      "learning_rate": 9.588918677390528e-06,
      "loss": 0.0578,
      "step": 1073
    },
    {
      "epoch": 0.28801287208366855,
      "grad_norm": 0.762027099211578,
      "learning_rate": 9.597855227882038e-06,
      "loss": 0.0654,
      "step": 1074
    },
    {
      "epoch": 0.2882810404934299,
      "grad_norm": 0.46947107649955794,
      "learning_rate": 9.606791778373549e-06,
      "loss": 0.051,
      "step": 1075
    },
    {
      "epoch": 0.2885492089031912,
      "grad_norm": 0.5221337872140267,
      "learning_rate": 9.61572832886506e-06,
      "loss": 0.0515,
      "step": 1076
    },
    {
      "epoch": 0.28881737731295254,
      "grad_norm": 0.9578724975879174,
      "learning_rate": 9.62466487935657e-06,
      "loss": 0.0443,
      "step": 1077
    },
    {
      "epoch": 0.2890855457227139,
      "grad_norm": 0.5763965641719264,
      "learning_rate": 9.63360142984808e-06,
      "loss": 0.0451,
      "step": 1078
    },
    {
      "epoch": 0.2893537141324752,
      "grad_norm": 0.7010837187212227,
      "learning_rate": 9.64253798033959e-06,
      "loss": 0.0521,
      "step": 1079
    },
    {
      "epoch": 0.28962188254223653,
      "grad_norm": 0.4707298750872888,
      "learning_rate": 9.651474530831099e-06,
      "loss": 0.0447,
      "step": 1080
    },
    {
      "epoch": 0.2898900509519979,
      "grad_norm": 0.8713811621183434,
      "learning_rate": 9.66041108132261e-06,
      "loss": 0.0606,
      "step": 1081
    },
    {
      "epoch": 0.2901582193617592,
      "grad_norm": 0.8172399257965689,
      "learning_rate": 9.669347631814121e-06,
      "loss": 0.0542,
      "step": 1082
    },
    {
      "epoch": 0.2904263877715205,
      "grad_norm": 0.8277305311428933,
      "learning_rate": 9.67828418230563e-06,
      "loss": 0.0465,
      "step": 1083
    },
    {
      "epoch": 0.29069455618128187,
      "grad_norm": 0.5957409430682489,
      "learning_rate": 9.687220732797141e-06,
      "loss": 0.0509,
      "step": 1084
    },
    {
      "epoch": 0.29096272459104316,
      "grad_norm": 0.7388381696095824,
      "learning_rate": 9.696157283288652e-06,
      "loss": 0.0627,
      "step": 1085
    },
    {
      "epoch": 0.2912308930008045,
      "grad_norm": 0.9208301495223172,
      "learning_rate": 9.705093833780162e-06,
      "loss": 0.0498,
      "step": 1086
    },
    {
      "epoch": 0.29149906141056586,
      "grad_norm": 0.6326590013322599,
      "learning_rate": 9.714030384271673e-06,
      "loss": 0.0441,
      "step": 1087
    },
    {
      "epoch": 0.29176722982032716,
      "grad_norm": 0.6967249536075673,
      "learning_rate": 9.722966934763182e-06,
      "loss": 0.0597,
      "step": 1088
    },
    {
      "epoch": 0.2920353982300885,
      "grad_norm": 1.1541930624146808,
      "learning_rate": 9.731903485254692e-06,
      "loss": 0.0584,
      "step": 1089
    },
    {
      "epoch": 0.2923035666398498,
      "grad_norm": 0.7784515320622181,
      "learning_rate": 9.740840035746203e-06,
      "loss": 0.0426,
      "step": 1090
    },
    {
      "epoch": 0.29257173504961115,
      "grad_norm": 0.7452996845248616,
      "learning_rate": 9.749776586237712e-06,
      "loss": 0.0697,
      "step": 1091
    },
    {
      "epoch": 0.2928399034593725,
      "grad_norm": 1.0257377184335728,
      "learning_rate": 9.758713136729223e-06,
      "loss": 0.0549,
      "step": 1092
    },
    {
      "epoch": 0.2931080718691338,
      "grad_norm": 0.7274660477799325,
      "learning_rate": 9.767649687220734e-06,
      "loss": 0.0596,
      "step": 1093
    },
    {
      "epoch": 0.29337624027889514,
      "grad_norm": 0.7357046563041781,
      "learning_rate": 9.776586237712243e-06,
      "loss": 0.0708,
      "step": 1094
    },
    {
      "epoch": 0.2936444086886565,
      "grad_norm": 0.9121530513103581,
      "learning_rate": 9.785522788203754e-06,
      "loss": 0.0643,
      "step": 1095
    },
    {
      "epoch": 0.2939125770984178,
      "grad_norm": 1.0739227451064046,
      "learning_rate": 9.794459338695265e-06,
      "loss": 0.0594,
      "step": 1096
    },
    {
      "epoch": 0.29418074550817913,
      "grad_norm": 0.9766897376558196,
      "learning_rate": 9.803395889186775e-06,
      "loss": 0.0699,
      "step": 1097
    },
    {
      "epoch": 0.2944489139179405,
      "grad_norm": 0.49341165796093067,
      "learning_rate": 9.812332439678284e-06,
      "loss": 0.0405,
      "step": 1098
    },
    {
      "epoch": 0.2947170823277018,
      "grad_norm": 0.7330904635525741,
      "learning_rate": 9.821268990169795e-06,
      "loss": 0.0592,
      "step": 1099
    },
    {
      "epoch": 0.2949852507374631,
      "grad_norm": 0.6662615996827013,
      "learning_rate": 9.830205540661305e-06,
      "loss": 0.0512,
      "step": 1100
    },
    {
      "epoch": 0.29525341914722447,
      "grad_norm": 0.4622151507107239,
      "learning_rate": 9.839142091152816e-06,
      "loss": 0.0577,
      "step": 1101
    },
    {
      "epoch": 0.29552158755698577,
      "grad_norm": 0.6377067812080595,
      "learning_rate": 9.848078641644327e-06,
      "loss": 0.058,
      "step": 1102
    },
    {
      "epoch": 0.2957897559667471,
      "grad_norm": 0.8201111139968565,
      "learning_rate": 9.857015192135836e-06,
      "loss": 0.0622,
      "step": 1103
    },
    {
      "epoch": 0.29605792437650846,
      "grad_norm": 0.6608748012257967,
      "learning_rate": 9.865951742627347e-06,
      "loss": 0.0583,
      "step": 1104
    },
    {
      "epoch": 0.29632609278626976,
      "grad_norm": 0.6701380246533643,
      "learning_rate": 9.874888293118856e-06,
      "loss": 0.0649,
      "step": 1105
    },
    {
      "epoch": 0.2965942611960311,
      "grad_norm": 0.6727116846777095,
      "learning_rate": 9.883824843610367e-06,
      "loss": 0.0574,
      "step": 1106
    },
    {
      "epoch": 0.29686242960579245,
      "grad_norm": 0.5562309106392723,
      "learning_rate": 9.892761394101877e-06,
      "loss": 0.0492,
      "step": 1107
    },
    {
      "epoch": 0.29713059801555375,
      "grad_norm": 0.6445083808230152,
      "learning_rate": 9.901697944593388e-06,
      "loss": 0.0449,
      "step": 1108
    },
    {
      "epoch": 0.2973987664253151,
      "grad_norm": 0.5978559151947976,
      "learning_rate": 9.910634495084897e-06,
      "loss": 0.0557,
      "step": 1109
    },
    {
      "epoch": 0.29766693483507645,
      "grad_norm": 0.6484753688220354,
      "learning_rate": 9.919571045576408e-06,
      "loss": 0.0569,
      "step": 1110
    },
    {
      "epoch": 0.29793510324483774,
      "grad_norm": 0.8588180038306833,
      "learning_rate": 9.92850759606792e-06,
      "loss": 0.0851,
      "step": 1111
    },
    {
      "epoch": 0.2982032716545991,
      "grad_norm": 0.9150265504351235,
      "learning_rate": 9.937444146559429e-06,
      "loss": 0.0727,
      "step": 1112
    },
    {
      "epoch": 0.29847144006436044,
      "grad_norm": 0.5880704450963565,
      "learning_rate": 9.94638069705094e-06,
      "loss": 0.0558,
      "step": 1113
    },
    {
      "epoch": 0.29873960847412173,
      "grad_norm": 0.7413578943581071,
      "learning_rate": 9.955317247542449e-06,
      "loss": 0.0558,
      "step": 1114
    },
    {
      "epoch": 0.2990077768838831,
      "grad_norm": 0.7418022916728801,
      "learning_rate": 9.96425379803396e-06,
      "loss": 0.044,
      "step": 1115
    },
    {
      "epoch": 0.29927594529364443,
      "grad_norm": 0.6806110597988253,
      "learning_rate": 9.97319034852547e-06,
      "loss": 0.0682,
      "step": 1116
    },
    {
      "epoch": 0.2995441137034057,
      "grad_norm": 0.5294203016313889,
      "learning_rate": 9.98212689901698e-06,
      "loss": 0.0497,
      "step": 1117
    },
    {
      "epoch": 0.2998122821131671,
      "grad_norm": 0.6555745503096044,
      "learning_rate": 9.99106344950849e-06,
      "loss": 0.0592,
      "step": 1118
    },
    {
      "epoch": 0.3000804505229284,
      "grad_norm": 0.6349988874894202,
      "learning_rate": 1e-05,
      "loss": 0.0453,
      "step": 1119
    },
    {
      "epoch": 0.3003486189326897,
      "grad_norm": 0.6338997340109369,
      "learning_rate": 9.999999756581638e-06,
      "loss": 0.0463,
      "step": 1120
    },
    {
      "epoch": 0.30061678734245106,
      "grad_norm": 0.8333822777464575,
      "learning_rate": 9.999999026326575e-06,
      "loss": 0.0682,
      "step": 1121
    },
    {
      "epoch": 0.3008849557522124,
      "grad_norm": 0.5896573059518729,
      "learning_rate": 9.999997809234879e-06,
      "loss": 0.0485,
      "step": 1122
    },
    {
      "epoch": 0.3011531241619737,
      "grad_norm": 0.6374461687270109,
      "learning_rate": 9.999996105306673e-06,
      "loss": 0.0538,
      "step": 1123
    },
    {
      "epoch": 0.30142129257173506,
      "grad_norm": 0.7762849683222337,
      "learning_rate": 9.99999391454212e-06,
      "loss": 0.0625,
      "step": 1124
    },
    {
      "epoch": 0.3016894609814964,
      "grad_norm": 0.7384412788392885,
      "learning_rate": 9.999991236941436e-06,
      "loss": 0.0582,
      "step": 1125
    },
    {
      "epoch": 0.3019576293912577,
      "grad_norm": 0.6880480519340163,
      "learning_rate": 9.999988072504878e-06,
      "loss": 0.0598,
      "step": 1126
    },
    {
      "epoch": 0.30222579780101905,
      "grad_norm": 0.5579098832408893,
      "learning_rate": 9.999984421232756e-06,
      "loss": 0.0616,
      "step": 1127
    },
    {
      "epoch": 0.3024939662107804,
      "grad_norm": 0.4298755433135871,
      "learning_rate": 9.999980283125427e-06,
      "loss": 0.041,
      "step": 1128
    },
    {
      "epoch": 0.3027621346205417,
      "grad_norm": 0.39961239646229213,
      "learning_rate": 9.999975658183292e-06,
      "loss": 0.0361,
      "step": 1129
    },
    {
      "epoch": 0.30303030303030304,
      "grad_norm": 0.7320445080181024,
      "learning_rate": 9.9999705464068e-06,
      "loss": 0.0584,
      "step": 1130
    },
    {
      "epoch": 0.3032984714400644,
      "grad_norm": 0.5745009853485151,
      "learning_rate": 9.999964947796453e-06,
      "loss": 0.0454,
      "step": 1131
    },
    {
      "epoch": 0.3035666398498257,
      "grad_norm": 0.5469506729601884,
      "learning_rate": 9.999958862352793e-06,
      "loss": 0.0593,
      "step": 1132
    },
    {
      "epoch": 0.30383480825958703,
      "grad_norm": 0.7118954016306777,
      "learning_rate": 9.999952290076414e-06,
      "loss": 0.0495,
      "step": 1133
    },
    {
      "epoch": 0.3041029766693483,
      "grad_norm": 0.4022427423278802,
      "learning_rate": 9.999945230967953e-06,
      "loss": 0.0397,
      "step": 1134
    },
    {
      "epoch": 0.3043711450791097,
      "grad_norm": 0.7483138244062656,
      "learning_rate": 9.999937685028101e-06,
      "loss": 0.0407,
      "step": 1135
    },
    {
      "epoch": 0.304639313488871,
      "grad_norm": 0.7307167930130598,
      "learning_rate": 9.999929652257592e-06,
      "loss": 0.0479,
      "step": 1136
    },
    {
      "epoch": 0.3049074818986323,
      "grad_norm": 0.653058552018561,
      "learning_rate": 9.999921132657206e-06,
      "loss": 0.0494,
      "step": 1137
    },
    {
      "epoch": 0.30517565030839366,
      "grad_norm": 0.4217059301508981,
      "learning_rate": 9.999912126227774e-06,
      "loss": 0.0403,
      "step": 1138
    },
    {
      "epoch": 0.305443818718155,
      "grad_norm": 0.5579569473617421,
      "learning_rate": 9.999902632970174e-06,
      "loss": 0.0706,
      "step": 1139
    },
    {
      "epoch": 0.3057119871279163,
      "grad_norm": 0.6524567103369456,
      "learning_rate": 9.999892652885325e-06,
      "loss": 0.0544,
      "step": 1140
    },
    {
      "epoch": 0.30598015553767766,
      "grad_norm": 0.6066458913716382,
      "learning_rate": 9.999882185974208e-06,
      "loss": 0.0473,
      "step": 1141
    },
    {
      "epoch": 0.306248323947439,
      "grad_norm": 0.6673392372739456,
      "learning_rate": 9.999871232237835e-06,
      "loss": 0.0606,
      "step": 1142
    },
    {
      "epoch": 0.3065164923572003,
      "grad_norm": 0.8725173658516677,
      "learning_rate": 9.999859791677274e-06,
      "loss": 0.0549,
      "step": 1143
    },
    {
      "epoch": 0.30678466076696165,
      "grad_norm": 0.552648634582255,
      "learning_rate": 9.99984786429364e-06,
      "loss": 0.0488,
      "step": 1144
    },
    {
      "epoch": 0.307052829176723,
      "grad_norm": 0.48734345870093704,
      "learning_rate": 9.999835450088093e-06,
      "loss": 0.0439,
      "step": 1145
    },
    {
      "epoch": 0.3073209975864843,
      "grad_norm": 0.4156890641948285,
      "learning_rate": 9.999822549061843e-06,
      "loss": 0.041,
      "step": 1146
    },
    {
      "epoch": 0.30758916599624564,
      "grad_norm": 0.610066849580524,
      "learning_rate": 9.999809161216145e-06,
      "loss": 0.0631,
      "step": 1147
    },
    {
      "epoch": 0.307857334406007,
      "grad_norm": 0.4806768380383877,
      "learning_rate": 9.999795286552305e-06,
      "loss": 0.0455,
      "step": 1148
    },
    {
      "epoch": 0.3081255028157683,
      "grad_norm": 1.3768498811039278,
      "learning_rate": 9.99978092507167e-06,
      "loss": 0.055,
      "step": 1149
    },
    {
      "epoch": 0.30839367122552963,
      "grad_norm": 0.5343046337991524,
      "learning_rate": 9.999766076775641e-06,
      "loss": 0.0445,
      "step": 1150
    },
    {
      "epoch": 0.308661839635291,
      "grad_norm": 0.53634179559045,
      "learning_rate": 9.999750741665664e-06,
      "loss": 0.0486,
      "step": 1151
    },
    {
      "epoch": 0.3089300080450523,
      "grad_norm": 0.6620571078765336,
      "learning_rate": 9.999734919743231e-06,
      "loss": 0.057,
      "step": 1152
    },
    {
      "epoch": 0.3091981764548136,
      "grad_norm": 0.5574830943974115,
      "learning_rate": 9.999718611009884e-06,
      "loss": 0.0501,
      "step": 1153
    },
    {
      "epoch": 0.30946634486457497,
      "grad_norm": 0.3974204726716722,
      "learning_rate": 9.999701815467209e-06,
      "loss": 0.0441,
      "step": 1154
    },
    {
      "epoch": 0.30973451327433627,
      "grad_norm": 0.6109708637738438,
      "learning_rate": 9.999684533116843e-06,
      "loss": 0.053,
      "step": 1155
    },
    {
      "epoch": 0.3100026816840976,
      "grad_norm": 0.5923972158843595,
      "learning_rate": 9.999666763960468e-06,
      "loss": 0.0519,
      "step": 1156
    },
    {
      "epoch": 0.31027085009385896,
      "grad_norm": 0.6207914836159666,
      "learning_rate": 9.99964850799981e-06,
      "loss": 0.0594,
      "step": 1157
    },
    {
      "epoch": 0.31053901850362026,
      "grad_norm": 0.6971266018166004,
      "learning_rate": 9.999629765236655e-06,
      "loss": 0.0606,
      "step": 1158
    },
    {
      "epoch": 0.3108071869133816,
      "grad_norm": 0.7627432317040791,
      "learning_rate": 9.999610535672822e-06,
      "loss": 0.0672,
      "step": 1159
    },
    {
      "epoch": 0.31107535532314295,
      "grad_norm": 1.0462355989146774,
      "learning_rate": 9.999590819310185e-06,
      "loss": 0.055,
      "step": 1160
    },
    {
      "epoch": 0.31134352373290425,
      "grad_norm": 0.505303486568486,
      "learning_rate": 9.999570616150664e-06,
      "loss": 0.0473,
      "step": 1161
    },
    {
      "epoch": 0.3116116921426656,
      "grad_norm": 0.6403514606799499,
      "learning_rate": 9.999549926196225e-06,
      "loss": 0.0495,
      "step": 1162
    },
    {
      "epoch": 0.31187986055242695,
      "grad_norm": 0.4974097145780926,
      "learning_rate": 9.999528749448883e-06,
      "loss": 0.0448,
      "step": 1163
    },
    {
      "epoch": 0.31214802896218824,
      "grad_norm": 1.1813164642164455,
      "learning_rate": 9.999507085910702e-06,
      "loss": 0.1163,
      "step": 1164
    },
    {
      "epoch": 0.3124161973719496,
      "grad_norm": 1.602332724454136,
      "learning_rate": 9.999484935583787e-06,
      "loss": 0.0749,
      "step": 1165
    },
    {
      "epoch": 0.31268436578171094,
      "grad_norm": 0.4867184233528548,
      "learning_rate": 9.9994622984703e-06,
      "loss": 0.0433,
      "step": 1166
    },
    {
      "epoch": 0.31295253419147223,
      "grad_norm": 0.8302492851395153,
      "learning_rate": 9.999439174572441e-06,
      "loss": 0.0656,
      "step": 1167
    },
    {
      "epoch": 0.3132207026012336,
      "grad_norm": 0.6085673534634279,
      "learning_rate": 9.999415563892461e-06,
      "loss": 0.0473,
      "step": 1168
    },
    {
      "epoch": 0.31348887101099493,
      "grad_norm": 0.48774901026953765,
      "learning_rate": 9.999391466432663e-06,
      "loss": 0.0492,
      "step": 1169
    },
    {
      "epoch": 0.3137570394207562,
      "grad_norm": 0.49479692933604774,
      "learning_rate": 9.99936688219539e-06,
      "loss": 0.0427,
      "step": 1170
    },
    {
      "epoch": 0.31402520783051757,
      "grad_norm": 1.0947714375321784,
      "learning_rate": 9.999341811183036e-06,
      "loss": 0.0689,
      "step": 1171
    },
    {
      "epoch": 0.3142933762402789,
      "grad_norm": 0.3917382750014761,
      "learning_rate": 9.999316253398046e-06,
      "loss": 0.0423,
      "step": 1172
    },
    {
      "epoch": 0.3145615446500402,
      "grad_norm": 0.5745478880230427,
      "learning_rate": 9.999290208842902e-06,
      "loss": 0.0692,
      "step": 1173
    },
    {
      "epoch": 0.31482971305980156,
      "grad_norm": 0.6362465976943918,
      "learning_rate": 9.999263677520146e-06,
      "loss": 0.0477,
      "step": 1174
    },
    {
      "epoch": 0.3150978814695629,
      "grad_norm": 0.6101117032982308,
      "learning_rate": 9.999236659432357e-06,
      "loss": 0.056,
      "step": 1175
    },
    {
      "epoch": 0.3153660498793242,
      "grad_norm": 0.6489249654946811,
      "learning_rate": 9.999209154582166e-06,
      "loss": 0.0569,
      "step": 1176
    },
    {
      "epoch": 0.31563421828908556,
      "grad_norm": 0.8783984427421814,
      "learning_rate": 9.999181162972253e-06,
      "loss": 0.0837,
      "step": 1177
    },
    {
      "epoch": 0.31590238669884685,
      "grad_norm": 0.5240421221217416,
      "learning_rate": 9.999152684605345e-06,
      "loss": 0.0508,
      "step": 1178
    },
    {
      "epoch": 0.3161705551086082,
      "grad_norm": 0.658332854238612,
      "learning_rate": 9.999123719484209e-06,
      "loss": 0.0522,
      "step": 1179
    },
    {
      "epoch": 0.31643872351836955,
      "grad_norm": 0.8568673677041208,
      "learning_rate": 9.999094267611671e-06,
      "loss": 0.0714,
      "step": 1180
    },
    {
      "epoch": 0.31670689192813084,
      "grad_norm": 0.7389052637372896,
      "learning_rate": 9.999064328990596e-06,
      "loss": 0.0541,
      "step": 1181
    },
    {
      "epoch": 0.3169750603378922,
      "grad_norm": 0.7070077968214553,
      "learning_rate": 9.9990339036239e-06,
      "loss": 0.0609,
      "step": 1182
    },
    {
      "epoch": 0.31724322874765354,
      "grad_norm": 0.670723418855042,
      "learning_rate": 9.999002991514544e-06,
      "loss": 0.0458,
      "step": 1183
    },
    {
      "epoch": 0.31751139715741483,
      "grad_norm": 0.8827426677524723,
      "learning_rate": 9.99897159266554e-06,
      "loss": 0.0648,
      "step": 1184
    },
    {
      "epoch": 0.3177795655671762,
      "grad_norm": 0.48443309147074826,
      "learning_rate": 9.998939707079943e-06,
      "loss": 0.0371,
      "step": 1185
    },
    {
      "epoch": 0.31804773397693753,
      "grad_norm": 0.47214541872025856,
      "learning_rate": 9.998907334760859e-06,
      "loss": 0.0526,
      "step": 1186
    },
    {
      "epoch": 0.3183159023866988,
      "grad_norm": 0.6325525109954492,
      "learning_rate": 9.998874475711439e-06,
      "loss": 0.0425,
      "step": 1187
    },
    {
      "epoch": 0.3185840707964602,
      "grad_norm": 0.5281246235985751,
      "learning_rate": 9.998841129934885e-06,
      "loss": 0.054,
      "step": 1188
    },
    {
      "epoch": 0.3188522392062215,
      "grad_norm": 0.4669553258907945,
      "learning_rate": 9.998807297434441e-06,
      "loss": 0.0422,
      "step": 1189
    },
    {
      "epoch": 0.3191204076159828,
      "grad_norm": 0.44491007764543106,
      "learning_rate": 9.9987729782134e-06,
      "loss": 0.0478,
      "step": 1190
    },
    {
      "epoch": 0.31938857602574416,
      "grad_norm": 0.574355405116833,
      "learning_rate": 9.99873817227511e-06,
      "loss": 0.0488,
      "step": 1191
    },
    {
      "epoch": 0.3196567444355055,
      "grad_norm": 0.5312550040991334,
      "learning_rate": 9.998702879622952e-06,
      "loss": 0.0407,
      "step": 1192
    },
    {
      "epoch": 0.3199249128452668,
      "grad_norm": 0.7391872260788728,
      "learning_rate": 9.998667100260367e-06,
      "loss": 0.0501,
      "step": 1193
    },
    {
      "epoch": 0.32019308125502816,
      "grad_norm": 0.7315235721401155,
      "learning_rate": 9.998630834190837e-06,
      "loss": 0.0721,
      "step": 1194
    },
    {
      "epoch": 0.3204612496647895,
      "grad_norm": 0.4374554188285498,
      "learning_rate": 9.998594081417894e-06,
      "loss": 0.0511,
      "step": 1195
    },
    {
      "epoch": 0.3207294180745508,
      "grad_norm": 0.7754960133985275,
      "learning_rate": 9.998556841945118e-06,
      "loss": 0.0611,
      "step": 1196
    },
    {
      "epoch": 0.32099758648431215,
      "grad_norm": 1.0268822408348215,
      "learning_rate": 9.99851911577613e-06,
      "loss": 0.0717,
      "step": 1197
    },
    {
      "epoch": 0.3212657548940735,
      "grad_norm": 0.5614775493313844,
      "learning_rate": 9.998480902914609e-06,
      "loss": 0.0431,
      "step": 1198
    },
    {
      "epoch": 0.3215339233038348,
      "grad_norm": 0.6282347904587501,
      "learning_rate": 9.998442203364271e-06,
      "loss": 0.0479,
      "step": 1199
    },
    {
      "epoch": 0.32180209171359614,
      "grad_norm": 0.7929797452066377,
      "learning_rate": 9.998403017128889e-06,
      "loss": 0.0557,
      "step": 1200
    },
    {
      "epoch": 0.3220702601233575,
      "grad_norm": 0.47353556889844406,
      "learning_rate": 9.998363344212275e-06,
      "loss": 0.0464,
      "step": 1201
    },
    {
      "epoch": 0.3223384285331188,
      "grad_norm": 0.5415896452679779,
      "learning_rate": 9.99832318461829e-06,
      "loss": 0.0424,
      "step": 1202
    },
    {
      "epoch": 0.32260659694288013,
      "grad_norm": 0.5802447111370217,
      "learning_rate": 9.998282538350849e-06,
      "loss": 0.0565,
      "step": 1203
    },
    {
      "epoch": 0.3228747653526415,
      "grad_norm": 0.5861409225922304,
      "learning_rate": 9.998241405413908e-06,
      "loss": 0.0454,
      "step": 1204
    },
    {
      "epoch": 0.3231429337624028,
      "grad_norm": 0.42444180614709814,
      "learning_rate": 9.998199785811472e-06,
      "loss": 0.0434,
      "step": 1205
    },
    {
      "epoch": 0.3234111021721641,
      "grad_norm": 0.7093359796241029,
      "learning_rate": 9.99815767954759e-06,
      "loss": 0.0498,
      "step": 1206
    },
    {
      "epoch": 0.32367927058192547,
      "grad_norm": 0.5168428815853503,
      "learning_rate": 9.998115086626367e-06,
      "loss": 0.0397,
      "step": 1207
    },
    {
      "epoch": 0.32394743899168676,
      "grad_norm": 0.48611064237149737,
      "learning_rate": 9.998072007051946e-06,
      "loss": 0.0442,
      "step": 1208
    },
    {
      "epoch": 0.3242156074014481,
      "grad_norm": 0.6441423210736196,
      "learning_rate": 9.998028440828524e-06,
      "loss": 0.0538,
      "step": 1209
    },
    {
      "epoch": 0.32448377581120946,
      "grad_norm": 0.5386982162337676,
      "learning_rate": 9.997984387960342e-06,
      "loss": 0.0548,
      "step": 1210
    },
    {
      "epoch": 0.32475194422097076,
      "grad_norm": 0.5447310769884404,
      "learning_rate": 9.99793984845169e-06,
      "loss": 0.0637,
      "step": 1211
    },
    {
      "epoch": 0.3250201126307321,
      "grad_norm": 0.5920110680526529,
      "learning_rate": 9.997894822306903e-06,
      "loss": 0.0422,
      "step": 1212
    },
    {
      "epoch": 0.32528828104049345,
      "grad_norm": 0.6131999907550065,
      "learning_rate": 9.997849309530369e-06,
      "loss": 0.0466,
      "step": 1213
    },
    {
      "epoch": 0.32555644945025475,
      "grad_norm": 0.5980967950419176,
      "learning_rate": 9.997803310126515e-06,
      "loss": 0.069,
      "step": 1214
    },
    {
      "epoch": 0.3258246178600161,
      "grad_norm": 0.6150266055109079,
      "learning_rate": 9.997756824099822e-06,
      "loss": 0.0469,
      "step": 1215
    },
    {
      "epoch": 0.32609278626977745,
      "grad_norm": 0.7447691919561921,
      "learning_rate": 9.997709851454815e-06,
      "loss": 0.0701,
      "step": 1216
    },
    {
      "epoch": 0.32636095467953874,
      "grad_norm": 1.0724543438698761,
      "learning_rate": 9.99766239219607e-06,
      "loss": 0.0652,
      "step": 1217
    },
    {
      "epoch": 0.3266291230893001,
      "grad_norm": 0.5885863722442383,
      "learning_rate": 9.997614446328205e-06,
      "loss": 0.0468,
      "step": 1218
    },
    {
      "epoch": 0.32689729149906144,
      "grad_norm": 0.5698531398006067,
      "learning_rate": 9.997566013855891e-06,
      "loss": 0.046,
      "step": 1219
    },
    {
      "epoch": 0.32716545990882273,
      "grad_norm": 0.6531407221404981,
      "learning_rate": 9.997517094783843e-06,
      "loss": 0.0568,
      "step": 1220
    },
    {
      "epoch": 0.3274336283185841,
      "grad_norm": 0.980803681536748,
      "learning_rate": 9.99746768911682e-06,
      "loss": 0.0551,
      "step": 1221
    },
    {
      "epoch": 0.3277017967283454,
      "grad_norm": 0.5509925108741643,
      "learning_rate": 9.997417796859638e-06,
      "loss": 0.0663,
      "step": 1222
    },
    {
      "epoch": 0.3279699651381067,
      "grad_norm": 0.4396352512594954,
      "learning_rate": 9.997367418017154e-06,
      "loss": 0.0437,
      "step": 1223
    },
    {
      "epoch": 0.32823813354786807,
      "grad_norm": 0.597977681561772,
      "learning_rate": 9.997316552594271e-06,
      "loss": 0.0484,
      "step": 1224
    },
    {
      "epoch": 0.32850630195762937,
      "grad_norm": 0.6206789935584044,
      "learning_rate": 9.997265200595944e-06,
      "loss": 0.065,
      "step": 1225
    },
    {
      "epoch": 0.3287744703673907,
      "grad_norm": 0.5457175018181399,
      "learning_rate": 9.997213362027171e-06,
      "loss": 0.0449,
      "step": 1226
    },
    {
      "epoch": 0.32904263877715206,
      "grad_norm": 1.4069740679095741,
      "learning_rate": 9.997161036893001e-06,
      "loss": 0.0496,
      "step": 1227
    },
    {
      "epoch": 0.32931080718691336,
      "grad_norm": 0.42957803259939653,
      "learning_rate": 9.997108225198528e-06,
      "loss": 0.0481,
      "step": 1228
    },
    {
      "epoch": 0.3295789755966747,
      "grad_norm": 0.689776942083758,
      "learning_rate": 9.997054926948895e-06,
      "loss": 0.0599,
      "step": 1229
    },
    {
      "epoch": 0.32984714400643605,
      "grad_norm": 0.6644283461062305,
      "learning_rate": 9.99700114214929e-06,
      "loss": 0.0847,
      "step": 1230
    },
    {
      "epoch": 0.33011531241619735,
      "grad_norm": 0.5652240018221643,
      "learning_rate": 9.99694687080495e-06,
      "loss": 0.0621,
      "step": 1231
    },
    {
      "epoch": 0.3303834808259587,
      "grad_norm": 0.6717250716810702,
      "learning_rate": 9.996892112921161e-06,
      "loss": 0.0627,
      "step": 1232
    },
    {
      "epoch": 0.33065164923572005,
      "grad_norm": 0.686197907154484,
      "learning_rate": 9.996836868503253e-06,
      "loss": 0.054,
      "step": 1233
    },
    {
      "epoch": 0.33091981764548134,
      "grad_norm": 0.4377274461909573,
      "learning_rate": 9.996781137556604e-06,
      "loss": 0.0412,
      "step": 1234
    },
    {
      "epoch": 0.3311879860552427,
      "grad_norm": 0.44317086271523176,
      "learning_rate": 9.996724920086643e-06,
      "loss": 0.0458,
      "step": 1235
    },
    {
      "epoch": 0.33145615446500404,
      "grad_norm": 0.6155713385364404,
      "learning_rate": 9.996668216098841e-06,
      "loss": 0.0627,
      "step": 1236
    },
    {
      "epoch": 0.33172432287476533,
      "grad_norm": 0.7323427849479378,
      "learning_rate": 9.996611025598723e-06,
      "loss": 0.0609,
      "step": 1237
    },
    {
      "epoch": 0.3319924912845267,
      "grad_norm": 0.8651074583277407,
      "learning_rate": 9.996553348591856e-06,
      "loss": 0.055,
      "step": 1238
    },
    {
      "epoch": 0.33226065969428803,
      "grad_norm": 0.7603584456656953,
      "learning_rate": 9.996495185083853e-06,
      "loss": 0.0594,
      "step": 1239
    },
    {
      "epoch": 0.3325288281040493,
      "grad_norm": 0.45835268560219034,
      "learning_rate": 9.996436535080379e-06,
      "loss": 0.0428,
      "step": 1240
    },
    {
      "epoch": 0.3327969965138107,
      "grad_norm": 0.5440838001305944,
      "learning_rate": 9.996377398587146e-06,
      "loss": 0.0442,
      "step": 1241
    },
    {
      "epoch": 0.333065164923572,
      "grad_norm": 0.7047541313901163,
      "learning_rate": 9.996317775609909e-06,
      "loss": 0.0612,
      "step": 1242
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 0.47697292098096294,
      "learning_rate": 9.996257666154477e-06,
      "loss": 0.0382,
      "step": 1243
    },
    {
      "epoch": 0.33360150174309466,
      "grad_norm": 1.2807016967998974,
      "learning_rate": 9.996197070226701e-06,
      "loss": 0.0595,
      "step": 1244
    },
    {
      "epoch": 0.333869670152856,
      "grad_norm": 0.874449200275916,
      "learning_rate": 9.996135987832481e-06,
      "loss": 0.0591,
      "step": 1245
    },
    {
      "epoch": 0.3341378385626173,
      "grad_norm": 0.47341171170424673,
      "learning_rate": 9.996074418977762e-06,
      "loss": 0.0458,
      "step": 1246
    },
    {
      "epoch": 0.33440600697237866,
      "grad_norm": 1.0609136534081791,
      "learning_rate": 9.996012363668544e-06,
      "loss": 0.0462,
      "step": 1247
    },
    {
      "epoch": 0.33467417538214,
      "grad_norm": 0.602144045036516,
      "learning_rate": 9.995949821910864e-06,
      "loss": 0.0612,
      "step": 1248
    },
    {
      "epoch": 0.3349423437919013,
      "grad_norm": 0.5365518375145811,
      "learning_rate": 9.995886793710816e-06,
      "loss": 0.0476,
      "step": 1249
    },
    {
      "epoch": 0.33521051220166265,
      "grad_norm": 0.5062873122293781,
      "learning_rate": 9.995823279074532e-06,
      "loss": 0.0479,
      "step": 1250
    },
    {
      "epoch": 0.335478680611424,
      "grad_norm": 0.6166015609290926,
      "learning_rate": 9.995759278008202e-06,
      "loss": 0.0557,
      "step": 1251
    },
    {
      "epoch": 0.3357468490211853,
      "grad_norm": 0.41152630913083493,
      "learning_rate": 9.995694790518052e-06,
      "loss": 0.0412,
      "step": 1252
    },
    {
      "epoch": 0.33601501743094664,
      "grad_norm": 0.6861178050546083,
      "learning_rate": 9.995629816610365e-06,
      "loss": 0.0377,
      "step": 1253
    },
    {
      "epoch": 0.336283185840708,
      "grad_norm": 0.6458812180499209,
      "learning_rate": 9.995564356291466e-06,
      "loss": 0.0504,
      "step": 1254
    },
    {
      "epoch": 0.3365513542504693,
      "grad_norm": 0.6029266492400795,
      "learning_rate": 9.995498409567727e-06,
      "loss": 0.0557,
      "step": 1255
    },
    {
      "epoch": 0.33681952266023063,
      "grad_norm": 0.5395568920510841,
      "learning_rate": 9.995431976445572e-06,
      "loss": 0.0476,
      "step": 1256
    },
    {
      "epoch": 0.337087691069992,
      "grad_norm": 0.6062976502019732,
      "learning_rate": 9.99536505693147e-06,
      "loss": 0.0601,
      "step": 1257
    },
    {
      "epoch": 0.3373558594797533,
      "grad_norm": 0.5870956103032208,
      "learning_rate": 9.995297651031931e-06,
      "loss": 0.0603,
      "step": 1258
    },
    {
      "epoch": 0.3376240278895146,
      "grad_norm": 0.5980065779499646,
      "learning_rate": 9.995229758753523e-06,
      "loss": 0.0637,
      "step": 1259
    },
    {
      "epoch": 0.33789219629927597,
      "grad_norm": 0.5176078545077247,
      "learning_rate": 9.995161380102857e-06,
      "loss": 0.0435,
      "step": 1260
    },
    {
      "epoch": 0.33816036470903726,
      "grad_norm": 0.5980781610960592,
      "learning_rate": 9.99509251508659e-06,
      "loss": 0.0465,
      "step": 1261
    },
    {
      "epoch": 0.3384285331187986,
      "grad_norm": 0.653243173535905,
      "learning_rate": 9.995023163711424e-06,
      "loss": 0.0527,
      "step": 1262
    },
    {
      "epoch": 0.33869670152855996,
      "grad_norm": 0.5851340216883456,
      "learning_rate": 9.994953325984116e-06,
      "loss": 0.0474,
      "step": 1263
    },
    {
      "epoch": 0.33896486993832126,
      "grad_norm": 0.7005734563979943,
      "learning_rate": 9.994883001911464e-06,
      "loss": 0.0691,
      "step": 1264
    },
    {
      "epoch": 0.3392330383480826,
      "grad_norm": 0.6783434105871767,
      "learning_rate": 9.994812191500313e-06,
      "loss": 0.0685,
      "step": 1265
    },
    {
      "epoch": 0.3395012067578439,
      "grad_norm": 0.6003280977215824,
      "learning_rate": 9.994740894757562e-06,
      "loss": 0.0511,
      "step": 1266
    },
    {
      "epoch": 0.33976937516760525,
      "grad_norm": 1.0974303054537655,
      "learning_rate": 9.99466911169015e-06,
      "loss": 0.0442,
      "step": 1267
    },
    {
      "epoch": 0.3400375435773666,
      "grad_norm": 0.6827668179140555,
      "learning_rate": 9.994596842305067e-06,
      "loss": 0.0585,
      "step": 1268
    },
    {
      "epoch": 0.3403057119871279,
      "grad_norm": 0.8259226948894844,
      "learning_rate": 9.994524086609349e-06,
      "loss": 0.0717,
      "step": 1269
    },
    {
      "epoch": 0.34057388039688924,
      "grad_norm": 0.7346163895712556,
      "learning_rate": 9.994450844610085e-06,
      "loss": 0.045,
      "step": 1270
    },
    {
      "epoch": 0.3408420488066506,
      "grad_norm": 0.39543826481643735,
      "learning_rate": 9.994377116314397e-06,
      "loss": 0.0344,
      "step": 1271
    },
    {
      "epoch": 0.3411102172164119,
      "grad_norm": 0.5800373328720898,
      "learning_rate": 9.994302901729472e-06,
      "loss": 0.0556,
      "step": 1272
    },
    {
      "epoch": 0.34137838562617323,
      "grad_norm": 0.5519684158816232,
      "learning_rate": 9.994228200862532e-06,
      "loss": 0.0438,
      "step": 1273
    },
    {
      "epoch": 0.3416465540359346,
      "grad_norm": 0.5751398448320688,
      "learning_rate": 9.994153013720853e-06,
      "loss": 0.0467,
      "step": 1274
    },
    {
      "epoch": 0.3419147224456959,
      "grad_norm": 0.642724842344229,
      "learning_rate": 9.994077340311751e-06,
      "loss": 0.0686,
      "step": 1275
    },
    {
      "epoch": 0.3421828908554572,
      "grad_norm": 0.49356457083655547,
      "learning_rate": 9.9940011806426e-06,
      "loss": 0.0505,
      "step": 1276
    },
    {
      "epoch": 0.34245105926521857,
      "grad_norm": 0.46694832922304197,
      "learning_rate": 9.993924534720814e-06,
      "loss": 0.0618,
      "step": 1277
    },
    {
      "epoch": 0.34271922767497986,
      "grad_norm": 0.36982061817828604,
      "learning_rate": 9.993847402553854e-06,
      "loss": 0.0501,
      "step": 1278
    },
    {
      "epoch": 0.3429873960847412,
      "grad_norm": 0.4995239486650329,
      "learning_rate": 9.993769784149229e-06,
      "loss": 0.0396,
      "step": 1279
    },
    {
      "epoch": 0.34325556449450256,
      "grad_norm": 0.6177456174818781,
      "learning_rate": 9.993691679514499e-06,
      "loss": 0.0542,
      "step": 1280
    },
    {
      "epoch": 0.34352373290426386,
      "grad_norm": 0.5107819308050413,
      "learning_rate": 9.993613088657267e-06,
      "loss": 0.0535,
      "step": 1281
    },
    {
      "epoch": 0.3437919013140252,
      "grad_norm": 0.5957240597566732,
      "learning_rate": 9.993534011585188e-06,
      "loss": 0.0703,
      "step": 1282
    },
    {
      "epoch": 0.34406006972378655,
      "grad_norm": 0.5658605239578426,
      "learning_rate": 9.993454448305961e-06,
      "loss": 0.0525,
      "step": 1283
    },
    {
      "epoch": 0.34432823813354785,
      "grad_norm": 0.5167561348024428,
      "learning_rate": 9.99337439882733e-06,
      "loss": 0.0515,
      "step": 1284
    },
    {
      "epoch": 0.3445964065433092,
      "grad_norm": 0.5499404042489819,
      "learning_rate": 9.993293863157092e-06,
      "loss": 0.058,
      "step": 1285
    },
    {
      "epoch": 0.34486457495307055,
      "grad_norm": 0.5381475485063885,
      "learning_rate": 9.993212841303087e-06,
      "loss": 0.038,
      "step": 1286
    },
    {
      "epoch": 0.34513274336283184,
      "grad_norm": 0.47875268790706615,
      "learning_rate": 9.993131333273203e-06,
      "loss": 0.0477,
      "step": 1287
    },
    {
      "epoch": 0.3454009117725932,
      "grad_norm": 0.9039109418458877,
      "learning_rate": 9.99304933907538e-06,
      "loss": 0.0627,
      "step": 1288
    },
    {
      "epoch": 0.34566908018235454,
      "grad_norm": 0.5251831747499652,
      "learning_rate": 9.992966858717597e-06,
      "loss": 0.0466,
      "step": 1289
    },
    {
      "epoch": 0.34593724859211583,
      "grad_norm": 0.42591907741104934,
      "learning_rate": 9.992883892207888e-06,
      "loss": 0.0379,
      "step": 1290
    },
    {
      "epoch": 0.3462054170018772,
      "grad_norm": 0.4219803045185595,
      "learning_rate": 9.992800439554331e-06,
      "loss": 0.055,
      "step": 1291
    },
    {
      "epoch": 0.34647358541163853,
      "grad_norm": 0.5276466399247531,
      "learning_rate": 9.992716500765049e-06,
      "loss": 0.056,
      "step": 1292
    },
    {
      "epoch": 0.3467417538213998,
      "grad_norm": 0.49990087461416416,
      "learning_rate": 9.992632075848218e-06,
      "loss": 0.0527,
      "step": 1293
    },
    {
      "epoch": 0.34700992223116117,
      "grad_norm": 0.44478894410907177,
      "learning_rate": 9.992547164812056e-06,
      "loss": 0.0521,
      "step": 1294
    },
    {
      "epoch": 0.3472780906409225,
      "grad_norm": 0.7637729685973763,
      "learning_rate": 9.992461767664833e-06,
      "loss": 0.0788,
      "step": 1295
    },
    {
      "epoch": 0.3475462590506838,
      "grad_norm": 0.522570047005033,
      "learning_rate": 9.992375884414862e-06,
      "loss": 0.0543,
      "step": 1296
    },
    {
      "epoch": 0.34781442746044516,
      "grad_norm": 0.4562174693155051,
      "learning_rate": 9.992289515070506e-06,
      "loss": 0.0554,
      "step": 1297
    },
    {
      "epoch": 0.3480825958702065,
      "grad_norm": 0.49980977283571265,
      "learning_rate": 9.992202659640173e-06,
      "loss": 0.0639,
      "step": 1298
    },
    {
      "epoch": 0.3483507642799678,
      "grad_norm": 0.4277720363213342,
      "learning_rate": 9.99211531813232e-06,
      "loss": 0.047,
      "step": 1299
    },
    {
      "epoch": 0.34861893268972916,
      "grad_norm": 0.4520260442436689,
      "learning_rate": 9.992027490555454e-06,
      "loss": 0.0455,
      "step": 1300
    },
    {
      "epoch": 0.3488871010994905,
      "grad_norm": 0.46186180773458396,
      "learning_rate": 9.991939176918127e-06,
      "loss": 0.053,
      "step": 1301
    },
    {
      "epoch": 0.3491552695092518,
      "grad_norm": 0.4178861470283483,
      "learning_rate": 9.991850377228932e-06,
      "loss": 0.0436,
      "step": 1302
    },
    {
      "epoch": 0.34942343791901315,
      "grad_norm": 0.6055802130945195,
      "learning_rate": 9.99176109149652e-06,
      "loss": 0.0669,
      "step": 1303
    },
    {
      "epoch": 0.3496916063287745,
      "grad_norm": 0.397151347765842,
      "learning_rate": 9.991671319729583e-06,
      "loss": 0.0366,
      "step": 1304
    },
    {
      "epoch": 0.3499597747385358,
      "grad_norm": 0.37708390837146133,
      "learning_rate": 9.991581061936864e-06,
      "loss": 0.0433,
      "step": 1305
    },
    {
      "epoch": 0.35022794314829714,
      "grad_norm": 1.0498601351046697,
      "learning_rate": 9.991490318127149e-06,
      "loss": 0.0585,
      "step": 1306
    },
    {
      "epoch": 0.3504961115580585,
      "grad_norm": 0.6765797780022859,
      "learning_rate": 9.991399088309273e-06,
      "loss": 0.0583,
      "step": 1307
    },
    {
      "epoch": 0.3507642799678198,
      "grad_norm": 0.4840836771083987,
      "learning_rate": 9.991307372492118e-06,
      "loss": 0.0379,
      "step": 1308
    },
    {
      "epoch": 0.35103244837758113,
      "grad_norm": 0.6389956233628771,
      "learning_rate": 9.991215170684618e-06,
      "loss": 0.0437,
      "step": 1309
    },
    {
      "epoch": 0.3513006167873424,
      "grad_norm": 0.8518253753262383,
      "learning_rate": 9.991122482895748e-06,
      "loss": 0.0663,
      "step": 1310
    },
    {
      "epoch": 0.3515687851971038,
      "grad_norm": 1.0277398900534278,
      "learning_rate": 9.991029309134533e-06,
      "loss": 0.0481,
      "step": 1311
    },
    {
      "epoch": 0.3518369536068651,
      "grad_norm": 0.4634799855311521,
      "learning_rate": 9.990935649410046e-06,
      "loss": 0.0492,
      "step": 1312
    },
    {
      "epoch": 0.3521051220166264,
      "grad_norm": 0.4825046604865742,
      "learning_rate": 9.990841503731403e-06,
      "loss": 0.0378,
      "step": 1313
    },
    {
      "epoch": 0.35237329042638776,
      "grad_norm": 0.6459396059372403,
      "learning_rate": 9.990746872107775e-06,
      "loss": 0.0617,
      "step": 1314
    },
    {
      "epoch": 0.3526414588361491,
      "grad_norm": 0.3842373939407726,
      "learning_rate": 9.990651754548374e-06,
      "loss": 0.0387,
      "step": 1315
    },
    {
      "epoch": 0.3529096272459104,
      "grad_norm": 0.7060714051663174,
      "learning_rate": 9.990556151062461e-06,
      "loss": 0.0532,
      "step": 1316
    },
    {
      "epoch": 0.35317779565567176,
      "grad_norm": 0.778506076799676,
      "learning_rate": 9.990460061659346e-06,
      "loss": 0.0528,
      "step": 1317
    },
    {
      "epoch": 0.3534459640654331,
      "grad_norm": 0.5069258753947526,
      "learning_rate": 9.990363486348384e-06,
      "loss": 0.0475,
      "step": 1318
    },
    {
      "epoch": 0.3537141324751944,
      "grad_norm": 0.9320111810595793,
      "learning_rate": 9.990266425138979e-06,
      "loss": 0.0567,
      "step": 1319
    },
    {
      "epoch": 0.35398230088495575,
      "grad_norm": 0.7319769166641872,
      "learning_rate": 9.990168878040581e-06,
      "loss": 0.0737,
      "step": 1320
    },
    {
      "epoch": 0.3542504692947171,
      "grad_norm": 0.48965015588875527,
      "learning_rate": 9.990070845062687e-06,
      "loss": 0.0405,
      "step": 1321
    },
    {
      "epoch": 0.3545186377044784,
      "grad_norm": 0.57053143327353,
      "learning_rate": 9.989972326214842e-06,
      "loss": 0.051,
      "step": 1322
    },
    {
      "epoch": 0.35478680611423974,
      "grad_norm": 0.4705955014165205,
      "learning_rate": 9.989873321506643e-06,
      "loss": 0.0441,
      "step": 1323
    },
    {
      "epoch": 0.3550549745240011,
      "grad_norm": 0.4422692632885619,
      "learning_rate": 9.989773830947724e-06,
      "loss": 0.0469,
      "step": 1324
    },
    {
      "epoch": 0.3553231429337624,
      "grad_norm": 0.6531553954413803,
      "learning_rate": 9.989673854547778e-06,
      "loss": 0.0734,
      "step": 1325
    },
    {
      "epoch": 0.35559131134352373,
      "grad_norm": 0.6493501769789799,
      "learning_rate": 9.989573392316531e-06,
      "loss": 0.0594,
      "step": 1326
    },
    {
      "epoch": 0.3558594797532851,
      "grad_norm": 0.6505359519446975,
      "learning_rate": 9.989472444263773e-06,
      "loss": 0.0526,
      "step": 1327
    },
    {
      "epoch": 0.3561276481630464,
      "grad_norm": 0.5554740860009785,
      "learning_rate": 9.989371010399329e-06,
      "loss": 0.0539,
      "step": 1328
    },
    {
      "epoch": 0.3563958165728077,
      "grad_norm": 0.6903193022984716,
      "learning_rate": 9.989269090733078e-06,
      "loss": 0.0741,
      "step": 1329
    },
    {
      "epoch": 0.35666398498256907,
      "grad_norm": 0.48923714967170945,
      "learning_rate": 9.98916668527494e-06,
      "loss": 0.0483,
      "step": 1330
    },
    {
      "epoch": 0.35693215339233036,
      "grad_norm": 0.6731720070292243,
      "learning_rate": 9.98906379403489e-06,
      "loss": 0.0521,
      "step": 1331
    },
    {
      "epoch": 0.3572003218020917,
      "grad_norm": 0.6401578741863662,
      "learning_rate": 9.988960417022942e-06,
      "loss": 0.0665,
      "step": 1332
    },
    {
      "epoch": 0.35746849021185306,
      "grad_norm": 0.559259783618805,
      "learning_rate": 9.988856554249163e-06,
      "loss": 0.0413,
      "step": 1333
    },
    {
      "epoch": 0.35773665862161436,
      "grad_norm": 0.580708091016445,
      "learning_rate": 9.988752205723668e-06,
      "loss": 0.0526,
      "step": 1334
    },
    {
      "epoch": 0.3580048270313757,
      "grad_norm": 0.5785587006161642,
      "learning_rate": 9.988647371456614e-06,
      "loss": 0.0651,
      "step": 1335
    },
    {
      "epoch": 0.35827299544113705,
      "grad_norm": 0.4559278963571047,
      "learning_rate": 9.988542051458211e-06,
      "loss": 0.0488,
      "step": 1336
    },
    {
      "epoch": 0.35854116385089835,
      "grad_norm": 0.5337252755996704,
      "learning_rate": 9.988436245738714e-06,
      "loss": 0.054,
      "step": 1337
    },
    {
      "epoch": 0.3588093322606597,
      "grad_norm": 0.5822690385648438,
      "learning_rate": 9.988329954308423e-06,
      "loss": 0.0543,
      "step": 1338
    },
    {
      "epoch": 0.35907750067042105,
      "grad_norm": 0.5339882584104336,
      "learning_rate": 9.988223177177688e-06,
      "loss": 0.0513,
      "step": 1339
    },
    {
      "epoch": 0.35934566908018234,
      "grad_norm": 0.5741821274216442,
      "learning_rate": 9.988115914356906e-06,
      "loss": 0.0552,
      "step": 1340
    },
    {
      "epoch": 0.3596138374899437,
      "grad_norm": 0.6667213116760979,
      "learning_rate": 9.98800816585652e-06,
      "loss": 0.053,
      "step": 1341
    },
    {
      "epoch": 0.35988200589970504,
      "grad_norm": 0.46428363199712525,
      "learning_rate": 9.987899931687021e-06,
      "loss": 0.0406,
      "step": 1342
    },
    {
      "epoch": 0.36015017430946633,
      "grad_norm": 0.6755381552305845,
      "learning_rate": 9.98779121185895e-06,
      "loss": 0.0517,
      "step": 1343
    },
    {
      "epoch": 0.3604183427192277,
      "grad_norm": 0.6398551206927541,
      "learning_rate": 9.987682006382891e-06,
      "loss": 0.0498,
      "step": 1344
    },
    {
      "epoch": 0.36068651112898903,
      "grad_norm": 0.47982944839862945,
      "learning_rate": 9.987572315269475e-06,
      "loss": 0.0491,
      "step": 1345
    },
    {
      "epoch": 0.3609546795387503,
      "grad_norm": 0.474606418957216,
      "learning_rate": 9.987462138529387e-06,
      "loss": 0.0401,
      "step": 1346
    },
    {
      "epoch": 0.36122284794851167,
      "grad_norm": 0.4887012362353088,
      "learning_rate": 9.987351476173352e-06,
      "loss": 0.0409,
      "step": 1347
    },
    {
      "epoch": 0.361491016358273,
      "grad_norm": 0.5209333140026641,
      "learning_rate": 9.987240328212144e-06,
      "loss": 0.078,
      "step": 1348
    },
    {
      "epoch": 0.3617591847680343,
      "grad_norm": 0.5177846902182508,
      "learning_rate": 9.987128694656585e-06,
      "loss": 0.0493,
      "step": 1349
    },
    {
      "epoch": 0.36202735317779566,
      "grad_norm": 0.34744880810299267,
      "learning_rate": 9.987016575517546e-06,
      "loss": 0.0429,
      "step": 1350
    },
    {
      "epoch": 0.362295521587557,
      "grad_norm": 0.41574452069130186,
      "learning_rate": 9.986903970805944e-06,
      "loss": 0.0377,
      "step": 1351
    },
    {
      "epoch": 0.3625636899973183,
      "grad_norm": 0.4276759332150814,
      "learning_rate": 9.986790880532741e-06,
      "loss": 0.066,
      "step": 1352
    },
    {
      "epoch": 0.36283185840707965,
      "grad_norm": 0.3752553111902337,
      "learning_rate": 9.986677304708952e-06,
      "loss": 0.0386,
      "step": 1353
    },
    {
      "epoch": 0.36310002681684095,
      "grad_norm": 0.4675732142592896,
      "learning_rate": 9.986563243345633e-06,
      "loss": 0.0472,
      "step": 1354
    },
    {
      "epoch": 0.3633681952266023,
      "grad_norm": 0.49895234443550995,
      "learning_rate": 9.986448696453887e-06,
      "loss": 0.0437,
      "step": 1355
    },
    {
      "epoch": 0.36363636363636365,
      "grad_norm": 0.4759400809143723,
      "learning_rate": 9.986333664044874e-06,
      "loss": 0.0442,
      "step": 1356
    },
    {
      "epoch": 0.36390453204612494,
      "grad_norm": 0.7809012490443222,
      "learning_rate": 9.986218146129789e-06,
      "loss": 0.0539,
      "step": 1357
    },
    {
      "epoch": 0.3641727004558863,
      "grad_norm": 0.4419410675720373,
      "learning_rate": 9.986102142719881e-06,
      "loss": 0.0415,
      "step": 1358
    },
    {
      "epoch": 0.36444086886564764,
      "grad_norm": 0.44661705689939796,
      "learning_rate": 9.985985653826444e-06,
      "loss": 0.0488,
      "step": 1359
    },
    {
      "epoch": 0.36470903727540893,
      "grad_norm": 0.4420119341665048,
      "learning_rate": 9.985868679460824e-06,
      "loss": 0.0375,
      "step": 1360
    },
    {
      "epoch": 0.3649772056851703,
      "grad_norm": 0.4735353599702838,
      "learning_rate": 9.985751219634406e-06,
      "loss": 0.0437,
      "step": 1361
    },
    {
      "epoch": 0.36524537409493163,
      "grad_norm": 0.6893757938273076,
      "learning_rate": 9.98563327435863e-06,
      "loss": 0.0712,
      "step": 1362
    },
    {
      "epoch": 0.3655135425046929,
      "grad_norm": 0.4074776746040115,
      "learning_rate": 9.985514843644978e-06,
      "loss": 0.0489,
      "step": 1363
    },
    {
      "epoch": 0.36578171091445427,
      "grad_norm": 0.4898356575508194,
      "learning_rate": 9.985395927504981e-06,
      "loss": 0.041,
      "step": 1364
    },
    {
      "epoch": 0.3660498793242156,
      "grad_norm": 0.5550625430428274,
      "learning_rate": 9.985276525950221e-06,
      "loss": 0.0404,
      "step": 1365
    },
    {
      "epoch": 0.3663180477339769,
      "grad_norm": 0.7342254774305252,
      "learning_rate": 9.98515663899232e-06,
      "loss": 0.0437,
      "step": 1366
    },
    {
      "epoch": 0.36658621614373826,
      "grad_norm": 0.47754662716095736,
      "learning_rate": 9.985036266642952e-06,
      "loss": 0.0416,
      "step": 1367
    },
    {
      "epoch": 0.3668543845534996,
      "grad_norm": 0.5840497009801184,
      "learning_rate": 9.984915408913837e-06,
      "loss": 0.0435,
      "step": 1368
    },
    {
      "epoch": 0.3671225529632609,
      "grad_norm": 0.490322351220109,
      "learning_rate": 9.984794065816744e-06,
      "loss": 0.0551,
      "step": 1369
    },
    {
      "epoch": 0.36739072137302226,
      "grad_norm": 0.43913971012292896,
      "learning_rate": 9.984672237363487e-06,
      "loss": 0.0431,
      "step": 1370
    },
    {
      "epoch": 0.3676588897827836,
      "grad_norm": 0.38038132117967227,
      "learning_rate": 9.98454992356593e-06,
      "loss": 0.0345,
      "step": 1371
    },
    {
      "epoch": 0.3679270581925449,
      "grad_norm": 0.5660507623698914,
      "learning_rate": 9.984427124435978e-06,
      "loss": 0.0426,
      "step": 1372
    },
    {
      "epoch": 0.36819522660230625,
      "grad_norm": 0.578623278729196,
      "learning_rate": 9.984303839985593e-06,
      "loss": 0.0534,
      "step": 1373
    },
    {
      "epoch": 0.3684633950120676,
      "grad_norm": 0.5730871549562503,
      "learning_rate": 9.984180070226774e-06,
      "loss": 0.0535,
      "step": 1374
    },
    {
      "epoch": 0.3687315634218289,
      "grad_norm": 0.9317908754052814,
      "learning_rate": 9.984055815171576e-06,
      "loss": 0.066,
      "step": 1375
    },
    {
      "epoch": 0.36899973183159024,
      "grad_norm": 0.44931134962916097,
      "learning_rate": 9.983931074832094e-06,
      "loss": 0.0431,
      "step": 1376
    },
    {
      "epoch": 0.3692679002413516,
      "grad_norm": 0.42163778082235204,
      "learning_rate": 9.983805849220475e-06,
      "loss": 0.0337,
      "step": 1377
    },
    {
      "epoch": 0.3695360686511129,
      "grad_norm": 0.5783753666495938,
      "learning_rate": 9.983680138348914e-06,
      "loss": 0.0419,
      "step": 1378
    },
    {
      "epoch": 0.36980423706087423,
      "grad_norm": 0.6042800248274317,
      "learning_rate": 9.983553942229649e-06,
      "loss": 0.0442,
      "step": 1379
    },
    {
      "epoch": 0.3700724054706356,
      "grad_norm": 0.5290235600246085,
      "learning_rate": 9.983427260874967e-06,
      "loss": 0.0503,
      "step": 1380
    },
    {
      "epoch": 0.3703405738803969,
      "grad_norm": 0.47350076586267714,
      "learning_rate": 9.983300094297205e-06,
      "loss": 0.0468,
      "step": 1381
    },
    {
      "epoch": 0.3706087422901582,
      "grad_norm": 0.5343609301608477,
      "learning_rate": 9.983172442508743e-06,
      "loss": 0.0471,
      "step": 1382
    },
    {
      "epoch": 0.37087691069991957,
      "grad_norm": 0.4194077126788963,
      "learning_rate": 9.983044305522007e-06,
      "loss": 0.027,
      "step": 1383
    },
    {
      "epoch": 0.37114507910968086,
      "grad_norm": 0.6802081209523023,
      "learning_rate": 9.98291568334948e-06,
      "loss": 0.0506,
      "step": 1384
    },
    {
      "epoch": 0.3714132475194422,
      "grad_norm": 0.5234856291234419,
      "learning_rate": 9.982786576003682e-06,
      "loss": 0.0514,
      "step": 1385
    },
    {
      "epoch": 0.37168141592920356,
      "grad_norm": 0.5926422302263622,
      "learning_rate": 9.982656983497183e-06,
      "loss": 0.0468,
      "step": 1386
    },
    {
      "epoch": 0.37194958433896486,
      "grad_norm": 0.5520763702431307,
      "learning_rate": 9.982526905842601e-06,
      "loss": 0.0417,
      "step": 1387
    },
    {
      "epoch": 0.3722177527487262,
      "grad_norm": 0.3956913429740374,
      "learning_rate": 9.982396343052606e-06,
      "loss": 0.0342,
      "step": 1388
    },
    {
      "epoch": 0.37248592115848755,
      "grad_norm": 0.5835380426673424,
      "learning_rate": 9.982265295139906e-06,
      "loss": 0.0524,
      "step": 1389
    },
    {
      "epoch": 0.37275408956824885,
      "grad_norm": 0.9411097239192372,
      "learning_rate": 9.982133762117262e-06,
      "loss": 0.0762,
      "step": 1390
    },
    {
      "epoch": 0.3730222579780102,
      "grad_norm": 0.7319551021174818,
      "learning_rate": 9.982001743997477e-06,
      "loss": 0.0498,
      "step": 1391
    },
    {
      "epoch": 0.37329042638777155,
      "grad_norm": 0.4576709008205559,
      "learning_rate": 9.981869240793414e-06,
      "loss": 0.0385,
      "step": 1392
    },
    {
      "epoch": 0.37355859479753284,
      "grad_norm": 0.42831650216039885,
      "learning_rate": 9.981736252517967e-06,
      "loss": 0.0297,
      "step": 1393
    },
    {
      "epoch": 0.3738267632072942,
      "grad_norm": 0.4726164380036066,
      "learning_rate": 9.981602779184089e-06,
      "loss": 0.0488,
      "step": 1394
    },
    {
      "epoch": 0.37409493161705554,
      "grad_norm": 0.4643216536315915,
      "learning_rate": 9.981468820804774e-06,
      "loss": 0.039,
      "step": 1395
    },
    {
      "epoch": 0.37436310002681683,
      "grad_norm": 0.6958952257581869,
      "learning_rate": 9.981334377393063e-06,
      "loss": 0.0559,
      "step": 1396
    },
    {
      "epoch": 0.3746312684365782,
      "grad_norm": 0.5852392534460145,
      "learning_rate": 9.98119944896205e-06,
      "loss": 0.0505,
      "step": 1397
    },
    {
      "epoch": 0.37489943684633953,
      "grad_norm": 0.3958672556069444,
      "learning_rate": 9.981064035524875e-06,
      "loss": 0.0366,
      "step": 1398
    },
    {
      "epoch": 0.3751676052561008,
      "grad_norm": 0.49623659509855444,
      "learning_rate": 9.980928137094716e-06,
      "loss": 0.0401,
      "step": 1399
    },
    {
      "epoch": 0.37543577366586217,
      "grad_norm": 0.7581586890739327,
      "learning_rate": 9.980791753684809e-06,
      "loss": 0.0793,
      "step": 1400
    },
    {
      "epoch": 0.37570394207562346,
      "grad_norm": 0.3510886591633425,
      "learning_rate": 9.980654885308433e-06,
      "loss": 0.0355,
      "step": 1401
    },
    {
      "epoch": 0.3759721104853848,
      "grad_norm": 0.7732746990894042,
      "learning_rate": 9.980517531978914e-06,
      "loss": 0.0637,
      "step": 1402
    },
    {
      "epoch": 0.37624027889514616,
      "grad_norm": 0.45666944170885615,
      "learning_rate": 9.980379693709626e-06,
      "loss": 0.0427,
      "step": 1403
    },
    {
      "epoch": 0.37650844730490746,
      "grad_norm": 0.4787066788974094,
      "learning_rate": 9.980241370513989e-06,
      "loss": 0.0568,
      "step": 1404
    },
    {
      "epoch": 0.3767766157146688,
      "grad_norm": 0.4654427281543683,
      "learning_rate": 9.980102562405474e-06,
      "loss": 0.0561,
      "step": 1405
    },
    {
      "epoch": 0.37704478412443015,
      "grad_norm": 0.5599225065243438,
      "learning_rate": 9.979963269397592e-06,
      "loss": 0.0484,
      "step": 1406
    },
    {
      "epoch": 0.37731295253419145,
      "grad_norm": 0.3823018157604615,
      "learning_rate": 9.979823491503909e-06,
      "loss": 0.0343,
      "step": 1407
    },
    {
      "epoch": 0.3775811209439528,
      "grad_norm": 0.6441600031989471,
      "learning_rate": 9.979683228738034e-06,
      "loss": 0.0428,
      "step": 1408
    },
    {
      "epoch": 0.37784928935371415,
      "grad_norm": 0.48365263685157994,
      "learning_rate": 9.979542481113625e-06,
      "loss": 0.042,
      "step": 1409
    },
    {
      "epoch": 0.37811745776347544,
      "grad_norm": 0.4933179841794089,
      "learning_rate": 9.979401248644383e-06,
      "loss": 0.042,
      "step": 1410
    },
    {
      "epoch": 0.3783856261732368,
      "grad_norm": 1.0468988150171847,
      "learning_rate": 9.979259531344062e-06,
      "loss": 0.0524,
      "step": 1411
    },
    {
      "epoch": 0.37865379458299814,
      "grad_norm": 0.46632216206808014,
      "learning_rate": 9.97911732922646e-06,
      "loss": 0.045,
      "step": 1412
    },
    {
      "epoch": 0.37892196299275943,
      "grad_norm": 0.46767908699365535,
      "learning_rate": 9.978974642305424e-06,
      "loss": 0.0727,
      "step": 1413
    },
    {
      "epoch": 0.3791901314025208,
      "grad_norm": 0.38075460376642184,
      "learning_rate": 9.978831470594846e-06,
      "loss": 0.0535,
      "step": 1414
    },
    {
      "epoch": 0.37945829981228213,
      "grad_norm": 0.4659791271109738,
      "learning_rate": 9.978687814108666e-06,
      "loss": 0.0416,
      "step": 1415
    },
    {
      "epoch": 0.3797264682220434,
      "grad_norm": 0.4240134075415706,
      "learning_rate": 9.97854367286087e-06,
      "loss": 0.0438,
      "step": 1416
    },
    {
      "epoch": 0.37999463663180477,
      "grad_norm": 0.4538111104862816,
      "learning_rate": 9.978399046865498e-06,
      "loss": 0.0717,
      "step": 1417
    },
    {
      "epoch": 0.3802628050415661,
      "grad_norm": 0.4100673087654287,
      "learning_rate": 9.978253936136624e-06,
      "loss": 0.0436,
      "step": 1418
    },
    {
      "epoch": 0.3805309734513274,
      "grad_norm": 0.4799623130181287,
      "learning_rate": 9.978108340688383e-06,
      "loss": 0.0458,
      "step": 1419
    },
    {
      "epoch": 0.38079914186108876,
      "grad_norm": 0.4443899332532248,
      "learning_rate": 9.97796226053495e-06,
      "loss": 0.0377,
      "step": 1420
    },
    {
      "epoch": 0.3810673102708501,
      "grad_norm": 0.5300463512289733,
      "learning_rate": 9.977815695690547e-06,
      "loss": 0.0431,
      "step": 1421
    },
    {
      "epoch": 0.3813354786806114,
      "grad_norm": 0.5234572709973816,
      "learning_rate": 9.977668646169447e-06,
      "loss": 0.061,
      "step": 1422
    },
    {
      "epoch": 0.38160364709037276,
      "grad_norm": 0.38089329821921597,
      "learning_rate": 9.977521111985965e-06,
      "loss": 0.041,
      "step": 1423
    },
    {
      "epoch": 0.3818718155001341,
      "grad_norm": 0.5109836358567956,
      "learning_rate": 9.977373093154468e-06,
      "loss": 0.0458,
      "step": 1424
    },
    {
      "epoch": 0.3821399839098954,
      "grad_norm": 0.3724071955563545,
      "learning_rate": 9.977224589689366e-06,
      "loss": 0.0368,
      "step": 1425
    },
    {
      "epoch": 0.38240815231965675,
      "grad_norm": 0.4487793140625272,
      "learning_rate": 9.977075601605123e-06,
      "loss": 0.0355,
      "step": 1426
    },
    {
      "epoch": 0.3826763207294181,
      "grad_norm": 0.5629154801988521,
      "learning_rate": 9.976926128916242e-06,
      "loss": 0.051,
      "step": 1427
    },
    {
      "epoch": 0.3829444891391794,
      "grad_norm": 0.5217214389959313,
      "learning_rate": 9.976776171637275e-06,
      "loss": 0.0726,
      "step": 1428
    },
    {
      "epoch": 0.38321265754894074,
      "grad_norm": 0.5314739590759987,
      "learning_rate": 9.976625729782827e-06,
      "loss": 0.0511,
      "step": 1429
    },
    {
      "epoch": 0.3834808259587021,
      "grad_norm": 0.4960070344457156,
      "learning_rate": 9.976474803367544e-06,
      "loss": 0.0439,
      "step": 1430
    },
    {
      "epoch": 0.3837489943684634,
      "grad_norm": 0.38587853940109235,
      "learning_rate": 9.976323392406122e-06,
      "loss": 0.0479,
      "step": 1431
    },
    {
      "epoch": 0.38401716277822473,
      "grad_norm": 0.5398012707240039,
      "learning_rate": 9.976171496913303e-06,
      "loss": 0.0401,
      "step": 1432
    },
    {
      "epoch": 0.3842853311879861,
      "grad_norm": 0.4346807089344981,
      "learning_rate": 9.976019116903875e-06,
      "loss": 0.0439,
      "step": 1433
    },
    {
      "epoch": 0.3845534995977474,
      "grad_norm": 0.6565075836443373,
      "learning_rate": 9.975866252392678e-06,
      "loss": 0.0615,
      "step": 1434
    },
    {
      "epoch": 0.3848216680075087,
      "grad_norm": 0.46036894970990455,
      "learning_rate": 9.975712903394597e-06,
      "loss": 0.0521,
      "step": 1435
    },
    {
      "epoch": 0.38508983641727007,
      "grad_norm": 0.38701226068556727,
      "learning_rate": 9.975559069924558e-06,
      "loss": 0.0415,
      "step": 1436
    },
    {
      "epoch": 0.38535800482703136,
      "grad_norm": 0.36945310510925883,
      "learning_rate": 9.975404751997543e-06,
      "loss": 0.0356,
      "step": 1437
    },
    {
      "epoch": 0.3856261732367927,
      "grad_norm": 0.5141870173591446,
      "learning_rate": 9.975249949628576e-06,
      "loss": 0.0449,
      "step": 1438
    },
    {
      "epoch": 0.38589434164655406,
      "grad_norm": 0.5391929841638313,
      "learning_rate": 9.975094662832732e-06,
      "loss": 0.0463,
      "step": 1439
    },
    {
      "epoch": 0.38616251005631536,
      "grad_norm": 0.5242149219576956,
      "learning_rate": 9.974938891625128e-06,
      "loss": 0.0547,
      "step": 1440
    },
    {
      "epoch": 0.3864306784660767,
      "grad_norm": 0.37537260138139494,
      "learning_rate": 9.974782636020933e-06,
      "loss": 0.0296,
      "step": 1441
    },
    {
      "epoch": 0.38669884687583805,
      "grad_norm": 0.5324900321642044,
      "learning_rate": 9.974625896035361e-06,
      "loss": 0.0421,
      "step": 1442
    },
    {
      "epoch": 0.38696701528559935,
      "grad_norm": 0.9984801550917732,
      "learning_rate": 9.974468671683673e-06,
      "loss": 0.0613,
      "step": 1443
    },
    {
      "epoch": 0.3872351836953607,
      "grad_norm": 0.4202347866884358,
      "learning_rate": 9.974310962981176e-06,
      "loss": 0.0504,
      "step": 1444
    },
    {
      "epoch": 0.387503352105122,
      "grad_norm": 0.49957061268931924,
      "learning_rate": 9.974152769943227e-06,
      "loss": 0.0517,
      "step": 1445
    },
    {
      "epoch": 0.38777152051488334,
      "grad_norm": 0.479976258927118,
      "learning_rate": 9.97399409258523e-06,
      "loss": 0.0566,
      "step": 1446
    },
    {
      "epoch": 0.3880396889246447,
      "grad_norm": 0.7692274486872865,
      "learning_rate": 9.973834930922634e-06,
      "loss": 0.0729,
      "step": 1447
    },
    {
      "epoch": 0.388307857334406,
      "grad_norm": 0.4691759416819609,
      "learning_rate": 9.973675284970936e-06,
      "loss": 0.0425,
      "step": 1448
    },
    {
      "epoch": 0.38857602574416733,
      "grad_norm": 0.6746321261388285,
      "learning_rate": 9.973515154745679e-06,
      "loss": 0.0627,
      "step": 1449
    },
    {
      "epoch": 0.3888441941539287,
      "grad_norm": 0.7383042564015457,
      "learning_rate": 9.973354540262456e-06,
      "loss": 0.0573,
      "step": 1450
    },
    {
      "epoch": 0.38911236256369,
      "grad_norm": 0.6949022748509741,
      "learning_rate": 9.973193441536905e-06,
      "loss": 0.0623,
      "step": 1451
    },
    {
      "epoch": 0.3893805309734513,
      "grad_norm": 0.6076571369328968,
      "learning_rate": 9.973031858584716e-06,
      "loss": 0.0536,
      "step": 1452
    },
    {
      "epoch": 0.38964869938321267,
      "grad_norm": 0.4937626334395001,
      "learning_rate": 9.972869791421615e-06,
      "loss": 0.0469,
      "step": 1453
    },
    {
      "epoch": 0.38991686779297396,
      "grad_norm": 0.6935468145804896,
      "learning_rate": 9.972707240063385e-06,
      "loss": 0.0492,
      "step": 1454
    },
    {
      "epoch": 0.3901850362027353,
      "grad_norm": 0.5307630703036336,
      "learning_rate": 9.972544204525853e-06,
      "loss": 0.0456,
      "step": 1455
    },
    {
      "epoch": 0.39045320461249666,
      "grad_norm": 0.4598150390437496,
      "learning_rate": 9.972380684824896e-06,
      "loss": 0.0522,
      "step": 1456
    },
    {
      "epoch": 0.39072137302225796,
      "grad_norm": 0.729621194188739,
      "learning_rate": 9.97221668097643e-06,
      "loss": 0.0689,
      "step": 1457
    },
    {
      "epoch": 0.3909895414320193,
      "grad_norm": 0.621999751544139,
      "learning_rate": 9.97205219299643e-06,
      "loss": 0.0578,
      "step": 1458
    },
    {
      "epoch": 0.39125770984178065,
      "grad_norm": 0.48484683547037954,
      "learning_rate": 9.971887220900907e-06,
      "loss": 0.0586,
      "step": 1459
    },
    {
      "epoch": 0.39152587825154195,
      "grad_norm": 0.7533900547061578,
      "learning_rate": 9.971721764705924e-06,
      "loss": 0.0674,
      "step": 1460
    },
    {
      "epoch": 0.3917940466613033,
      "grad_norm": 0.4832192375594444,
      "learning_rate": 9.971555824427594e-06,
      "loss": 0.0509,
      "step": 1461
    },
    {
      "epoch": 0.39206221507106465,
      "grad_norm": 0.5832665691678989,
      "learning_rate": 9.971389400082073e-06,
      "loss": 0.0585,
      "step": 1462
    },
    {
      "epoch": 0.39233038348082594,
      "grad_norm": 0.5559653938928762,
      "learning_rate": 9.971222491685564e-06,
      "loss": 0.0401,
      "step": 1463
    },
    {
      "epoch": 0.3925985518905873,
      "grad_norm": 0.40470074450263593,
      "learning_rate": 9.971055099254318e-06,
      "loss": 0.0357,
      "step": 1464
    },
    {
      "epoch": 0.39286672030034864,
      "grad_norm": 0.6743024056813168,
      "learning_rate": 9.970887222804637e-06,
      "loss": 0.0711,
      "step": 1465
    },
    {
      "epoch": 0.39313488871010993,
      "grad_norm": 1.0236978797170537,
      "learning_rate": 9.970718862352865e-06,
      "loss": 0.0513,
      "step": 1466
    },
    {
      "epoch": 0.3934030571198713,
      "grad_norm": 0.6387977085505682,
      "learning_rate": 9.970550017915393e-06,
      "loss": 0.0575,
      "step": 1467
    },
    {
      "epoch": 0.39367122552963263,
      "grad_norm": 0.6916774413735859,
      "learning_rate": 9.970380689508664e-06,
      "loss": 0.0488,
      "step": 1468
    },
    {
      "epoch": 0.3939393939393939,
      "grad_norm": 0.7584007830815458,
      "learning_rate": 9.97021087714916e-06,
      "loss": 0.0597,
      "step": 1469
    },
    {
      "epoch": 0.39420756234915527,
      "grad_norm": 0.3948690896021368,
      "learning_rate": 9.970040580853422e-06,
      "loss": 0.0398,
      "step": 1470
    },
    {
      "epoch": 0.3944757307589166,
      "grad_norm": 0.3631523775322716,
      "learning_rate": 9.969869800638026e-06,
      "loss": 0.032,
      "step": 1471
    },
    {
      "epoch": 0.3947438991686779,
      "grad_norm": 0.6044347965634627,
      "learning_rate": 9.969698536519603e-06,
      "loss": 0.0457,
      "step": 1472
    },
    {
      "epoch": 0.39501206757843926,
      "grad_norm": 0.4279184496651392,
      "learning_rate": 9.969526788514827e-06,
      "loss": 0.0351,
      "step": 1473
    },
    {
      "epoch": 0.3952802359882006,
      "grad_norm": 0.5568081474423444,
      "learning_rate": 9.969354556640423e-06,
      "loss": 0.049,
      "step": 1474
    },
    {
      "epoch": 0.3955484043979619,
      "grad_norm": 0.5232715356198936,
      "learning_rate": 9.969181840913159e-06,
      "loss": 0.0486,
      "step": 1475
    },
    {
      "epoch": 0.39581657280772325,
      "grad_norm": 0.6600744673638719,
      "learning_rate": 9.969008641349848e-06,
      "loss": 0.0397,
      "step": 1476
    },
    {
      "epoch": 0.3960847412174846,
      "grad_norm": 0.7289736681617289,
      "learning_rate": 9.968834957967363e-06,
      "loss": 0.0759,
      "step": 1477
    },
    {
      "epoch": 0.3963529096272459,
      "grad_norm": 0.48884858872863574,
      "learning_rate": 9.968660790782608e-06,
      "loss": 0.0411,
      "step": 1478
    },
    {
      "epoch": 0.39662107803700725,
      "grad_norm": 0.5627631949884025,
      "learning_rate": 9.968486139812544e-06,
      "loss": 0.0441,
      "step": 1479
    },
    {
      "epoch": 0.3968892464467686,
      "grad_norm": 0.6120268467122656,
      "learning_rate": 9.968311005074175e-06,
      "loss": 0.045,
      "step": 1480
    },
    {
      "epoch": 0.3971574148565299,
      "grad_norm": 0.6758895689277288,
      "learning_rate": 9.968135386584554e-06,
      "loss": 0.042,
      "step": 1481
    },
    {
      "epoch": 0.39742558326629124,
      "grad_norm": 0.5316117909588363,
      "learning_rate": 9.967959284360781e-06,
      "loss": 0.0453,
      "step": 1482
    },
    {
      "epoch": 0.3976937516760526,
      "grad_norm": 0.4211777878796908,
      "learning_rate": 9.967782698420004e-06,
      "loss": 0.0427,
      "step": 1483
    },
    {
      "epoch": 0.3979619200858139,
      "grad_norm": 0.6619456950817497,
      "learning_rate": 9.967605628779412e-06,
      "loss": 0.0422,
      "step": 1484
    },
    {
      "epoch": 0.39823008849557523,
      "grad_norm": 1.1739600897471947,
      "learning_rate": 9.967428075456248e-06,
      "loss": 0.0541,
      "step": 1485
    },
    {
      "epoch": 0.3984982569053366,
      "grad_norm": 0.511679976345095,
      "learning_rate": 9.967250038467803e-06,
      "loss": 0.0563,
      "step": 1486
    },
    {
      "epoch": 0.39876642531509787,
      "grad_norm": 0.6363511043776823,
      "learning_rate": 9.967071517831411e-06,
      "loss": 0.0537,
      "step": 1487
    },
    {
      "epoch": 0.3990345937248592,
      "grad_norm": 0.4250145312251278,
      "learning_rate": 9.96689251356445e-06,
      "loss": 0.0529,
      "step": 1488
    },
    {
      "epoch": 0.3993027621346205,
      "grad_norm": 0.7341479308550573,
      "learning_rate": 9.966713025684351e-06,
      "loss": 0.0545,
      "step": 1489
    },
    {
      "epoch": 0.39957093054438186,
      "grad_norm": 0.7234099827193285,
      "learning_rate": 9.966533054208593e-06,
      "loss": 0.0722,
      "step": 1490
    },
    {
      "epoch": 0.3998390989541432,
      "grad_norm": 0.5179774960511757,
      "learning_rate": 9.966352599154697e-06,
      "loss": 0.0404,
      "step": 1491
    },
    {
      "epoch": 0.4001072673639045,
      "grad_norm": 0.6020615267730577,
      "learning_rate": 9.966171660540233e-06,
      "loss": 0.0476,
      "step": 1492
    },
    {
      "epoch": 0.40037543577366586,
      "grad_norm": 0.4817371378206565,
      "learning_rate": 9.96599023838282e-06,
      "loss": 0.0506,
      "step": 1493
    },
    {
      "epoch": 0.4006436041834272,
      "grad_norm": 0.44723070637766194,
      "learning_rate": 9.965808332700122e-06,
      "loss": 0.045,
      "step": 1494
    },
    {
      "epoch": 0.4009117725931885,
      "grad_norm": 0.6039138079646764,
      "learning_rate": 9.965625943509851e-06,
      "loss": 0.0729,
      "step": 1495
    },
    {
      "epoch": 0.40117994100294985,
      "grad_norm": 0.3655912305770551,
      "learning_rate": 9.965443070829763e-06,
      "loss": 0.0348,
      "step": 1496
    },
    {
      "epoch": 0.4014481094127112,
      "grad_norm": 0.7820269867978273,
      "learning_rate": 9.965259714677668e-06,
      "loss": 0.0518,
      "step": 1497
    },
    {
      "epoch": 0.4017162778224725,
      "grad_norm": 0.3538558107447198,
      "learning_rate": 9.965075875071417e-06,
      "loss": 0.0375,
      "step": 1498
    },
    {
      "epoch": 0.40198444623223384,
      "grad_norm": 0.6178258602842533,
      "learning_rate": 9.964891552028911e-06,
      "loss": 0.0417,
      "step": 1499
    },
    {
      "epoch": 0.4022526146419952,
      "grad_norm": 0.4536696421404748,
      "learning_rate": 9.964706745568096e-06,
      "loss": 0.0479,
      "step": 1500
    },
    {
      "epoch": 0.4025207830517565,
      "grad_norm": 0.5099670672020854,
      "learning_rate": 9.964521455706965e-06,
      "loss": 0.0399,
      "step": 1501
    },
    {
      "epoch": 0.40278895146151783,
      "grad_norm": 0.5779780026066804,
      "learning_rate": 9.964335682463561e-06,
      "loss": 0.0558,
      "step": 1502
    },
    {
      "epoch": 0.4030571198712792,
      "grad_norm": 0.8210955795531958,
      "learning_rate": 9.964149425855971e-06,
      "loss": 0.0525,
      "step": 1503
    },
    {
      "epoch": 0.4033252882810405,
      "grad_norm": 1.0478277529182631,
      "learning_rate": 9.963962685902331e-06,
      "loss": 0.0611,
      "step": 1504
    },
    {
      "epoch": 0.4035934566908018,
      "grad_norm": 0.505294396513007,
      "learning_rate": 9.963775462620825e-06,
      "loss": 0.0454,
      "step": 1505
    },
    {
      "epoch": 0.40386162510056317,
      "grad_norm": 0.9066308955405175,
      "learning_rate": 9.963587756029679e-06,
      "loss": 0.0465,
      "step": 1506
    },
    {
      "epoch": 0.40412979351032446,
      "grad_norm": 0.5019543675031959,
      "learning_rate": 9.963399566147172e-06,
      "loss": 0.0572,
      "step": 1507
    },
    {
      "epoch": 0.4043979619200858,
      "grad_norm": 0.40156110640085935,
      "learning_rate": 9.963210892991626e-06,
      "loss": 0.0363,
      "step": 1508
    },
    {
      "epoch": 0.40466613032984716,
      "grad_norm": 0.5787893379514217,
      "learning_rate": 9.963021736581413e-06,
      "loss": 0.0494,
      "step": 1509
    },
    {
      "epoch": 0.40493429873960846,
      "grad_norm": 0.6864524043023535,
      "learning_rate": 9.962832096934952e-06,
      "loss": 0.0612,
      "step": 1510
    },
    {
      "epoch": 0.4052024671493698,
      "grad_norm": 0.5714415032227125,
      "learning_rate": 9.962641974070702e-06,
      "loss": 0.0628,
      "step": 1511
    },
    {
      "epoch": 0.40547063555913115,
      "grad_norm": 0.6093447704244753,
      "learning_rate": 9.962451368007182e-06,
      "loss": 0.0629,
      "step": 1512
    },
    {
      "epoch": 0.40573880396889245,
      "grad_norm": 0.5614457167971783,
      "learning_rate": 9.962260278762946e-06,
      "loss": 0.0363,
      "step": 1513
    },
    {
      "epoch": 0.4060069723786538,
      "grad_norm": 0.4806050014375334,
      "learning_rate": 9.9620687063566e-06,
      "loss": 0.0395,
      "step": 1514
    },
    {
      "epoch": 0.40627514078841515,
      "grad_norm": 0.3849303782813662,
      "learning_rate": 9.961876650806799e-06,
      "loss": 0.0419,
      "step": 1515
    },
    {
      "epoch": 0.40654330919817644,
      "grad_norm": 0.4952169567558052,
      "learning_rate": 9.961684112132242e-06,
      "loss": 0.0348,
      "step": 1516
    },
    {
      "epoch": 0.4068114776079378,
      "grad_norm": 0.49944364398352076,
      "learning_rate": 9.961491090351676e-06,
      "loss": 0.0545,
      "step": 1517
    },
    {
      "epoch": 0.40707964601769914,
      "grad_norm": 0.4481913069439499,
      "learning_rate": 9.961297585483895e-06,
      "loss": 0.0494,
      "step": 1518
    },
    {
      "epoch": 0.40734781442746043,
      "grad_norm": 0.48121855869425484,
      "learning_rate": 9.961103597547741e-06,
      "loss": 0.0516,
      "step": 1519
    },
    {
      "epoch": 0.4076159828372218,
      "grad_norm": 0.5555965516580667,
      "learning_rate": 9.960909126562102e-06,
      "loss": 0.0613,
      "step": 1520
    },
    {
      "epoch": 0.40788415124698313,
      "grad_norm": 0.7679943289847897,
      "learning_rate": 9.96071417254591e-06,
      "loss": 0.0658,
      "step": 1521
    },
    {
      "epoch": 0.4081523196567444,
      "grad_norm": 0.3603558260148338,
      "learning_rate": 9.96051873551815e-06,
      "loss": 0.039,
      "step": 1522
    },
    {
      "epoch": 0.40842048806650577,
      "grad_norm": 0.4828801955429522,
      "learning_rate": 9.960322815497852e-06,
      "loss": 0.0498,
      "step": 1523
    },
    {
      "epoch": 0.4086886564762671,
      "grad_norm": 0.5000190310469207,
      "learning_rate": 9.960126412504089e-06,
      "loss": 0.0495,
      "step": 1524
    },
    {
      "epoch": 0.4089568248860284,
      "grad_norm": 0.5593440535457198,
      "learning_rate": 9.959929526555987e-06,
      "loss": 0.0536,
      "step": 1525
    },
    {
      "epoch": 0.40922499329578976,
      "grad_norm": 0.371495004109536,
      "learning_rate": 9.959732157672715e-06,
      "loss": 0.0424,
      "step": 1526
    },
    {
      "epoch": 0.4094931617055511,
      "grad_norm": 0.42931403654557876,
      "learning_rate": 9.95953430587349e-06,
      "loss": 0.0441,
      "step": 1527
    },
    {
      "epoch": 0.4097613301153124,
      "grad_norm": 0.48393681157996615,
      "learning_rate": 9.95933597117758e-06,
      "loss": 0.0579,
      "step": 1528
    },
    {
      "epoch": 0.41002949852507375,
      "grad_norm": 0.4026835220522151,
      "learning_rate": 9.95913715360429e-06,
      "loss": 0.0451,
      "step": 1529
    },
    {
      "epoch": 0.4102976669348351,
      "grad_norm": 0.45003484204788097,
      "learning_rate": 9.958937853172982e-06,
      "loss": 0.0554,
      "step": 1530
    },
    {
      "epoch": 0.4105658353445964,
      "grad_norm": 0.4828875688079087,
      "learning_rate": 9.958738069903062e-06,
      "loss": 0.0384,
      "step": 1531
    },
    {
      "epoch": 0.41083400375435775,
      "grad_norm": 0.4112170299303783,
      "learning_rate": 9.958537803813981e-06,
      "loss": 0.0442,
      "step": 1532
    },
    {
      "epoch": 0.41110217216411904,
      "grad_norm": 0.5178106943809002,
      "learning_rate": 9.958337054925239e-06,
      "loss": 0.0574,
      "step": 1533
    },
    {
      "epoch": 0.4113703405738804,
      "grad_norm": 0.4614124403584122,
      "learning_rate": 9.958135823256381e-06,
      "loss": 0.0501,
      "step": 1534
    },
    {
      "epoch": 0.41163850898364174,
      "grad_norm": 0.47731808924633257,
      "learning_rate": 9.957934108827003e-06,
      "loss": 0.052,
      "step": 1535
    },
    {
      "epoch": 0.41190667739340303,
      "grad_norm": 0.3426622014631583,
      "learning_rate": 9.957731911656743e-06,
      "loss": 0.0368,
      "step": 1536
    },
    {
      "epoch": 0.4121748458031644,
      "grad_norm": 0.34897719156197654,
      "learning_rate": 9.957529231765289e-06,
      "loss": 0.0345,
      "step": 1537
    },
    {
      "epoch": 0.41244301421292573,
      "grad_norm": 0.4462920273034147,
      "learning_rate": 9.957326069172378e-06,
      "loss": 0.0528,
      "step": 1538
    },
    {
      "epoch": 0.412711182622687,
      "grad_norm": 0.41206181916970774,
      "learning_rate": 9.957122423897786e-06,
      "loss": 0.0429,
      "step": 1539
    },
    {
      "epoch": 0.41297935103244837,
      "grad_norm": 0.6963846821272254,
      "learning_rate": 9.956918295961347e-06,
      "loss": 0.0728,
      "step": 1540
    },
    {
      "epoch": 0.4132475194422097,
      "grad_norm": 0.34280354412948955,
      "learning_rate": 9.956713685382931e-06,
      "loss": 0.042,
      "step": 1541
    },
    {
      "epoch": 0.413515687851971,
      "grad_norm": 0.36298017184811465,
      "learning_rate": 9.956508592182465e-06,
      "loss": 0.0409,
      "step": 1542
    },
    {
      "epoch": 0.41378385626173236,
      "grad_norm": 0.7288312084774158,
      "learning_rate": 9.956303016379916e-06,
      "loss": 0.0566,
      "step": 1543
    },
    {
      "epoch": 0.4140520246714937,
      "grad_norm": 0.3622871056181386,
      "learning_rate": 9.956096957995302e-06,
      "loss": 0.0291,
      "step": 1544
    },
    {
      "epoch": 0.414320193081255,
      "grad_norm": 0.5026766141301878,
      "learning_rate": 9.955890417048686e-06,
      "loss": 0.0568,
      "step": 1545
    },
    {
      "epoch": 0.41458836149101636,
      "grad_norm": 0.5093066010614714,
      "learning_rate": 9.955683393560176e-06,
      "loss": 0.0657,
      "step": 1546
    },
    {
      "epoch": 0.4148565299007777,
      "grad_norm": 0.5802828379215412,
      "learning_rate": 9.955475887549933e-06,
      "loss": 0.0674,
      "step": 1547
    },
    {
      "epoch": 0.415124698310539,
      "grad_norm": 0.5230095098745715,
      "learning_rate": 9.955267899038157e-06,
      "loss": 0.0415,
      "step": 1548
    },
    {
      "epoch": 0.41539286672030035,
      "grad_norm": 0.517732704419722,
      "learning_rate": 9.955059428045102e-06,
      "loss": 0.0545,
      "step": 1549
    },
    {
      "epoch": 0.4156610351300617,
      "grad_norm": 0.5387615935592449,
      "learning_rate": 9.954850474591066e-06,
      "loss": 0.0471,
      "step": 1550
    },
    {
      "epoch": 0.415929203539823,
      "grad_norm": 0.3883734640432954,
      "learning_rate": 9.954641038696395e-06,
      "loss": 0.0452,
      "step": 1551
    },
    {
      "epoch": 0.41619737194958434,
      "grad_norm": 0.4776342629723951,
      "learning_rate": 9.954431120381482e-06,
      "loss": 0.0465,
      "step": 1552
    },
    {
      "epoch": 0.4164655403593457,
      "grad_norm": 0.39763965851013067,
      "learning_rate": 9.954220719666761e-06,
      "loss": 0.0385,
      "step": 1553
    },
    {
      "epoch": 0.416733708769107,
      "grad_norm": 0.3795763593740213,
      "learning_rate": 9.954009836572725e-06,
      "loss": 0.0373,
      "step": 1554
    },
    {
      "epoch": 0.41700187717886833,
      "grad_norm": 0.3210735210097809,
      "learning_rate": 9.953798471119902e-06,
      "loss": 0.038,
      "step": 1555
    },
    {
      "epoch": 0.4172700455886297,
      "grad_norm": 0.332998182005263,
      "learning_rate": 9.953586623328875e-06,
      "loss": 0.0291,
      "step": 1556
    },
    {
      "epoch": 0.417538213998391,
      "grad_norm": 0.5053798458218474,
      "learning_rate": 9.953374293220268e-06,
      "loss": 0.0673,
      "step": 1557
    },
    {
      "epoch": 0.4178063824081523,
      "grad_norm": 0.3886065554822884,
      "learning_rate": 9.95316148081476e-06,
      "loss": 0.0386,
      "step": 1558
    },
    {
      "epoch": 0.41807455081791367,
      "grad_norm": 0.6169379555091556,
      "learning_rate": 9.952948186133066e-06,
      "loss": 0.0533,
      "step": 1559
    },
    {
      "epoch": 0.41834271922767496,
      "grad_norm": 0.51392758421929,
      "learning_rate": 9.952734409195959e-06,
      "loss": 0.0504,
      "step": 1560
    },
    {
      "epoch": 0.4186108876374363,
      "grad_norm": 0.9874120763985208,
      "learning_rate": 9.952520150024251e-06,
      "loss": 0.0564,
      "step": 1561
    },
    {
      "epoch": 0.41887905604719766,
      "grad_norm": 0.457472097900679,
      "learning_rate": 9.952305408638806e-06,
      "loss": 0.0437,
      "step": 1562
    },
    {
      "epoch": 0.41914722445695896,
      "grad_norm": 0.39183715518657786,
      "learning_rate": 9.952090185060528e-06,
      "loss": 0.0379,
      "step": 1563
    },
    {
      "epoch": 0.4194153928667203,
      "grad_norm": 0.5822137581104712,
      "learning_rate": 9.951874479310379e-06,
      "loss": 0.0419,
      "step": 1564
    },
    {
      "epoch": 0.41968356127648165,
      "grad_norm": 0.7028137838978532,
      "learning_rate": 9.951658291409358e-06,
      "loss": 0.0492,
      "step": 1565
    },
    {
      "epoch": 0.41995172968624295,
      "grad_norm": 0.6141577153719346,
      "learning_rate": 9.951441621378516e-06,
      "loss": 0.0622,
      "step": 1566
    },
    {
      "epoch": 0.4202198980960043,
      "grad_norm": 0.546514140691345,
      "learning_rate": 9.951224469238949e-06,
      "loss": 0.0422,
      "step": 1567
    },
    {
      "epoch": 0.42048806650576565,
      "grad_norm": 0.4318719798859912,
      "learning_rate": 9.951006835011801e-06,
      "loss": 0.0328,
      "step": 1568
    },
    {
      "epoch": 0.42075623491552694,
      "grad_norm": 0.5127036067420944,
      "learning_rate": 9.950788718718262e-06,
      "loss": 0.0613,
      "step": 1569
    },
    {
      "epoch": 0.4210244033252883,
      "grad_norm": 0.5753542784949802,
      "learning_rate": 9.95057012037957e-06,
      "loss": 0.0585,
      "step": 1570
    },
    {
      "epoch": 0.42129257173504964,
      "grad_norm": 0.5743911053517858,
      "learning_rate": 9.950351040017007e-06,
      "loss": 0.0397,
      "step": 1571
    },
    {
      "epoch": 0.42156074014481093,
      "grad_norm": 0.3936589974756621,
      "learning_rate": 9.950131477651909e-06,
      "loss": 0.0377,
      "step": 1572
    },
    {
      "epoch": 0.4218289085545723,
      "grad_norm": 0.4549551754493138,
      "learning_rate": 9.94991143330565e-06,
      "loss": 0.0455,
      "step": 1573
    },
    {
      "epoch": 0.42209707696433363,
      "grad_norm": 0.4737514804531119,
      "learning_rate": 9.949690906999656e-06,
      "loss": 0.0546,
      "step": 1574
    },
    {
      "epoch": 0.4223652453740949,
      "grad_norm": 0.43066367347249324,
      "learning_rate": 9.9494698987554e-06,
      "loss": 0.0436,
      "step": 1575
    },
    {
      "epoch": 0.42263341378385627,
      "grad_norm": 0.5178512345061418,
      "learning_rate": 9.9492484085944e-06,
      "loss": 0.0434,
      "step": 1576
    },
    {
      "epoch": 0.42290158219361756,
      "grad_norm": 0.44720641002369177,
      "learning_rate": 9.949026436538223e-06,
      "loss": 0.0446,
      "step": 1577
    },
    {
      "epoch": 0.4231697506033789,
      "grad_norm": 0.44016599414656754,
      "learning_rate": 9.948803982608482e-06,
      "loss": 0.0335,
      "step": 1578
    },
    {
      "epoch": 0.42343791901314026,
      "grad_norm": 1.2207721154373405,
      "learning_rate": 9.948581046826835e-06,
      "loss": 0.0458,
      "step": 1579
    },
    {
      "epoch": 0.42370608742290156,
      "grad_norm": 0.4692588099437912,
      "learning_rate": 9.948357629214991e-06,
      "loss": 0.0356,
      "step": 1580
    },
    {
      "epoch": 0.4239742558326629,
      "grad_norm": 0.45585523027207786,
      "learning_rate": 9.9481337297947e-06,
      "loss": 0.055,
      "step": 1581
    },
    {
      "epoch": 0.42424242424242425,
      "grad_norm": 0.5481561511247544,
      "learning_rate": 9.947909348587766e-06,
      "loss": 0.0466,
      "step": 1582
    },
    {
      "epoch": 0.42451059265218555,
      "grad_norm": 0.3940284385551459,
      "learning_rate": 9.947684485616036e-06,
      "loss": 0.0459,
      "step": 1583
    },
    {
      "epoch": 0.4247787610619469,
      "grad_norm": 0.6957996115252733,
      "learning_rate": 9.947459140901402e-06,
      "loss": 0.0603,
      "step": 1584
    },
    {
      "epoch": 0.42504692947170825,
      "grad_norm": 0.36842237803172034,
      "learning_rate": 9.947233314465807e-06,
      "loss": 0.0417,
      "step": 1585
    },
    {
      "epoch": 0.42531509788146954,
      "grad_norm": 0.9956851587603684,
      "learning_rate": 9.94700700633124e-06,
      "loss": 0.0605,
      "step": 1586
    },
    {
      "epoch": 0.4255832662912309,
      "grad_norm": 0.4588119715929885,
      "learning_rate": 9.946780216519734e-06,
      "loss": 0.0577,
      "step": 1587
    },
    {
      "epoch": 0.42585143470099224,
      "grad_norm": 0.8052413969500355,
      "learning_rate": 9.946552945053371e-06,
      "loss": 0.074,
      "step": 1588
    },
    {
      "epoch": 0.42611960311075353,
      "grad_norm": 0.41446893683361724,
      "learning_rate": 9.946325191954283e-06,
      "loss": 0.0423,
      "step": 1589
    },
    {
      "epoch": 0.4263877715205149,
      "grad_norm": 0.6248109615112485,
      "learning_rate": 9.946096957244641e-06,
      "loss": 0.0503,
      "step": 1590
    },
    {
      "epoch": 0.42665593993027623,
      "grad_norm": 0.3754820451072399,
      "learning_rate": 9.94586824094667e-06,
      "loss": 0.0354,
      "step": 1591
    },
    {
      "epoch": 0.4269241083400375,
      "grad_norm": 0.5458444562760195,
      "learning_rate": 9.94563904308264e-06,
      "loss": 0.0422,
      "step": 1592
    },
    {
      "epoch": 0.42719227674979887,
      "grad_norm": 0.5745767930221711,
      "learning_rate": 9.945409363674867e-06,
      "loss": 0.0508,
      "step": 1593
    },
    {
      "epoch": 0.4274604451595602,
      "grad_norm": 0.4711404522052878,
      "learning_rate": 9.945179202745713e-06,
      "loss": 0.044,
      "step": 1594
    },
    {
      "epoch": 0.4277286135693215,
      "grad_norm": 0.6747971332877517,
      "learning_rate": 9.944948560317592e-06,
      "loss": 0.0412,
      "step": 1595
    },
    {
      "epoch": 0.42799678197908286,
      "grad_norm": 0.759244857477804,
      "learning_rate": 9.944717436412956e-06,
      "loss": 0.0508,
      "step": 1596
    },
    {
      "epoch": 0.4282649503888442,
      "grad_norm": 0.6738618283942054,
      "learning_rate": 9.944485831054311e-06,
      "loss": 0.0499,
      "step": 1597
    },
    {
      "epoch": 0.4285331187986055,
      "grad_norm": 0.4285162583446723,
      "learning_rate": 9.944253744264209e-06,
      "loss": 0.0374,
      "step": 1598
    },
    {
      "epoch": 0.42880128720836685,
      "grad_norm": 0.8442385233061325,
      "learning_rate": 9.944021176065247e-06,
      "loss": 0.0621,
      "step": 1599
    },
    {
      "epoch": 0.4290694556181282,
      "grad_norm": 0.42337405318536786,
      "learning_rate": 9.943788126480068e-06,
      "loss": 0.042,
      "step": 1600
    },
    {
      "epoch": 0.4293376240278895,
      "grad_norm": 0.4690496757975748,
      "learning_rate": 9.943554595531364e-06,
      "loss": 0.0389,
      "step": 1601
    },
    {
      "epoch": 0.42960579243765085,
      "grad_norm": 0.44661641944769725,
      "learning_rate": 9.943320583241876e-06,
      "loss": 0.037,
      "step": 1602
    },
    {
      "epoch": 0.4298739608474122,
      "grad_norm": 0.6738844918007444,
      "learning_rate": 9.943086089634387e-06,
      "loss": 0.0484,
      "step": 1603
    },
    {
      "epoch": 0.4301421292571735,
      "grad_norm": 0.5299063485742996,
      "learning_rate": 9.942851114731726e-06,
      "loss": 0.05,
      "step": 1604
    },
    {
      "epoch": 0.43041029766693484,
      "grad_norm": 0.6381442442925218,
      "learning_rate": 9.942615658556778e-06,
      "loss": 0.0485,
      "step": 1605
    },
    {
      "epoch": 0.4306784660766962,
      "grad_norm": 0.5640439949105651,
      "learning_rate": 9.942379721132464e-06,
      "loss": 0.0424,
      "step": 1606
    },
    {
      "epoch": 0.4309466344864575,
      "grad_norm": 0.8609338246697618,
      "learning_rate": 9.942143302481759e-06,
      "loss": 0.0514,
      "step": 1607
    },
    {
      "epoch": 0.43121480289621883,
      "grad_norm": 0.5061651565168902,
      "learning_rate": 9.941906402627684e-06,
      "loss": 0.0534,
      "step": 1608
    },
    {
      "epoch": 0.4314829713059802,
      "grad_norm": 0.415662373088618,
      "learning_rate": 9.9416690215933e-06,
      "loss": 0.0538,
      "step": 1609
    },
    {
      "epoch": 0.43175113971574147,
      "grad_norm": 0.6675060128771009,
      "learning_rate": 9.941431159401725e-06,
      "loss": 0.0453,
      "step": 1610
    },
    {
      "epoch": 0.4320193081255028,
      "grad_norm": 0.555454595793176,
      "learning_rate": 9.941192816076114e-06,
      "loss": 0.0719,
      "step": 1611
    },
    {
      "epoch": 0.43228747653526417,
      "grad_norm": 0.4434391905217857,
      "learning_rate": 9.94095399163968e-06,
      "loss": 0.0529,
      "step": 1612
    },
    {
      "epoch": 0.43255564494502546,
      "grad_norm": 0.4432154150312489,
      "learning_rate": 9.940714686115674e-06,
      "loss": 0.0479,
      "step": 1613
    },
    {
      "epoch": 0.4328238133547868,
      "grad_norm": 0.5696991475671687,
      "learning_rate": 9.940474899527397e-06,
      "loss": 0.0415,
      "step": 1614
    },
    {
      "epoch": 0.43309198176454816,
      "grad_norm": 0.5934137566000263,
      "learning_rate": 9.940234631898193e-06,
      "loss": 0.0444,
      "step": 1615
    },
    {
      "epoch": 0.43336015017430946,
      "grad_norm": 0.6787309663515398,
      "learning_rate": 9.939993883251462e-06,
      "loss": 0.0548,
      "step": 1616
    },
    {
      "epoch": 0.4336283185840708,
      "grad_norm": 0.5298444752480082,
      "learning_rate": 9.939752653610639e-06,
      "loss": 0.0413,
      "step": 1617
    },
    {
      "epoch": 0.43389648699383215,
      "grad_norm": 0.544173022269774,
      "learning_rate": 9.939510942999218e-06,
      "loss": 0.0529,
      "step": 1618
    },
    {
      "epoch": 0.43416465540359345,
      "grad_norm": 0.8399925652779808,
      "learning_rate": 9.939268751440728e-06,
      "loss": 0.053,
      "step": 1619
    },
    {
      "epoch": 0.4344328238133548,
      "grad_norm": 0.8372998831208353,
      "learning_rate": 9.939026078958755e-06,
      "loss": 0.0617,
      "step": 1620
    },
    {
      "epoch": 0.4347009922231161,
      "grad_norm": 0.5481772332624137,
      "learning_rate": 9.938782925576925e-06,
      "loss": 0.0489,
      "step": 1621
    },
    {
      "epoch": 0.43496916063287744,
      "grad_norm": 0.3957594716097898,
      "learning_rate": 9.938539291318913e-06,
      "loss": 0.0553,
      "step": 1622
    },
    {
      "epoch": 0.4352373290426388,
      "grad_norm": 0.5159704723397789,
      "learning_rate": 9.938295176208441e-06,
      "loss": 0.0525,
      "step": 1623
    },
    {
      "epoch": 0.4355054974524001,
      "grad_norm": 0.4112912373675489,
      "learning_rate": 9.93805058026928e-06,
      "loss": 0.0417,
      "step": 1624
    },
    {
      "epoch": 0.43577366586216143,
      "grad_norm": 0.7328767049079161,
      "learning_rate": 9.937805503525244e-06,
      "loss": 0.0394,
      "step": 1625
    },
    {
      "epoch": 0.4360418342719228,
      "grad_norm": 0.5292512846454595,
      "learning_rate": 9.937559946000196e-06,
      "loss": 0.0385,
      "step": 1626
    },
    {
      "epoch": 0.4363100026816841,
      "grad_norm": 0.6468614827634845,
      "learning_rate": 9.937313907718046e-06,
      "loss": 0.0485,
      "step": 1627
    },
    {
      "epoch": 0.4365781710914454,
      "grad_norm": 0.4505443998986794,
      "learning_rate": 9.937067388702748e-06,
      "loss": 0.061,
      "step": 1628
    },
    {
      "epoch": 0.43684633950120677,
      "grad_norm": 0.5597095803322025,
      "learning_rate": 9.936820388978306e-06,
      "loss": 0.043,
      "step": 1629
    },
    {
      "epoch": 0.43711450791096806,
      "grad_norm": 0.4728603467284444,
      "learning_rate": 9.936572908568768e-06,
      "loss": 0.0647,
      "step": 1630
    },
    {
      "epoch": 0.4373826763207294,
      "grad_norm": 0.4722589663682352,
      "learning_rate": 9.936324947498237e-06,
      "loss": 0.0386,
      "step": 1631
    },
    {
      "epoch": 0.43765084473049076,
      "grad_norm": 0.5969013106142882,
      "learning_rate": 9.936076505790848e-06,
      "loss": 0.0482,
      "step": 1632
    },
    {
      "epoch": 0.43791901314025206,
      "grad_norm": 0.6084121950394153,
      "learning_rate": 9.935827583470793e-06,
      "loss": 0.05,
      "step": 1633
    },
    {
      "epoch": 0.4381871815500134,
      "grad_norm": 0.7002806716948767,
      "learning_rate": 9.935578180562315e-06,
      "loss": 0.0389,
      "step": 1634
    },
    {
      "epoch": 0.43845534995977475,
      "grad_norm": 0.5758641398579936,
      "learning_rate": 9.93532829708969e-06,
      "loss": 0.0625,
      "step": 1635
    },
    {
      "epoch": 0.43872351836953605,
      "grad_norm": 0.44420550676278925,
      "learning_rate": 9.935077933077252e-06,
      "loss": 0.0355,
      "step": 1636
    },
    {
      "epoch": 0.4389916867792974,
      "grad_norm": 0.5035070446091369,
      "learning_rate": 9.93482708854938e-06,
      "loss": 0.0519,
      "step": 1637
    },
    {
      "epoch": 0.43925985518905875,
      "grad_norm": 0.7675657327100923,
      "learning_rate": 9.934575763530496e-06,
      "loss": 0.0627,
      "step": 1638
    },
    {
      "epoch": 0.43952802359882004,
      "grad_norm": 0.417443837502166,
      "learning_rate": 9.934323958045069e-06,
      "loss": 0.045,
      "step": 1639
    },
    {
      "epoch": 0.4397961920085814,
      "grad_norm": 0.4514709191998314,
      "learning_rate": 9.93407167211762e-06,
      "loss": 0.0402,
      "step": 1640
    },
    {
      "epoch": 0.44006436041834274,
      "grad_norm": 0.3617548510653869,
      "learning_rate": 9.933818905772713e-06,
      "loss": 0.0372,
      "step": 1641
    },
    {
      "epoch": 0.44033252882810403,
      "grad_norm": 0.5013704477166068,
      "learning_rate": 9.933565659034955e-06,
      "loss": 0.0592,
      "step": 1642
    },
    {
      "epoch": 0.4406006972378654,
      "grad_norm": 1.1095673450338919,
      "learning_rate": 9.933311931929008e-06,
      "loss": 0.0652,
      "step": 1643
    },
    {
      "epoch": 0.44086886564762673,
      "grad_norm": 0.44111484308396137,
      "learning_rate": 9.933057724479578e-06,
      "loss": 0.0456,
      "step": 1644
    },
    {
      "epoch": 0.441137034057388,
      "grad_norm": 0.6813778350113352,
      "learning_rate": 9.932803036711413e-06,
      "loss": 0.0533,
      "step": 1645
    },
    {
      "epoch": 0.44140520246714937,
      "grad_norm": 0.6524824535315231,
      "learning_rate": 9.932547868649313e-06,
      "loss": 0.071,
      "step": 1646
    },
    {
      "epoch": 0.4416733708769107,
      "grad_norm": 0.40175275304674,
      "learning_rate": 9.932292220318121e-06,
      "loss": 0.0481,
      "step": 1647
    },
    {
      "epoch": 0.441941539286672,
      "grad_norm": 0.4050965128307958,
      "learning_rate": 9.932036091742732e-06,
      "loss": 0.0392,
      "step": 1648
    },
    {
      "epoch": 0.44220970769643336,
      "grad_norm": 0.5467004541457118,
      "learning_rate": 9.931779482948083e-06,
      "loss": 0.0491,
      "step": 1649
    },
    {
      "epoch": 0.4424778761061947,
      "grad_norm": 0.44115926393665356,
      "learning_rate": 9.93152239395916e-06,
      "loss": 0.0383,
      "step": 1650
    },
    {
      "epoch": 0.442746044515956,
      "grad_norm": 0.4863668581054675,
      "learning_rate": 9.931264824800995e-06,
      "loss": 0.0452,
      "step": 1651
    },
    {
      "epoch": 0.44301421292571735,
      "grad_norm": 0.6932025326781432,
      "learning_rate": 9.931006775498664e-06,
      "loss": 0.0543,
      "step": 1652
    },
    {
      "epoch": 0.4432823813354787,
      "grad_norm": 0.629272932778374,
      "learning_rate": 9.930748246077296e-06,
      "loss": 0.0622,
      "step": 1653
    },
    {
      "epoch": 0.44355054974524,
      "grad_norm": 0.5376602941608818,
      "learning_rate": 9.930489236562063e-06,
      "loss": 0.0486,
      "step": 1654
    },
    {
      "epoch": 0.44381871815500135,
      "grad_norm": 0.46613971273142707,
      "learning_rate": 9.930229746978181e-06,
      "loss": 0.0504,
      "step": 1655
    },
    {
      "epoch": 0.4440868865647627,
      "grad_norm": 0.49500407773854394,
      "learning_rate": 9.92996977735092e-06,
      "loss": 0.052,
      "step": 1656
    },
    {
      "epoch": 0.444355054974524,
      "grad_norm": 0.6253522537145251,
      "learning_rate": 9.92970932770559e-06,
      "loss": 0.0706,
      "step": 1657
    },
    {
      "epoch": 0.44462322338428534,
      "grad_norm": 0.5504296330605859,
      "learning_rate": 9.92944839806755e-06,
      "loss": 0.0502,
      "step": 1658
    },
    {
      "epoch": 0.4448913917940467,
      "grad_norm": 0.4766863316440304,
      "learning_rate": 9.929186988462208e-06,
      "loss": 0.0368,
      "step": 1659
    },
    {
      "epoch": 0.445159560203808,
      "grad_norm": 0.5014563876664402,
      "learning_rate": 9.928925098915016e-06,
      "loss": 0.0375,
      "step": 1660
    },
    {
      "epoch": 0.44542772861356933,
      "grad_norm": 0.49760274952258365,
      "learning_rate": 9.928662729451472e-06,
      "loss": 0.041,
      "step": 1661
    },
    {
      "epoch": 0.4456958970233307,
      "grad_norm": 0.5118654949774181,
      "learning_rate": 9.928399880097124e-06,
      "loss": 0.0444,
      "step": 1662
    },
    {
      "epoch": 0.44596406543309197,
      "grad_norm": 0.7442468080500084,
      "learning_rate": 9.928136550877565e-06,
      "loss": 0.0526,
      "step": 1663
    },
    {
      "epoch": 0.4462322338428533,
      "grad_norm": 0.3719104078439192,
      "learning_rate": 9.927872741818432e-06,
      "loss": 0.0351,
      "step": 1664
    },
    {
      "epoch": 0.4465004022526146,
      "grad_norm": 0.39610326639199905,
      "learning_rate": 9.927608452945413e-06,
      "loss": 0.0425,
      "step": 1665
    },
    {
      "epoch": 0.44676857066237596,
      "grad_norm": 0.44652217705156433,
      "learning_rate": 9.927343684284245e-06,
      "loss": 0.0341,
      "step": 1666
    },
    {
      "epoch": 0.4470367390721373,
      "grad_norm": 0.42568090636904093,
      "learning_rate": 9.927078435860702e-06,
      "loss": 0.0352,
      "step": 1667
    },
    {
      "epoch": 0.4473049074818986,
      "grad_norm": 0.4157220147822246,
      "learning_rate": 9.926812707700614e-06,
      "loss": 0.0349,
      "step": 1668
    },
    {
      "epoch": 0.44757307589165995,
      "grad_norm": 0.6036301514601518,
      "learning_rate": 9.926546499829853e-06,
      "loss": 0.0417,
      "step": 1669
    },
    {
      "epoch": 0.4478412443014213,
      "grad_norm": 0.40777734814952366,
      "learning_rate": 9.926279812274338e-06,
      "loss": 0.0468,
      "step": 1670
    },
    {
      "epoch": 0.4481094127111826,
      "grad_norm": 0.4870265793774108,
      "learning_rate": 9.926012645060037e-06,
      "loss": 0.0462,
      "step": 1671
    },
    {
      "epoch": 0.44837758112094395,
      "grad_norm": 0.4426130121868837,
      "learning_rate": 9.925744998212962e-06,
      "loss": 0.0386,
      "step": 1672
    },
    {
      "epoch": 0.4486457495307053,
      "grad_norm": 0.48820723774934593,
      "learning_rate": 9.925476871759177e-06,
      "loss": 0.0551,
      "step": 1673
    },
    {
      "epoch": 0.4489139179404666,
      "grad_norm": 0.5068056004251991,
      "learning_rate": 9.925208265724782e-06,
      "loss": 0.0474,
      "step": 1674
    },
    {
      "epoch": 0.44918208635022794,
      "grad_norm": 0.46199564110468133,
      "learning_rate": 9.924939180135938e-06,
      "loss": 0.0339,
      "step": 1675
    },
    {
      "epoch": 0.4494502547599893,
      "grad_norm": 0.6675978079712932,
      "learning_rate": 9.92466961501884e-06,
      "loss": 0.0448,
      "step": 1676
    },
    {
      "epoch": 0.4497184231697506,
      "grad_norm": 0.3429677272426827,
      "learning_rate": 9.924399570399737e-06,
      "loss": 0.0273,
      "step": 1677
    },
    {
      "epoch": 0.44998659157951193,
      "grad_norm": 0.4598651121029453,
      "learning_rate": 9.92412904630492e-06,
      "loss": 0.0404,
      "step": 1678
    },
    {
      "epoch": 0.4502547599892733,
      "grad_norm": 0.6511549196432105,
      "learning_rate": 9.923858042760732e-06,
      "loss": 0.0507,
      "step": 1679
    },
    {
      "epoch": 0.4505229283990346,
      "grad_norm": 0.4503890355111746,
      "learning_rate": 9.923586559793559e-06,
      "loss": 0.0368,
      "step": 1680
    },
    {
      "epoch": 0.4507910968087959,
      "grad_norm": 0.4385224459183911,
      "learning_rate": 9.923314597429834e-06,
      "loss": 0.0411,
      "step": 1681
    },
    {
      "epoch": 0.45105926521855727,
      "grad_norm": 0.5908317341630881,
      "learning_rate": 9.92304215569604e-06,
      "loss": 0.0487,
      "step": 1682
    },
    {
      "epoch": 0.45132743362831856,
      "grad_norm": 0.4569852608040678,
      "learning_rate": 9.9227692346187e-06,
      "loss": 0.0428,
      "step": 1683
    },
    {
      "epoch": 0.4515956020380799,
      "grad_norm": 0.4224631141995175,
      "learning_rate": 9.922495834224391e-06,
      "loss": 0.0368,
      "step": 1684
    },
    {
      "epoch": 0.45186377044784126,
      "grad_norm": 0.4727641560899038,
      "learning_rate": 9.922221954539728e-06,
      "loss": 0.0378,
      "step": 1685
    },
    {
      "epoch": 0.45213193885760256,
      "grad_norm": 0.3404200421747889,
      "learning_rate": 9.921947595591385e-06,
      "loss": 0.0368,
      "step": 1686
    },
    {
      "epoch": 0.4524001072673639,
      "grad_norm": 0.4576762270549373,
      "learning_rate": 9.92167275740607e-06,
      "loss": 0.0344,
      "step": 1687
    },
    {
      "epoch": 0.45266827567712525,
      "grad_norm": 0.5073019169766373,
      "learning_rate": 9.921397440010548e-06,
      "loss": 0.045,
      "step": 1688
    },
    {
      "epoch": 0.45293644408688655,
      "grad_norm": 0.6825397825776557,
      "learning_rate": 9.92112164343162e-06,
      "loss": 0.0527,
      "step": 1689
    },
    {
      "epoch": 0.4532046124966479,
      "grad_norm": 0.6321986630393275,
      "learning_rate": 9.920845367696145e-06,
      "loss": 0.0512,
      "step": 1690
    },
    {
      "epoch": 0.45347278090640925,
      "grad_norm": 0.43329048651791213,
      "learning_rate": 9.920568612831021e-06,
      "loss": 0.0489,
      "step": 1691
    },
    {
      "epoch": 0.45374094931617054,
      "grad_norm": 0.3764795309768748,
      "learning_rate": 9.920291378863193e-06,
      "loss": 0.0422,
      "step": 1692
    },
    {
      "epoch": 0.4540091177259319,
      "grad_norm": 0.5019906052233006,
      "learning_rate": 9.92001366581966e-06,
      "loss": 0.0385,
      "step": 1693
    },
    {
      "epoch": 0.45427728613569324,
      "grad_norm": 1.4052075032240339,
      "learning_rate": 9.919735473727457e-06,
      "loss": 0.0491,
      "step": 1694
    },
    {
      "epoch": 0.45454545454545453,
      "grad_norm": 0.4529497522787487,
      "learning_rate": 9.919456802613672e-06,
      "loss": 0.0521,
      "step": 1695
    },
    {
      "epoch": 0.4548136229552159,
      "grad_norm": 0.6191353712941574,
      "learning_rate": 9.91917765250544e-06,
      "loss": 0.051,
      "step": 1696
    },
    {
      "epoch": 0.45508179136497723,
      "grad_norm": 0.3268028784558863,
      "learning_rate": 9.91889802342994e-06,
      "loss": 0.0313,
      "step": 1697
    },
    {
      "epoch": 0.4553499597747385,
      "grad_norm": 0.4291915712611819,
      "learning_rate": 9.918617915414397e-06,
      "loss": 0.0418,
      "step": 1698
    },
    {
      "epoch": 0.45561812818449987,
      "grad_norm": 0.6413031865675974,
      "learning_rate": 9.918337328486088e-06,
      "loss": 0.0703,
      "step": 1699
    },
    {
      "epoch": 0.4558862965942612,
      "grad_norm": 0.5361991268706133,
      "learning_rate": 9.91805626267233e-06,
      "loss": 0.0595,
      "step": 1700
    },
    {
      "epoch": 0.4561544650040225,
      "grad_norm": 0.45859914276293234,
      "learning_rate": 9.917774718000494e-06,
      "loss": 0.0363,
      "step": 1701
    },
    {
      "epoch": 0.45642263341378386,
      "grad_norm": 0.40210384626960993,
      "learning_rate": 9.917492694497988e-06,
      "loss": 0.0366,
      "step": 1702
    },
    {
      "epoch": 0.4566908018235452,
      "grad_norm": 0.3537240608016449,
      "learning_rate": 9.917210192192273e-06,
      "loss": 0.0334,
      "step": 1703
    },
    {
      "epoch": 0.4569589702333065,
      "grad_norm": 0.6405893452238196,
      "learning_rate": 9.916927211110858e-06,
      "loss": 0.0643,
      "step": 1704
    },
    {
      "epoch": 0.45722713864306785,
      "grad_norm": 0.591703100228027,
      "learning_rate": 9.916643751281294e-06,
      "loss": 0.0439,
      "step": 1705
    },
    {
      "epoch": 0.4574953070528292,
      "grad_norm": 0.5620820011368417,
      "learning_rate": 9.916359812731183e-06,
      "loss": 0.0408,
      "step": 1706
    },
    {
      "epoch": 0.4577634754625905,
      "grad_norm": 0.6006057309355749,
      "learning_rate": 9.916075395488167e-06,
      "loss": 0.0484,
      "step": 1707
    },
    {
      "epoch": 0.45803164387235185,
      "grad_norm": 0.5364452630668544,
      "learning_rate": 9.915790499579944e-06,
      "loss": 0.0477,
      "step": 1708
    },
    {
      "epoch": 0.4582998122821132,
      "grad_norm": 1.2791693088141756,
      "learning_rate": 9.91550512503425e-06,
      "loss": 0.0504,
      "step": 1709
    },
    {
      "epoch": 0.4585679806918745,
      "grad_norm": 0.5055088373475854,
      "learning_rate": 9.915219271878873e-06,
      "loss": 0.0491,
      "step": 1710
    },
    {
      "epoch": 0.45883614910163584,
      "grad_norm": 0.7445202228130572,
      "learning_rate": 9.914932940141644e-06,
      "loss": 0.0515,
      "step": 1711
    },
    {
      "epoch": 0.45910431751139713,
      "grad_norm": 0.39566929014430885,
      "learning_rate": 9.914646129850445e-06,
      "loss": 0.042,
      "step": 1712
    },
    {
      "epoch": 0.4593724859211585,
      "grad_norm": 0.3038402514697219,
      "learning_rate": 9.9143588410332e-06,
      "loss": 0.0258,
      "step": 1713
    },
    {
      "epoch": 0.45964065433091983,
      "grad_norm": 0.5075452566281609,
      "learning_rate": 9.914071073717882e-06,
      "loss": 0.0496,
      "step": 1714
    },
    {
      "epoch": 0.4599088227406811,
      "grad_norm": 0.8474723888530709,
      "learning_rate": 9.913782827932509e-06,
      "loss": 0.0457,
      "step": 1715
    },
    {
      "epoch": 0.46017699115044247,
      "grad_norm": 0.42358759439897153,
      "learning_rate": 9.91349410370515e-06,
      "loss": 0.0398,
      "step": 1716
    },
    {
      "epoch": 0.4604451595602038,
      "grad_norm": 0.4176654890436293,
      "learning_rate": 9.913204901063915e-06,
      "loss": 0.0488,
      "step": 1717
    },
    {
      "epoch": 0.4607133279699651,
      "grad_norm": 0.34955004699368303,
      "learning_rate": 9.912915220036961e-06,
      "loss": 0.0415,
      "step": 1718
    },
    {
      "epoch": 0.46098149637972646,
      "grad_norm": 0.4963476647141926,
      "learning_rate": 9.912625060652496e-06,
      "loss": 0.0667,
      "step": 1719
    },
    {
      "epoch": 0.4612496647894878,
      "grad_norm": 0.6313162827924645,
      "learning_rate": 9.912334422938773e-06,
      "loss": 0.0465,
      "step": 1720
    },
    {
      "epoch": 0.4615178331992491,
      "grad_norm": 0.5378640728108217,
      "learning_rate": 9.912043306924088e-06,
      "loss": 0.0635,
      "step": 1721
    },
    {
      "epoch": 0.46178600160901045,
      "grad_norm": 0.35823355752758573,
      "learning_rate": 9.911751712636789e-06,
      "loss": 0.0355,
      "step": 1722
    },
    {
      "epoch": 0.4620541700187718,
      "grad_norm": 0.9441209544404643,
      "learning_rate": 9.911459640105266e-06,
      "loss": 0.0554,
      "step": 1723
    },
    {
      "epoch": 0.4623223384285331,
      "grad_norm": 0.49998182992054657,
      "learning_rate": 9.911167089357957e-06,
      "loss": 0.0484,
      "step": 1724
    },
    {
      "epoch": 0.46259050683829445,
      "grad_norm": 0.46891903368928706,
      "learning_rate": 9.910874060423345e-06,
      "loss": 0.0441,
      "step": 1725
    },
    {
      "epoch": 0.4628586752480558,
      "grad_norm": 0.49538066604237285,
      "learning_rate": 9.910580553329966e-06,
      "loss": 0.0354,
      "step": 1726
    },
    {
      "epoch": 0.4631268436578171,
      "grad_norm": 0.6188115327983452,
      "learning_rate": 9.910286568106398e-06,
      "loss": 0.0744,
      "step": 1727
    },
    {
      "epoch": 0.46339501206757844,
      "grad_norm": 0.6378434879759454,
      "learning_rate": 9.90999210478126e-06,
      "loss": 0.0448,
      "step": 1728
    },
    {
      "epoch": 0.4636631804773398,
      "grad_norm": 0.3240291798483186,
      "learning_rate": 9.909697163383228e-06,
      "loss": 0.0379,
      "step": 1729
    },
    {
      "epoch": 0.4639313488871011,
      "grad_norm": 0.4050012435184163,
      "learning_rate": 9.909401743941018e-06,
      "loss": 0.0399,
      "step": 1730
    },
    {
      "epoch": 0.46419951729686243,
      "grad_norm": 0.5529102795659663,
      "learning_rate": 9.909105846483394e-06,
      "loss": 0.045,
      "step": 1731
    },
    {
      "epoch": 0.4644676857066238,
      "grad_norm": 0.48061242299094764,
      "learning_rate": 9.908809471039168e-06,
      "loss": 0.0313,
      "step": 1732
    },
    {
      "epoch": 0.46473585411638507,
      "grad_norm": 0.35225216044219293,
      "learning_rate": 9.908512617637195e-06,
      "loss": 0.0337,
      "step": 1733
    },
    {
      "epoch": 0.4650040225261464,
      "grad_norm": 0.33861235300653747,
      "learning_rate": 9.908215286306381e-06,
      "loss": 0.0372,
      "step": 1734
    },
    {
      "epoch": 0.46527219093590777,
      "grad_norm": 0.7720102132987724,
      "learning_rate": 9.907917477075677e-06,
      "loss": 0.0591,
      "step": 1735
    },
    {
      "epoch": 0.46554035934566906,
      "grad_norm": 0.6628331823992449,
      "learning_rate": 9.907619189974078e-06,
      "loss": 0.056,
      "step": 1736
    },
    {
      "epoch": 0.4658085277554304,
      "grad_norm": 0.4207837347947557,
      "learning_rate": 9.907320425030627e-06,
      "loss": 0.0334,
      "step": 1737
    },
    {
      "epoch": 0.46607669616519176,
      "grad_norm": 0.3858327041238357,
      "learning_rate": 9.907021182274415e-06,
      "loss": 0.0425,
      "step": 1738
    },
    {
      "epoch": 0.46634486457495306,
      "grad_norm": 0.6451169304631398,
      "learning_rate": 9.906721461734579e-06,
      "loss": 0.0741,
      "step": 1739
    },
    {
      "epoch": 0.4666130329847144,
      "grad_norm": 1.058262038340549,
      "learning_rate": 9.9064212634403e-06,
      "loss": 0.0477,
      "step": 1740
    },
    {
      "epoch": 0.46688120139447575,
      "grad_norm": 0.72372869651805,
      "learning_rate": 9.906120587420811e-06,
      "loss": 0.0474,
      "step": 1741
    },
    {
      "epoch": 0.46714936980423705,
      "grad_norm": 0.39987774915037994,
      "learning_rate": 9.905819433705383e-06,
      "loss": 0.0329,
      "step": 1742
    },
    {
      "epoch": 0.4674175382139984,
      "grad_norm": 0.6041263065761336,
      "learning_rate": 9.905517802323345e-06,
      "loss": 0.0593,
      "step": 1743
    },
    {
      "epoch": 0.46768570662375974,
      "grad_norm": 0.6718618450696869,
      "learning_rate": 9.90521569330406e-06,
      "loss": 0.0611,
      "step": 1744
    },
    {
      "epoch": 0.46795387503352104,
      "grad_norm": 0.5251679506324479,
      "learning_rate": 9.904913106676948e-06,
      "loss": 0.0389,
      "step": 1745
    },
    {
      "epoch": 0.4682220434432824,
      "grad_norm": 0.4968956157918924,
      "learning_rate": 9.904610042471468e-06,
      "loss": 0.0379,
      "step": 1746
    },
    {
      "epoch": 0.46849021185304374,
      "grad_norm": 0.4786245883249044,
      "learning_rate": 9.904306500717129e-06,
      "loss": 0.0606,
      "step": 1747
    },
    {
      "epoch": 0.46875838026280503,
      "grad_norm": 0.33940635055566865,
      "learning_rate": 9.904002481443487e-06,
      "loss": 0.0289,
      "step": 1748
    },
    {
      "epoch": 0.4690265486725664,
      "grad_norm": 0.5693703898372524,
      "learning_rate": 9.903697984680143e-06,
      "loss": 0.0434,
      "step": 1749
    },
    {
      "epoch": 0.46929471708232773,
      "grad_norm": 0.8198444706976998,
      "learning_rate": 9.903393010456745e-06,
      "loss": 0.0429,
      "step": 1750
    },
    {
      "epoch": 0.469562885492089,
      "grad_norm": 0.6508211644006137,
      "learning_rate": 9.90308755880299e-06,
      "loss": 0.0654,
      "step": 1751
    },
    {
      "epoch": 0.46983105390185037,
      "grad_norm": 0.6950013915623617,
      "learning_rate": 9.902781629748614e-06,
      "loss": 0.0502,
      "step": 1752
    },
    {
      "epoch": 0.4700992223116117,
      "grad_norm": 0.4375822632606668,
      "learning_rate": 9.902475223323408e-06,
      "loss": 0.0378,
      "step": 1753
    },
    {
      "epoch": 0.470367390721373,
      "grad_norm": 0.5975537899050268,
      "learning_rate": 9.902168339557208e-06,
      "loss": 0.0352,
      "step": 1754
    },
    {
      "epoch": 0.47063555913113436,
      "grad_norm": 0.6701800057022425,
      "learning_rate": 9.901860978479889e-06,
      "loss": 0.0491,
      "step": 1755
    },
    {
      "epoch": 0.47090372754089566,
      "grad_norm": 0.6375782576353912,
      "learning_rate": 9.901553140121382e-06,
      "loss": 0.0466,
      "step": 1756
    },
    {
      "epoch": 0.471171895950657,
      "grad_norm": 0.7874985228200917,
      "learning_rate": 9.90124482451166e-06,
      "loss": 0.0387,
      "step": 1757
    },
    {
      "epoch": 0.47144006436041835,
      "grad_norm": 0.5944394117621088,
      "learning_rate": 9.900936031680739e-06,
      "loss": 0.0492,
      "step": 1758
    },
    {
      "epoch": 0.47170823277017965,
      "grad_norm": 0.517331602287784,
      "learning_rate": 9.900626761658691e-06,
      "loss": 0.0391,
      "step": 1759
    },
    {
      "epoch": 0.471976401179941,
      "grad_norm": 0.48643856120626966,
      "learning_rate": 9.900317014475624e-06,
      "loss": 0.0432,
      "step": 1760
    },
    {
      "epoch": 0.47224456958970235,
      "grad_norm": 0.4547546331936923,
      "learning_rate": 9.900006790161701e-06,
      "loss": 0.0434,
      "step": 1761
    },
    {
      "epoch": 0.47251273799946364,
      "grad_norm": 0.5777305007089715,
      "learning_rate": 9.899696088747127e-06,
      "loss": 0.0532,
      "step": 1762
    },
    {
      "epoch": 0.472780906409225,
      "grad_norm": 0.4327120754997299,
      "learning_rate": 9.899384910262151e-06,
      "loss": 0.0457,
      "step": 1763
    },
    {
      "epoch": 0.47304907481898634,
      "grad_norm": 0.5010195261192465,
      "learning_rate": 9.899073254737076e-06,
      "loss": 0.0445,
      "step": 1764
    },
    {
      "epoch": 0.47331724322874763,
      "grad_norm": 0.9244833450364123,
      "learning_rate": 9.898761122202242e-06,
      "loss": 0.0436,
      "step": 1765
    },
    {
      "epoch": 0.473585411638509,
      "grad_norm": 0.6626612406826882,
      "learning_rate": 9.898448512688046e-06,
      "loss": 0.0615,
      "step": 1766
    },
    {
      "epoch": 0.47385358004827033,
      "grad_norm": 0.32932030147486296,
      "learning_rate": 9.898135426224923e-06,
      "loss": 0.0296,
      "step": 1767
    },
    {
      "epoch": 0.4741217484580316,
      "grad_norm": 0.48407560467866817,
      "learning_rate": 9.897821862843357e-06,
      "loss": 0.0539,
      "step": 1768
    },
    {
      "epoch": 0.47438991686779297,
      "grad_norm": 0.54509267539949,
      "learning_rate": 9.89750782257388e-06,
      "loss": 0.0461,
      "step": 1769
    },
    {
      "epoch": 0.4746580852775543,
      "grad_norm": 0.596702990208063,
      "learning_rate": 9.89719330544707e-06,
      "loss": 0.0402,
      "step": 1770
    },
    {
      "epoch": 0.4749262536873156,
      "grad_norm": 0.3935959320536733,
      "learning_rate": 9.896878311493549e-06,
      "loss": 0.0387,
      "step": 1771
    },
    {
      "epoch": 0.47519442209707696,
      "grad_norm": 0.44863894068749766,
      "learning_rate": 9.896562840743989e-06,
      "loss": 0.0506,
      "step": 1772
    },
    {
      "epoch": 0.4754625905068383,
      "grad_norm": 0.6138091048140234,
      "learning_rate": 9.896246893229102e-06,
      "loss": 0.07,
      "step": 1773
    },
    {
      "epoch": 0.4757307589165996,
      "grad_norm": 0.3417915825115119,
      "learning_rate": 9.895930468979658e-06,
      "loss": 0.0299,
      "step": 1774
    },
    {
      "epoch": 0.47599892732636095,
      "grad_norm": 0.43282816637646726,
      "learning_rate": 9.89561356802646e-06,
      "loss": 0.0368,
      "step": 1775
    },
    {
      "epoch": 0.4762670957361223,
      "grad_norm": 0.45485755304282816,
      "learning_rate": 9.895296190400368e-06,
      "loss": 0.0378,
      "step": 1776
    },
    {
      "epoch": 0.4765352641458836,
      "grad_norm": 0.4769666416615768,
      "learning_rate": 9.894978336132281e-06,
      "loss": 0.0399,
      "step": 1777
    },
    {
      "epoch": 0.47680343255564495,
      "grad_norm": 0.5523558325377964,
      "learning_rate": 9.894660005253152e-06,
      "loss": 0.052,
      "step": 1778
    },
    {
      "epoch": 0.4770716009654063,
      "grad_norm": 0.6889942355286874,
      "learning_rate": 9.89434119779397e-06,
      "loss": 0.046,
      "step": 1779
    },
    {
      "epoch": 0.4773397693751676,
      "grad_norm": 0.3732427447516459,
      "learning_rate": 9.894021913785782e-06,
      "loss": 0.0313,
      "step": 1780
    },
    {
      "epoch": 0.47760793778492894,
      "grad_norm": 0.5736278985682072,
      "learning_rate": 9.893702153259674e-06,
      "loss": 0.0395,
      "step": 1781
    },
    {
      "epoch": 0.4778761061946903,
      "grad_norm": 0.4799894240219493,
      "learning_rate": 9.893381916246778e-06,
      "loss": 0.0405,
      "step": 1782
    },
    {
      "epoch": 0.4781442746044516,
      "grad_norm": 0.6258538662251879,
      "learning_rate": 9.893061202778278e-06,
      "loss": 0.0656,
      "step": 1783
    },
    {
      "epoch": 0.47841244301421293,
      "grad_norm": 0.5109226239200082,
      "learning_rate": 9.8927400128854e-06,
      "loss": 0.0441,
      "step": 1784
    },
    {
      "epoch": 0.4786806114239743,
      "grad_norm": 0.492922324867543,
      "learning_rate": 9.892418346599415e-06,
      "loss": 0.0364,
      "step": 1785
    },
    {
      "epoch": 0.47894877983373557,
      "grad_norm": 0.40374312419019354,
      "learning_rate": 9.892096203951646e-06,
      "loss": 0.0314,
      "step": 1786
    },
    {
      "epoch": 0.4792169482434969,
      "grad_norm": 0.4159149521602461,
      "learning_rate": 9.891773584973457e-06,
      "loss": 0.0311,
      "step": 1787
    },
    {
      "epoch": 0.47948511665325827,
      "grad_norm": 0.4636132528323613,
      "learning_rate": 9.891450489696262e-06,
      "loss": 0.0376,
      "step": 1788
    },
    {
      "epoch": 0.47975328506301956,
      "grad_norm": 0.6025106729415012,
      "learning_rate": 9.89112691815152e-06,
      "loss": 0.0374,
      "step": 1789
    },
    {
      "epoch": 0.4800214534727809,
      "grad_norm": 0.4393233777855714,
      "learning_rate": 9.890802870370735e-06,
      "loss": 0.0448,
      "step": 1790
    },
    {
      "epoch": 0.48028962188254226,
      "grad_norm": 0.5201045352328328,
      "learning_rate": 9.89047834638546e-06,
      "loss": 0.0532,
      "step": 1791
    },
    {
      "epoch": 0.48055779029230355,
      "grad_norm": 0.5664281761949992,
      "learning_rate": 9.890153346227291e-06,
      "loss": 0.0479,
      "step": 1792
    },
    {
      "epoch": 0.4808259587020649,
      "grad_norm": 0.36924989986991813,
      "learning_rate": 9.889827869927875e-06,
      "loss": 0.0383,
      "step": 1793
    },
    {
      "epoch": 0.48109412711182625,
      "grad_norm": 0.3925287220558661,
      "learning_rate": 9.889501917518903e-06,
      "loss": 0.038,
      "step": 1794
    },
    {
      "epoch": 0.48136229552158755,
      "grad_norm": 0.4753518632399133,
      "learning_rate": 9.889175489032107e-06,
      "loss": 0.0447,
      "step": 1795
    },
    {
      "epoch": 0.4816304639313489,
      "grad_norm": 0.8029924703751461,
      "learning_rate": 9.888848584499275e-06,
      "loss": 0.0505,
      "step": 1796
    },
    {
      "epoch": 0.48189863234111024,
      "grad_norm": 0.4922593191622795,
      "learning_rate": 9.888521203952238e-06,
      "loss": 0.0435,
      "step": 1797
    },
    {
      "epoch": 0.48216680075087154,
      "grad_norm": 0.39693139987327125,
      "learning_rate": 9.88819334742287e-06,
      "loss": 0.0341,
      "step": 1798
    },
    {
      "epoch": 0.4824349691606329,
      "grad_norm": 0.46920216394789727,
      "learning_rate": 9.887865014943093e-06,
      "loss": 0.0515,
      "step": 1799
    },
    {
      "epoch": 0.4827031375703942,
      "grad_norm": 1.2314364810929457,
      "learning_rate": 9.887536206544876e-06,
      "loss": 0.0496,
      "step": 1800
    },
    {
      "epoch": 0.48297130598015553,
      "grad_norm": 0.47841482873956864,
      "learning_rate": 9.887206922260234e-06,
      "loss": 0.0635,
      "step": 1801
    },
    {
      "epoch": 0.4832394743899169,
      "grad_norm": 0.3660634008024165,
      "learning_rate": 9.886877162121231e-06,
      "loss": 0.0337,
      "step": 1802
    },
    {
      "epoch": 0.4835076427996782,
      "grad_norm": 0.6591664315646486,
      "learning_rate": 9.886546926159972e-06,
      "loss": 0.0659,
      "step": 1803
    },
    {
      "epoch": 0.4837758112094395,
      "grad_norm": 0.5540911662550497,
      "learning_rate": 9.886216214408612e-06,
      "loss": 0.0466,
      "step": 1804
    },
    {
      "epoch": 0.48404397961920087,
      "grad_norm": 0.49402038189082687,
      "learning_rate": 9.885885026899353e-06,
      "loss": 0.0588,
      "step": 1805
    },
    {
      "epoch": 0.48431214802896216,
      "grad_norm": 0.6205308999880106,
      "learning_rate": 9.88555336366444e-06,
      "loss": 0.0556,
      "step": 1806
    },
    {
      "epoch": 0.4845803164387235,
      "grad_norm": 0.39004279198349906,
      "learning_rate": 9.885221224736168e-06,
      "loss": 0.0413,
      "step": 1807
    },
    {
      "epoch": 0.48484848484848486,
      "grad_norm": 0.5759200245907066,
      "learning_rate": 9.884888610146875e-06,
      "loss": 0.0482,
      "step": 1808
    },
    {
      "epoch": 0.48511665325824616,
      "grad_norm": 0.3489770836013558,
      "learning_rate": 9.884555519928946e-06,
      "loss": 0.0399,
      "step": 1809
    },
    {
      "epoch": 0.4853848216680075,
      "grad_norm": 0.3878163578482397,
      "learning_rate": 9.884221954114814e-06,
      "loss": 0.0386,
      "step": 1810
    },
    {
      "epoch": 0.48565299007776885,
      "grad_norm": 0.7810601410169329,
      "learning_rate": 9.883887912736959e-06,
      "loss": 0.0428,
      "step": 1811
    },
    {
      "epoch": 0.48592115848753015,
      "grad_norm": 0.5659781609761343,
      "learning_rate": 9.883553395827905e-06,
      "loss": 0.0363,
      "step": 1812
    },
    {
      "epoch": 0.4861893268972915,
      "grad_norm": 0.44707113790463116,
      "learning_rate": 9.88321840342022e-06,
      "loss": 0.0417,
      "step": 1813
    },
    {
      "epoch": 0.48645749530705285,
      "grad_norm": 0.822240914556162,
      "learning_rate": 9.882882935546525e-06,
      "loss": 0.0379,
      "step": 1814
    },
    {
      "epoch": 0.48672566371681414,
      "grad_norm": 0.4147258047315414,
      "learning_rate": 9.882546992239483e-06,
      "loss": 0.0399,
      "step": 1815
    },
    {
      "epoch": 0.4869938321265755,
      "grad_norm": 0.4615974285102183,
      "learning_rate": 9.882210573531803e-06,
      "loss": 0.0423,
      "step": 1816
    },
    {
      "epoch": 0.48726200053633684,
      "grad_norm": 0.5103617488938941,
      "learning_rate": 9.881873679456243e-06,
      "loss": 0.0392,
      "step": 1817
    },
    {
      "epoch": 0.48753016894609813,
      "grad_norm": 0.3251048453102637,
      "learning_rate": 9.8815363100456e-06,
      "loss": 0.0311,
      "step": 1818
    },
    {
      "epoch": 0.4877983373558595,
      "grad_norm": 0.5939575424655739,
      "learning_rate": 9.88119846533273e-06,
      "loss": 0.0624,
      "step": 1819
    },
    {
      "epoch": 0.48806650576562083,
      "grad_norm": 0.48243322721865456,
      "learning_rate": 9.880860145350525e-06,
      "loss": 0.0485,
      "step": 1820
    },
    {
      "epoch": 0.4883346741753821,
      "grad_norm": 0.5092356182948604,
      "learning_rate": 9.880521350131925e-06,
      "loss": 0.0406,
      "step": 1821
    },
    {
      "epoch": 0.48860284258514347,
      "grad_norm": 0.48925062079129156,
      "learning_rate": 9.880182079709917e-06,
      "loss": 0.0384,
      "step": 1822
    },
    {
      "epoch": 0.4888710109949048,
      "grad_norm": 0.5668046082032859,
      "learning_rate": 9.879842334117538e-06,
      "loss": 0.0424,
      "step": 1823
    },
    {
      "epoch": 0.4891391794046661,
      "grad_norm": 0.439398472948795,
      "learning_rate": 9.879502113387867e-06,
      "loss": 0.0493,
      "step": 1824
    },
    {
      "epoch": 0.48940734781442746,
      "grad_norm": 0.36057799895253606,
      "learning_rate": 9.879161417554031e-06,
      "loss": 0.0302,
      "step": 1825
    },
    {
      "epoch": 0.4896755162241888,
      "grad_norm": 0.6085256437879389,
      "learning_rate": 9.878820246649198e-06,
      "loss": 0.0392,
      "step": 1826
    },
    {
      "epoch": 0.4899436846339501,
      "grad_norm": 0.4552925694922547,
      "learning_rate": 9.878478600706595e-06,
      "loss": 0.0496,
      "step": 1827
    },
    {
      "epoch": 0.49021185304371145,
      "grad_norm": 0.34872506471865444,
      "learning_rate": 9.87813647975948e-06,
      "loss": 0.0364,
      "step": 1828
    },
    {
      "epoch": 0.4904800214534728,
      "grad_norm": 0.4826648926075266,
      "learning_rate": 9.877793883841169e-06,
      "loss": 0.0437,
      "step": 1829
    },
    {
      "epoch": 0.4907481898632341,
      "grad_norm": 0.42159502591897136,
      "learning_rate": 9.877450812985016e-06,
      "loss": 0.0388,
      "step": 1830
    },
    {
      "epoch": 0.49101635827299545,
      "grad_norm": 0.391279464763829,
      "learning_rate": 9.877107267224429e-06,
      "loss": 0.0385,
      "step": 1831
    },
    {
      "epoch": 0.4912845266827568,
      "grad_norm": 0.6040449871499166,
      "learning_rate": 9.876763246592854e-06,
      "loss": 0.05,
      "step": 1832
    },
    {
      "epoch": 0.4915526950925181,
      "grad_norm": 0.5017377787875782,
      "learning_rate": 9.876418751123788e-06,
      "loss": 0.0396,
      "step": 1833
    },
    {
      "epoch": 0.49182086350227944,
      "grad_norm": 0.4467692897043541,
      "learning_rate": 9.876073780850776e-06,
      "loss": 0.036,
      "step": 1834
    },
    {
      "epoch": 0.4920890319120408,
      "grad_norm": 0.5069890223066578,
      "learning_rate": 9.875728335807407e-06,
      "loss": 0.0301,
      "step": 1835
    },
    {
      "epoch": 0.4923572003218021,
      "grad_norm": 0.5293469628442357,
      "learning_rate": 9.875382416027314e-06,
      "loss": 0.0345,
      "step": 1836
    },
    {
      "epoch": 0.49262536873156343,
      "grad_norm": 0.4030484546203759,
      "learning_rate": 9.87503602154418e-06,
      "loss": 0.0427,
      "step": 1837
    },
    {
      "epoch": 0.4928935371413248,
      "grad_norm": 0.38772730450355747,
      "learning_rate": 9.874689152391729e-06,
      "loss": 0.0374,
      "step": 1838
    },
    {
      "epoch": 0.49316170555108607,
      "grad_norm": 0.5485050135750426,
      "learning_rate": 9.87434180860374e-06,
      "loss": 0.0427,
      "step": 1839
    },
    {
      "epoch": 0.4934298739608474,
      "grad_norm": 0.895425894542109,
      "learning_rate": 9.873993990214028e-06,
      "loss": 0.0552,
      "step": 1840
    },
    {
      "epoch": 0.49369804237060877,
      "grad_norm": 0.4356913690541344,
      "learning_rate": 9.873645697256463e-06,
      "loss": 0.0489,
      "step": 1841
    },
    {
      "epoch": 0.49396621078037006,
      "grad_norm": 0.6255653046576253,
      "learning_rate": 9.873296929764956e-06,
      "loss": 0.0561,
      "step": 1842
    },
    {
      "epoch": 0.4942343791901314,
      "grad_norm": 0.41068129417844157,
      "learning_rate": 9.872947687773464e-06,
      "loss": 0.0514,
      "step": 1843
    },
    {
      "epoch": 0.4945025475998927,
      "grad_norm": 0.4759611275145552,
      "learning_rate": 9.872597971315994e-06,
      "loss": 0.0395,
      "step": 1844
    },
    {
      "epoch": 0.49477071600965405,
      "grad_norm": 1.8774257778098693,
      "learning_rate": 9.872247780426595e-06,
      "loss": 0.0504,
      "step": 1845
    },
    {
      "epoch": 0.4950388844194154,
      "grad_norm": 0.4792237876346564,
      "learning_rate": 9.871897115139367e-06,
      "loss": 0.0345,
      "step": 1846
    },
    {
      "epoch": 0.4953070528291767,
      "grad_norm": 0.5397751928617246,
      "learning_rate": 9.871545975488448e-06,
      "loss": 0.0406,
      "step": 1847
    },
    {
      "epoch": 0.49557522123893805,
      "grad_norm": 0.5254123324079035,
      "learning_rate": 9.871194361508034e-06,
      "loss": 0.0494,
      "step": 1848
    },
    {
      "epoch": 0.4958433896486994,
      "grad_norm": 0.5963120818612687,
      "learning_rate": 9.870842273232358e-06,
      "loss": 0.049,
      "step": 1849
    },
    {
      "epoch": 0.4961115580584607,
      "grad_norm": 0.41331531584818376,
      "learning_rate": 9.8704897106957e-06,
      "loss": 0.042,
      "step": 1850
    },
    {
      "epoch": 0.49637972646822204,
      "grad_norm": 0.42699918316620317,
      "learning_rate": 9.87013667393239e-06,
      "loss": 0.0261,
      "step": 1851
    },
    {
      "epoch": 0.4966478948779834,
      "grad_norm": 0.469610714399729,
      "learning_rate": 9.869783162976803e-06,
      "loss": 0.0445,
      "step": 1852
    },
    {
      "epoch": 0.4969160632877447,
      "grad_norm": 0.8678444719875582,
      "learning_rate": 9.869429177863358e-06,
      "loss": 0.0674,
      "step": 1853
    },
    {
      "epoch": 0.49718423169750603,
      "grad_norm": 0.4383735925094647,
      "learning_rate": 9.86907471862652e-06,
      "loss": 0.0397,
      "step": 1854
    },
    {
      "epoch": 0.4974524001072674,
      "grad_norm": 0.3869606055517455,
      "learning_rate": 9.868719785300808e-06,
      "loss": 0.0349,
      "step": 1855
    },
    {
      "epoch": 0.49772056851702867,
      "grad_norm": 0.5074712119960155,
      "learning_rate": 9.868364377920773e-06,
      "loss": 0.0426,
      "step": 1856
    },
    {
      "epoch": 0.49798873692679,
      "grad_norm": 0.3480526457978337,
      "learning_rate": 9.868008496521027e-06,
      "loss": 0.0308,
      "step": 1857
    },
    {
      "epoch": 0.49825690533655137,
      "grad_norm": 0.5862240516067672,
      "learning_rate": 9.867652141136217e-06,
      "loss": 0.0509,
      "step": 1858
    },
    {
      "epoch": 0.49852507374631266,
      "grad_norm": 0.9092458268113179,
      "learning_rate": 9.86729531180104e-06,
      "loss": 0.0491,
      "step": 1859
    },
    {
      "epoch": 0.498793242156074,
      "grad_norm": 0.5614843788500468,
      "learning_rate": 9.866938008550241e-06,
      "loss": 0.044,
      "step": 1860
    },
    {
      "epoch": 0.49906141056583536,
      "grad_norm": 0.4485914744424546,
      "learning_rate": 9.866580231418612e-06,
      "loss": 0.0425,
      "step": 1861
    },
    {
      "epoch": 0.49932957897559666,
      "grad_norm": 0.40731958733621093,
      "learning_rate": 9.866221980440983e-06,
      "loss": 0.0376,
      "step": 1862
    },
    {
      "epoch": 0.499597747385358,
      "grad_norm": 0.5380937722838296,
      "learning_rate": 9.865863255652242e-06,
      "loss": 0.0318,
      "step": 1863
    },
    {
      "epoch": 0.49986591579511935,
      "grad_norm": 0.7654845250567571,
      "learning_rate": 9.865504057087313e-06,
      "loss": 0.057,
      "step": 1864
    },
    {
      "epoch": 0.5001340842048807,
      "grad_norm": 0.5566431707799038,
      "learning_rate": 9.865144384781171e-06,
      "loss": 0.0537,
      "step": 1865
    },
    {
      "epoch": 0.500402252614642,
      "grad_norm": 0.41807511265658265,
      "learning_rate": 9.864784238768839e-06,
      "loss": 0.0365,
      "step": 1866
    },
    {
      "epoch": 0.5006704210244033,
      "grad_norm": 0.7394143471398104,
      "learning_rate": 9.86442361908538e-06,
      "loss": 0.0544,
      "step": 1867
    },
    {
      "epoch": 0.5009385894341647,
      "grad_norm": 0.31639467889448114,
      "learning_rate": 9.864062525765909e-06,
      "loss": 0.0286,
      "step": 1868
    },
    {
      "epoch": 0.501206757843926,
      "grad_norm": 0.5109002128410765,
      "learning_rate": 9.863700958845582e-06,
      "loss": 0.0518,
      "step": 1869
    },
    {
      "epoch": 0.5014749262536873,
      "grad_norm": 0.507758032128084,
      "learning_rate": 9.863338918359606e-06,
      "loss": 0.0478,
      "step": 1870
    },
    {
      "epoch": 0.5017430946634487,
      "grad_norm": 0.41086170239629327,
      "learning_rate": 9.862976404343234e-06,
      "loss": 0.0492,
      "step": 1871
    },
    {
      "epoch": 0.50201126307321,
      "grad_norm": 0.4581922378723714,
      "learning_rate": 9.862613416831757e-06,
      "loss": 0.0346,
      "step": 1872
    },
    {
      "epoch": 0.5022794314829713,
      "grad_norm": 0.464125572725527,
      "learning_rate": 9.862249955860524e-06,
      "loss": 0.0358,
      "step": 1873
    },
    {
      "epoch": 0.5025475998927327,
      "grad_norm": 0.47219795030124384,
      "learning_rate": 9.86188602146492e-06,
      "loss": 0.0522,
      "step": 1874
    },
    {
      "epoch": 0.502815768302494,
      "grad_norm": 1.0254670954831642,
      "learning_rate": 9.861521613680384e-06,
      "loss": 0.037,
      "step": 1875
    },
    {
      "epoch": 0.5030839367122553,
      "grad_norm": 0.7118083427317898,
      "learning_rate": 9.861156732542393e-06,
      "loss": 0.0509,
      "step": 1876
    },
    {
      "epoch": 0.5033521051220167,
      "grad_norm": 1.005742751152509,
      "learning_rate": 9.86079137808648e-06,
      "loss": 0.0553,
      "step": 1877
    },
    {
      "epoch": 0.503620273531778,
      "grad_norm": 0.43639645321982073,
      "learning_rate": 9.860425550348214e-06,
      "loss": 0.0322,
      "step": 1878
    },
    {
      "epoch": 0.5038884419415393,
      "grad_norm": 0.39540392914836325,
      "learning_rate": 9.860059249363217e-06,
      "loss": 0.0349,
      "step": 1879
    },
    {
      "epoch": 0.5041566103513007,
      "grad_norm": 0.30716063712663144,
      "learning_rate": 9.859692475167153e-06,
      "loss": 0.0414,
      "step": 1880
    },
    {
      "epoch": 0.504424778761062,
      "grad_norm": 0.42916293937995925,
      "learning_rate": 9.859325227795736e-06,
      "loss": 0.0463,
      "step": 1881
    },
    {
      "epoch": 0.5046929471708232,
      "grad_norm": 0.42059838880493067,
      "learning_rate": 9.858957507284723e-06,
      "loss": 0.0541,
      "step": 1882
    },
    {
      "epoch": 0.5049611155805847,
      "grad_norm": 0.6780767624438282,
      "learning_rate": 9.858589313669918e-06,
      "loss": 0.0446,
      "step": 1883
    },
    {
      "epoch": 0.505229283990346,
      "grad_norm": 0.6026634806555805,
      "learning_rate": 9.858220646987168e-06,
      "loss": 0.048,
      "step": 1884
    },
    {
      "epoch": 0.5054974524001072,
      "grad_norm": 0.5533851408987825,
      "learning_rate": 9.857851507272376e-06,
      "loss": 0.0608,
      "step": 1885
    },
    {
      "epoch": 0.5057656208098686,
      "grad_norm": 0.6812708088287186,
      "learning_rate": 9.857481894561478e-06,
      "loss": 0.0376,
      "step": 1886
    },
    {
      "epoch": 0.5060337892196299,
      "grad_norm": 0.6507469169042163,
      "learning_rate": 9.857111808890465e-06,
      "loss": 0.0385,
      "step": 1887
    },
    {
      "epoch": 0.5063019576293912,
      "grad_norm": 0.3843719663581745,
      "learning_rate": 9.85674125029537e-06,
      "loss": 0.0428,
      "step": 1888
    },
    {
      "epoch": 0.5065701260391526,
      "grad_norm": 0.7375810130340558,
      "learning_rate": 9.856370218812274e-06,
      "loss": 0.0622,
      "step": 1889
    },
    {
      "epoch": 0.5068382944489139,
      "grad_norm": 0.4450931021728746,
      "learning_rate": 9.855998714477302e-06,
      "loss": 0.0434,
      "step": 1890
    },
    {
      "epoch": 0.5071064628586752,
      "grad_norm": 0.3779130332249112,
      "learning_rate": 9.85562673732663e-06,
      "loss": 0.0384,
      "step": 1891
    },
    {
      "epoch": 0.5073746312684366,
      "grad_norm": 0.49767523085243587,
      "learning_rate": 9.855254287396473e-06,
      "loss": 0.0591,
      "step": 1892
    },
    {
      "epoch": 0.5076427996781979,
      "grad_norm": 0.3987772224031206,
      "learning_rate": 9.854881364723094e-06,
      "loss": 0.0451,
      "step": 1893
    },
    {
      "epoch": 0.5079109680879592,
      "grad_norm": 0.44421769020093776,
      "learning_rate": 9.854507969342808e-06,
      "loss": 0.0512,
      "step": 1894
    },
    {
      "epoch": 0.5081791364977206,
      "grad_norm": 0.4481980708263646,
      "learning_rate": 9.854134101291972e-06,
      "loss": 0.0516,
      "step": 1895
    },
    {
      "epoch": 0.5084473049074819,
      "grad_norm": 0.5109881101545968,
      "learning_rate": 9.853759760606982e-06,
      "loss": 0.0393,
      "step": 1896
    },
    {
      "epoch": 0.5087154733172432,
      "grad_norm": 0.3534620215630057,
      "learning_rate": 9.853384947324293e-06,
      "loss": 0.0328,
      "step": 1897
    },
    {
      "epoch": 0.5089836417270046,
      "grad_norm": 0.3811009870491632,
      "learning_rate": 9.853009661480397e-06,
      "loss": 0.0409,
      "step": 1898
    },
    {
      "epoch": 0.5092518101367659,
      "grad_norm": 0.4789581048963292,
      "learning_rate": 9.852633903111834e-06,
      "loss": 0.0441,
      "step": 1899
    },
    {
      "epoch": 0.5095199785465272,
      "grad_norm": 0.4601266127667206,
      "learning_rate": 9.852257672255193e-06,
      "loss": 0.0409,
      "step": 1900
    },
    {
      "epoch": 0.5097881469562886,
      "grad_norm": 0.5089488473649736,
      "learning_rate": 9.851880968947102e-06,
      "loss": 0.0428,
      "step": 1901
    },
    {
      "epoch": 0.5100563153660499,
      "grad_norm": 0.4878199552228014,
      "learning_rate": 9.851503793224244e-06,
      "loss": 0.033,
      "step": 1902
    },
    {
      "epoch": 0.5103244837758112,
      "grad_norm": 0.4493974613474292,
      "learning_rate": 9.851126145123341e-06,
      "loss": 0.0495,
      "step": 1903
    },
    {
      "epoch": 0.5105926521855726,
      "grad_norm": 0.3125771381799123,
      "learning_rate": 9.850748024681168e-06,
      "loss": 0.0314,
      "step": 1904
    },
    {
      "epoch": 0.5108608205953339,
      "grad_norm": 0.40382139296307484,
      "learning_rate": 9.850369431934536e-06,
      "loss": 0.0359,
      "step": 1905
    },
    {
      "epoch": 0.5111289890050952,
      "grad_norm": 0.7241205269506823,
      "learning_rate": 9.84999036692031e-06,
      "loss": 0.0526,
      "step": 1906
    },
    {
      "epoch": 0.5113971574148566,
      "grad_norm": 0.3107369658748318,
      "learning_rate": 9.849610829675398e-06,
      "loss": 0.0307,
      "step": 1907
    },
    {
      "epoch": 0.5116653258246179,
      "grad_norm": 0.33351763330345985,
      "learning_rate": 9.849230820236756e-06,
      "loss": 0.0445,
      "step": 1908
    },
    {
      "epoch": 0.5119334942343792,
      "grad_norm": 0.5749430815443008,
      "learning_rate": 9.848850338641382e-06,
      "loss": 0.0421,
      "step": 1909
    },
    {
      "epoch": 0.5122016626441405,
      "grad_norm": 0.4833725106367499,
      "learning_rate": 9.848469384926328e-06,
      "loss": 0.0456,
      "step": 1910
    },
    {
      "epoch": 0.5124698310539019,
      "grad_norm": 0.5156372110936757,
      "learning_rate": 9.848087959128679e-06,
      "loss": 0.0498,
      "step": 1911
    },
    {
      "epoch": 0.5127379994636632,
      "grad_norm": 0.6176247978298052,
      "learning_rate": 9.847706061285575e-06,
      "loss": 0.0357,
      "step": 1912
    },
    {
      "epoch": 0.5130061678734245,
      "grad_norm": 0.43171110707880106,
      "learning_rate": 9.847323691434205e-06,
      "loss": 0.0448,
      "step": 1913
    },
    {
      "epoch": 0.5132743362831859,
      "grad_norm": 0.3990706233838123,
      "learning_rate": 9.846940849611797e-06,
      "loss": 0.0366,
      "step": 1914
    },
    {
      "epoch": 0.5135425046929472,
      "grad_norm": 0.404395381243116,
      "learning_rate": 9.846557535855625e-06,
      "loss": 0.0294,
      "step": 1915
    },
    {
      "epoch": 0.5138106731027084,
      "grad_norm": 0.42657056067898025,
      "learning_rate": 9.846173750203014e-06,
      "loss": 0.027,
      "step": 1916
    },
    {
      "epoch": 0.5140788415124699,
      "grad_norm": 0.5766397342309836,
      "learning_rate": 9.845789492691332e-06,
      "loss": 0.0402,
      "step": 1917
    },
    {
      "epoch": 0.5143470099222311,
      "grad_norm": 0.3729853346069816,
      "learning_rate": 9.845404763357991e-06,
      "loss": 0.0387,
      "step": 1918
    },
    {
      "epoch": 0.5146151783319924,
      "grad_norm": 0.5106127211076743,
      "learning_rate": 9.845019562240455e-06,
      "loss": 0.0481,
      "step": 1919
    },
    {
      "epoch": 0.5148833467417538,
      "grad_norm": 0.3897415463160873,
      "learning_rate": 9.844633889376225e-06,
      "loss": 0.0368,
      "step": 1920
    },
    {
      "epoch": 0.5151515151515151,
      "grad_norm": 0.5860622588651289,
      "learning_rate": 9.844247744802857e-06,
      "loss": 0.0553,
      "step": 1921
    },
    {
      "epoch": 0.5154196835612764,
      "grad_norm": 0.37246291251504515,
      "learning_rate": 9.843861128557945e-06,
      "loss": 0.0373,
      "step": 1922
    },
    {
      "epoch": 0.5156878519710378,
      "grad_norm": 0.5745569495626512,
      "learning_rate": 9.843474040679137e-06,
      "loss": 0.036,
      "step": 1923
    },
    {
      "epoch": 0.5159560203807991,
      "grad_norm": 0.6707355482823639,
      "learning_rate": 9.84308648120412e-06,
      "loss": 0.0559,
      "step": 1924
    },
    {
      "epoch": 0.5162241887905604,
      "grad_norm": 0.5100638170327184,
      "learning_rate": 9.842698450170631e-06,
      "loss": 0.0453,
      "step": 1925
    },
    {
      "epoch": 0.5164923572003218,
      "grad_norm": 0.4907191306181842,
      "learning_rate": 9.842309947616451e-06,
      "loss": 0.0474,
      "step": 1926
    },
    {
      "epoch": 0.5167605256100831,
      "grad_norm": 0.42121510319549144,
      "learning_rate": 9.841920973579408e-06,
      "loss": 0.0482,
      "step": 1927
    },
    {
      "epoch": 0.5170286940198444,
      "grad_norm": 0.9185573422712652,
      "learning_rate": 9.841531528097374e-06,
      "loss": 0.0771,
      "step": 1928
    },
    {
      "epoch": 0.5172968624296058,
      "grad_norm": 0.5518653802424296,
      "learning_rate": 9.84114161120827e-06,
      "loss": 0.053,
      "step": 1929
    },
    {
      "epoch": 0.5175650308393671,
      "grad_norm": 0.6237041078165506,
      "learning_rate": 9.84075122295006e-06,
      "loss": 0.045,
      "step": 1930
    },
    {
      "epoch": 0.5178331992491284,
      "grad_norm": 0.35822703425892305,
      "learning_rate": 9.840360363360755e-06,
      "loss": 0.036,
      "step": 1931
    },
    {
      "epoch": 0.5181013676588898,
      "grad_norm": 0.3286410773732383,
      "learning_rate": 9.839969032478412e-06,
      "loss": 0.0385,
      "step": 1932
    },
    {
      "epoch": 0.5183695360686511,
      "grad_norm": 0.6133069238667195,
      "learning_rate": 9.839577230341135e-06,
      "loss": 0.0411,
      "step": 1933
    },
    {
      "epoch": 0.5186377044784124,
      "grad_norm": 0.6423314141769016,
      "learning_rate": 9.839184956987072e-06,
      "loss": 0.0531,
      "step": 1934
    },
    {
      "epoch": 0.5189058728881738,
      "grad_norm": 0.4309056236846783,
      "learning_rate": 9.838792212454416e-06,
      "loss": 0.043,
      "step": 1935
    },
    {
      "epoch": 0.5191740412979351,
      "grad_norm": 0.7235472835134716,
      "learning_rate": 9.838398996781411e-06,
      "loss": 0.048,
      "step": 1936
    },
    {
      "epoch": 0.5194422097076964,
      "grad_norm": 0.5042307332850442,
      "learning_rate": 9.83800531000634e-06,
      "loss": 0.0468,
      "step": 1937
    },
    {
      "epoch": 0.5197103781174578,
      "grad_norm": 0.44779758949145526,
      "learning_rate": 9.837611152167537e-06,
      "loss": 0.0455,
      "step": 1938
    },
    {
      "epoch": 0.5199785465272191,
      "grad_norm": 0.6389948644499553,
      "learning_rate": 9.83721652330338e-06,
      "loss": 0.0431,
      "step": 1939
    },
    {
      "epoch": 0.5202467149369804,
      "grad_norm": 0.46670359205564366,
      "learning_rate": 9.836821423452293e-06,
      "loss": 0.0344,
      "step": 1940
    },
    {
      "epoch": 0.5205148833467418,
      "grad_norm": 0.4901862331847499,
      "learning_rate": 9.836425852652744e-06,
      "loss": 0.0593,
      "step": 1941
    },
    {
      "epoch": 0.5207830517565031,
      "grad_norm": 0.5864491269815173,
      "learning_rate": 9.836029810943252e-06,
      "loss": 0.0706,
      "step": 1942
    },
    {
      "epoch": 0.5210512201662644,
      "grad_norm": 0.33757569663805637,
      "learning_rate": 9.835633298362377e-06,
      "loss": 0.0337,
      "step": 1943
    },
    {
      "epoch": 0.5213193885760258,
      "grad_norm": 0.43250520818770744,
      "learning_rate": 9.835236314948724e-06,
      "loss": 0.0417,
      "step": 1944
    },
    {
      "epoch": 0.5215875569857871,
      "grad_norm": 0.6846583128704155,
      "learning_rate": 9.834838860740949e-06,
      "loss": 0.0429,
      "step": 1945
    },
    {
      "epoch": 0.5218557253955484,
      "grad_norm": 0.3503436281744212,
      "learning_rate": 9.834440935777751e-06,
      "loss": 0.03,
      "step": 1946
    },
    {
      "epoch": 0.5221238938053098,
      "grad_norm": 0.41595731129623564,
      "learning_rate": 9.834042540097875e-06,
      "loss": 0.0393,
      "step": 1947
    },
    {
      "epoch": 0.5223920622150711,
      "grad_norm": 0.3496022950857249,
      "learning_rate": 9.833643673740109e-06,
      "loss": 0.0336,
      "step": 1948
    },
    {
      "epoch": 0.5226602306248324,
      "grad_norm": 0.6312545212138936,
      "learning_rate": 9.833244336743293e-06,
      "loss": 0.0391,
      "step": 1949
    },
    {
      "epoch": 0.5229283990345938,
      "grad_norm": 0.4021598407955603,
      "learning_rate": 9.832844529146307e-06,
      "loss": 0.0403,
      "step": 1950
    },
    {
      "epoch": 0.523196567444355,
      "grad_norm": 0.38071431845456905,
      "learning_rate": 9.83244425098808e-06,
      "loss": 0.0406,
      "step": 1951
    },
    {
      "epoch": 0.5234647358541163,
      "grad_norm": 0.39321990686693775,
      "learning_rate": 9.832043502307586e-06,
      "loss": 0.046,
      "step": 1952
    },
    {
      "epoch": 0.5237329042638778,
      "grad_norm": 0.6248491736936698,
      "learning_rate": 9.831642283143847e-06,
      "loss": 0.0474,
      "step": 1953
    },
    {
      "epoch": 0.524001072673639,
      "grad_norm": 0.3378895479291673,
      "learning_rate": 9.831240593535926e-06,
      "loss": 0.036,
      "step": 1954
    },
    {
      "epoch": 0.5242692410834003,
      "grad_norm": 0.9819261143089283,
      "learning_rate": 9.830838433522934e-06,
      "loss": 0.0521,
      "step": 1955
    },
    {
      "epoch": 0.5245374094931617,
      "grad_norm": 0.681714251541869,
      "learning_rate": 9.83043580314403e-06,
      "loss": 0.0693,
      "step": 1956
    },
    {
      "epoch": 0.524805577902923,
      "grad_norm": 1.5256861818451446,
      "learning_rate": 9.830032702438416e-06,
      "loss": 0.0541,
      "step": 1957
    },
    {
      "epoch": 0.5250737463126843,
      "grad_norm": 0.329938399394509,
      "learning_rate": 9.829629131445342e-06,
      "loss": 0.0281,
      "step": 1958
    },
    {
      "epoch": 0.5253419147224457,
      "grad_norm": 0.47282152029280494,
      "learning_rate": 9.829225090204102e-06,
      "loss": 0.0417,
      "step": 1959
    },
    {
      "epoch": 0.525610083132207,
      "grad_norm": 0.41137365489454303,
      "learning_rate": 9.828820578754036e-06,
      "loss": 0.0437,
      "step": 1960
    },
    {
      "epoch": 0.5258782515419683,
      "grad_norm": 0.5122239422513344,
      "learning_rate": 9.82841559713453e-06,
      "loss": 0.0597,
      "step": 1961
    },
    {
      "epoch": 0.5261464199517297,
      "grad_norm": 0.48809842965384104,
      "learning_rate": 9.828010145385017e-06,
      "loss": 0.0514,
      "step": 1962
    },
    {
      "epoch": 0.526414588361491,
      "grad_norm": 0.44217222530563616,
      "learning_rate": 9.827604223544973e-06,
      "loss": 0.046,
      "step": 1963
    },
    {
      "epoch": 0.5266827567712523,
      "grad_norm": 0.7315258258220281,
      "learning_rate": 9.827197831653926e-06,
      "loss": 0.0607,
      "step": 1964
    },
    {
      "epoch": 0.5269509251810137,
      "grad_norm": 0.3509288366635582,
      "learning_rate": 9.826790969751438e-06,
      "loss": 0.0393,
      "step": 1965
    },
    {
      "epoch": 0.527219093590775,
      "grad_norm": 0.3528880637316147,
      "learning_rate": 9.826383637877131e-06,
      "loss": 0.0317,
      "step": 1966
    },
    {
      "epoch": 0.5274872620005363,
      "grad_norm": 0.5689149692030911,
      "learning_rate": 9.825975836070662e-06,
      "loss": 0.0381,
      "step": 1967
    },
    {
      "epoch": 0.5277554304102977,
      "grad_norm": 0.6289910115824681,
      "learning_rate": 9.825567564371738e-06,
      "loss": 0.0572,
      "step": 1968
    },
    {
      "epoch": 0.528023598820059,
      "grad_norm": 0.3384676975157835,
      "learning_rate": 9.825158822820113e-06,
      "loss": 0.0307,
      "step": 1969
    },
    {
      "epoch": 0.5282917672298203,
      "grad_norm": 0.6597062114764977,
      "learning_rate": 9.824749611455583e-06,
      "loss": 0.0424,
      "step": 1970
    },
    {
      "epoch": 0.5285599356395817,
      "grad_norm": 0.4474356497119647,
      "learning_rate": 9.824339930317994e-06,
      "loss": 0.04,
      "step": 1971
    },
    {
      "epoch": 0.528828104049343,
      "grad_norm": 0.4425259584233051,
      "learning_rate": 9.823929779447235e-06,
      "loss": 0.0423,
      "step": 1972
    },
    {
      "epoch": 0.5290962724591043,
      "grad_norm": 0.39778137771121064,
      "learning_rate": 9.823519158883237e-06,
      "loss": 0.0372,
      "step": 1973
    },
    {
      "epoch": 0.5293644408688657,
      "grad_norm": 0.3458657863402165,
      "learning_rate": 9.823108068665987e-06,
      "loss": 0.0337,
      "step": 1974
    },
    {
      "epoch": 0.529632609278627,
      "grad_norm": 0.4328796269012874,
      "learning_rate": 9.822696508835511e-06,
      "loss": 0.0452,
      "step": 1975
    },
    {
      "epoch": 0.5299007776883883,
      "grad_norm": 2.090740379595086,
      "learning_rate": 9.822284479431878e-06,
      "loss": 0.0495,
      "step": 1976
    },
    {
      "epoch": 0.5301689460981497,
      "grad_norm": 0.41427382342182656,
      "learning_rate": 9.821871980495208e-06,
      "loss": 0.0465,
      "step": 1977
    },
    {
      "epoch": 0.530437114507911,
      "grad_norm": 0.6721291614225887,
      "learning_rate": 9.821459012065666e-06,
      "loss": 0.0364,
      "step": 1978
    },
    {
      "epoch": 0.5307052829176723,
      "grad_norm": 0.42888362415783904,
      "learning_rate": 9.821045574183461e-06,
      "loss": 0.0319,
      "step": 1979
    },
    {
      "epoch": 0.5309734513274337,
      "grad_norm": 0.6734723561228295,
      "learning_rate": 9.820631666888847e-06,
      "loss": 0.0495,
      "step": 1980
    },
    {
      "epoch": 0.531241619737195,
      "grad_norm": 0.3538424065302501,
      "learning_rate": 9.820217290222127e-06,
      "loss": 0.0298,
      "step": 1981
    },
    {
      "epoch": 0.5315097881469563,
      "grad_norm": 0.5062514779579452,
      "learning_rate": 9.819802444223647e-06,
      "loss": 0.039,
      "step": 1982
    },
    {
      "epoch": 0.5317779565567177,
      "grad_norm": 0.5016167979760771,
      "learning_rate": 9.819387128933799e-06,
      "loss": 0.05,
      "step": 1983
    },
    {
      "epoch": 0.532046124966479,
      "grad_norm": 0.537226155910277,
      "learning_rate": 9.81897134439302e-06,
      "loss": 0.0479,
      "step": 1984
    },
    {
      "epoch": 0.5323142933762403,
      "grad_norm": 0.5413373632027572,
      "learning_rate": 9.818555090641797e-06,
      "loss": 0.042,
      "step": 1985
    },
    {
      "epoch": 0.5325824617860017,
      "grad_norm": 0.6882519895178072,
      "learning_rate": 9.818138367720657e-06,
      "loss": 0.0511,
      "step": 1986
    },
    {
      "epoch": 0.532850630195763,
      "grad_norm": 0.5362041863974267,
      "learning_rate": 9.817721175670176e-06,
      "loss": 0.0458,
      "step": 1987
    },
    {
      "epoch": 0.5331187986055242,
      "grad_norm": 0.46323507083640103,
      "learning_rate": 9.817303514530975e-06,
      "loss": 0.0358,
      "step": 1988
    },
    {
      "epoch": 0.5333869670152857,
      "grad_norm": 0.46288558438143196,
      "learning_rate": 9.816885384343722e-06,
      "loss": 0.0412,
      "step": 1989
    },
    {
      "epoch": 0.533655135425047,
      "grad_norm": 0.4645334753230899,
      "learning_rate": 9.816466785149127e-06,
      "loss": 0.0427,
      "step": 1990
    },
    {
      "epoch": 0.5339233038348082,
      "grad_norm": 0.3971151377830419,
      "learning_rate": 9.81604771698795e-06,
      "loss": 0.0387,
      "step": 1991
    },
    {
      "epoch": 0.5341914722445696,
      "grad_norm": 0.4267761556992693,
      "learning_rate": 9.815628179900988e-06,
      "loss": 0.0416,
      "step": 1992
    },
    {
      "epoch": 0.5344596406543309,
      "grad_norm": 0.6567419634869496,
      "learning_rate": 9.8152081739291e-06,
      "loss": 0.0358,
      "step": 1993
    },
    {
      "epoch": 0.5347278090640922,
      "grad_norm": 0.45082081992977835,
      "learning_rate": 9.814787699113175e-06,
      "loss": 0.0466,
      "step": 1994
    },
    {
      "epoch": 0.5349959774738536,
      "grad_norm": 0.5310439962914222,
      "learning_rate": 9.814366755494155e-06,
      "loss": 0.0435,
      "step": 1995
    },
    {
      "epoch": 0.5352641458836149,
      "grad_norm": 0.553501840553156,
      "learning_rate": 9.813945343113026e-06,
      "loss": 0.0436,
      "step": 1996
    },
    {
      "epoch": 0.5355323142933762,
      "grad_norm": 0.4259732765278402,
      "learning_rate": 9.813523462010819e-06,
      "loss": 0.0377,
      "step": 1997
    },
    {
      "epoch": 0.5358004827031375,
      "grad_norm": 1.5511140657561304,
      "learning_rate": 9.813101112228613e-06,
      "loss": 0.0411,
      "step": 1998
    },
    {
      "epoch": 0.5360686511128989,
      "grad_norm": 0.47376425038306474,
      "learning_rate": 9.81267829380753e-06,
      "loss": 0.0416,
      "step": 1999
    },
    {
      "epoch": 0.5363368195226602,
      "grad_norm": 0.6550935872750709,
      "learning_rate": 9.812255006788738e-06,
      "loss": 0.0469,
      "step": 2000
    },
    {
      "epoch": 0.5366049879324215,
      "grad_norm": 0.6607356353045636,
      "learning_rate": 9.811831251213453e-06,
      "loss": 0.0452,
      "step": 2001
    },
    {
      "epoch": 0.5368731563421829,
      "grad_norm": 0.4446414671365038,
      "learning_rate": 9.811407027122935e-06,
      "loss": 0.0375,
      "step": 2002
    },
    {
      "epoch": 0.5371413247519442,
      "grad_norm": 0.3949614749945395,
      "learning_rate": 9.810982334558487e-06,
      "loss": 0.0397,
      "step": 2003
    },
    {
      "epoch": 0.5374094931617055,
      "grad_norm": 0.4427131013536356,
      "learning_rate": 9.810557173561464e-06,
      "loss": 0.0414,
      "step": 2004
    },
    {
      "epoch": 0.5376776615714669,
      "grad_norm": 0.35130823272872097,
      "learning_rate": 9.810131544173258e-06,
      "loss": 0.0367,
      "step": 2005
    },
    {
      "epoch": 0.5379458299812282,
      "grad_norm": 0.8107306647559662,
      "learning_rate": 9.809705446435317e-06,
      "loss": 0.061,
      "step": 2006
    },
    {
      "epoch": 0.5382139983909895,
      "grad_norm": 0.5106172543697388,
      "learning_rate": 9.809278880389126e-06,
      "loss": 0.0445,
      "step": 2007
    },
    {
      "epoch": 0.5384821668007509,
      "grad_norm": 0.5370949472531165,
      "learning_rate": 9.808851846076218e-06,
      "loss": 0.05,
      "step": 2008
    },
    {
      "epoch": 0.5387503352105122,
      "grad_norm": 0.4537942000808546,
      "learning_rate": 9.808424343538172e-06,
      "loss": 0.0471,
      "step": 2009
    },
    {
      "epoch": 0.5390185036202735,
      "grad_norm": 0.47430974170344364,
      "learning_rate": 9.807996372816614e-06,
      "loss": 0.0398,
      "step": 2010
    },
    {
      "epoch": 0.5392866720300349,
      "grad_norm": 0.3367586635058746,
      "learning_rate": 9.807567933953217e-06,
      "loss": 0.0364,
      "step": 2011
    },
    {
      "epoch": 0.5395548404397962,
      "grad_norm": 0.504305868405743,
      "learning_rate": 9.807139026989691e-06,
      "loss": 0.0465,
      "step": 2012
    },
    {
      "epoch": 0.5398230088495575,
      "grad_norm": 0.3781998561433705,
      "learning_rate": 9.806709651967802e-06,
      "loss": 0.0379,
      "step": 2013
    },
    {
      "epoch": 0.5400911772593189,
      "grad_norm": 0.45412341800651856,
      "learning_rate": 9.806279808929355e-06,
      "loss": 0.0675,
      "step": 2014
    },
    {
      "epoch": 0.5403593456690802,
      "grad_norm": 0.8238655463195337,
      "learning_rate": 9.805849497916205e-06,
      "loss": 0.0449,
      "step": 2015
    },
    {
      "epoch": 0.5406275140788415,
      "grad_norm": 0.44307730767133374,
      "learning_rate": 9.805418718970248e-06,
      "loss": 0.0473,
      "step": 2016
    },
    {
      "epoch": 0.5408956824886029,
      "grad_norm": 0.6771360137143813,
      "learning_rate": 9.804987472133427e-06,
      "loss": 0.0474,
      "step": 2017
    },
    {
      "epoch": 0.5411638508983642,
      "grad_norm": 0.4298039556547534,
      "learning_rate": 9.804555757447734e-06,
      "loss": 0.0354,
      "step": 2018
    },
    {
      "epoch": 0.5414320193081255,
      "grad_norm": 0.4922060094158956,
      "learning_rate": 9.804123574955202e-06,
      "loss": 0.048,
      "step": 2019
    },
    {
      "epoch": 0.5417001877178869,
      "grad_norm": 0.38747284837817864,
      "learning_rate": 9.803690924697913e-06,
      "loss": 0.0363,
      "step": 2020
    },
    {
      "epoch": 0.5419683561276482,
      "grad_norm": 0.5680746067080892,
      "learning_rate": 9.803257806717993e-06,
      "loss": 0.0425,
      "step": 2021
    },
    {
      "epoch": 0.5422365245374094,
      "grad_norm": 0.30652112214799315,
      "learning_rate": 9.802824221057613e-06,
      "loss": 0.03,
      "step": 2022
    },
    {
      "epoch": 0.5425046929471709,
      "grad_norm": 0.6247593189557609,
      "learning_rate": 9.802390167758987e-06,
      "loss": 0.0371,
      "step": 2023
    },
    {
      "epoch": 0.5427728613569321,
      "grad_norm": 0.39695721378562643,
      "learning_rate": 9.801955646864383e-06,
      "loss": 0.0338,
      "step": 2024
    },
    {
      "epoch": 0.5430410297666934,
      "grad_norm": 0.5010609172765217,
      "learning_rate": 9.801520658416105e-06,
      "loss": 0.0407,
      "step": 2025
    },
    {
      "epoch": 0.5433091981764548,
      "grad_norm": 0.6081915900344399,
      "learning_rate": 9.80108520245651e-06,
      "loss": 0.0433,
      "step": 2026
    },
    {
      "epoch": 0.5435773665862161,
      "grad_norm": 0.36573294103600473,
      "learning_rate": 9.800649279027994e-06,
      "loss": 0.0321,
      "step": 2027
    },
    {
      "epoch": 0.5438455349959774,
      "grad_norm": 2.2304779803443497,
      "learning_rate": 9.800212888173005e-06,
      "loss": 0.0385,
      "step": 2028
    },
    {
      "epoch": 0.5441137034057388,
      "grad_norm": 0.32192436541585506,
      "learning_rate": 9.799776029934029e-06,
      "loss": 0.0331,
      "step": 2029
    },
    {
      "epoch": 0.5443818718155001,
      "grad_norm": 0.40096091621010693,
      "learning_rate": 9.799338704353605e-06,
      "loss": 0.0418,
      "step": 2030
    },
    {
      "epoch": 0.5446500402252614,
      "grad_norm": 0.575553366921704,
      "learning_rate": 9.798900911474315e-06,
      "loss": 0.0563,
      "step": 2031
    },
    {
      "epoch": 0.5449182086350228,
      "grad_norm": 0.5158739079181425,
      "learning_rate": 9.798462651338782e-06,
      "loss": 0.0487,
      "step": 2032
    },
    {
      "epoch": 0.5451863770447841,
      "grad_norm": 0.36719436387320414,
      "learning_rate": 9.79802392398968e-06,
      "loss": 0.0301,
      "step": 2033
    },
    {
      "epoch": 0.5454545454545454,
      "grad_norm": 0.6538025558989532,
      "learning_rate": 9.797584729469728e-06,
      "loss": 0.0655,
      "step": 2034
    },
    {
      "epoch": 0.5457227138643068,
      "grad_norm": 0.8195894836188615,
      "learning_rate": 9.797145067821689e-06,
      "loss": 0.0488,
      "step": 2035
    },
    {
      "epoch": 0.5459908822740681,
      "grad_norm": 0.35510011274434927,
      "learning_rate": 9.79670493908837e-06,
      "loss": 0.0306,
      "step": 2036
    },
    {
      "epoch": 0.5462590506838294,
      "grad_norm": 0.5362784949062973,
      "learning_rate": 9.796264343312626e-06,
      "loss": 0.0528,
      "step": 2037
    },
    {
      "epoch": 0.5465272190935908,
      "grad_norm": 0.29060534024733253,
      "learning_rate": 9.795823280537358e-06,
      "loss": 0.026,
      "step": 2038
    },
    {
      "epoch": 0.5467953875033521,
      "grad_norm": 0.6519799557829945,
      "learning_rate": 9.795381750805508e-06,
      "loss": 0.0512,
      "step": 2039
    },
    {
      "epoch": 0.5470635559131134,
      "grad_norm": 0.5900074585378169,
      "learning_rate": 9.794939754160069e-06,
      "loss": 0.0586,
      "step": 2040
    },
    {
      "epoch": 0.5473317243228748,
      "grad_norm": 0.43756819270212416,
      "learning_rate": 9.794497290644076e-06,
      "loss": 0.0427,
      "step": 2041
    },
    {
      "epoch": 0.5475998927326361,
      "grad_norm": 0.4906247245972443,
      "learning_rate": 9.794054360300612e-06,
      "loss": 0.0328,
      "step": 2042
    },
    {
      "epoch": 0.5478680611423974,
      "grad_norm": 0.5929987261893424,
      "learning_rate": 9.793610963172802e-06,
      "loss": 0.0462,
      "step": 2043
    },
    {
      "epoch": 0.5481362295521588,
      "grad_norm": 0.4337172730837812,
      "learning_rate": 9.793167099303821e-06,
      "loss": 0.0406,
      "step": 2044
    },
    {
      "epoch": 0.5484043979619201,
      "grad_norm": 0.4838839310179972,
      "learning_rate": 9.792722768736885e-06,
      "loss": 0.0561,
      "step": 2045
    },
    {
      "epoch": 0.5486725663716814,
      "grad_norm": 0.4786640842564696,
      "learning_rate": 9.792277971515255e-06,
      "loss": 0.0435,
      "step": 2046
    },
    {
      "epoch": 0.5489407347814428,
      "grad_norm": 0.4166148286841388,
      "learning_rate": 9.791832707682242e-06,
      "loss": 0.0369,
      "step": 2047
    },
    {
      "epoch": 0.5492089031912041,
      "grad_norm": 0.36192826392679633,
      "learning_rate": 9.791386977281203e-06,
      "loss": 0.0493,
      "step": 2048
    },
    {
      "epoch": 0.5494770716009654,
      "grad_norm": 0.4842276034441402,
      "learning_rate": 9.790940780355534e-06,
      "loss": 0.0409,
      "step": 2049
    },
    {
      "epoch": 0.5497452400107268,
      "grad_norm": 0.4699337757958404,
      "learning_rate": 9.790494116948681e-06,
      "loss": 0.0406,
      "step": 2050
    },
    {
      "epoch": 0.5500134084204881,
      "grad_norm": 0.5737432895283681,
      "learning_rate": 9.790046987104133e-06,
      "loss": 0.0589,
      "step": 2051
    },
    {
      "epoch": 0.5502815768302494,
      "grad_norm": 0.3615656440886661,
      "learning_rate": 9.78959939086543e-06,
      "loss": 0.0359,
      "step": 2052
    },
    {
      "epoch": 0.5505497452400108,
      "grad_norm": 0.6521552816726939,
      "learning_rate": 9.789151328276147e-06,
      "loss": 0.0387,
      "step": 2053
    },
    {
      "epoch": 0.5508179136497721,
      "grad_norm": 0.3515190780840208,
      "learning_rate": 9.788702799379916e-06,
      "loss": 0.0316,
      "step": 2054
    },
    {
      "epoch": 0.5510860820595334,
      "grad_norm": 0.45202218978759595,
      "learning_rate": 9.78825380422041e-06,
      "loss": 0.0429,
      "step": 2055
    },
    {
      "epoch": 0.5513542504692948,
      "grad_norm": 0.5250935924423318,
      "learning_rate": 9.787804342841338e-06,
      "loss": 0.0565,
      "step": 2056
    },
    {
      "epoch": 0.551622418879056,
      "grad_norm": 0.47788448557594704,
      "learning_rate": 9.787354415286472e-06,
      "loss": 0.053,
      "step": 2057
    },
    {
      "epoch": 0.5518905872888173,
      "grad_norm": 0.3251582506228371,
      "learning_rate": 9.786904021599617e-06,
      "loss": 0.0301,
      "step": 2058
    },
    {
      "epoch": 0.5521587556985788,
      "grad_norm": 0.3484218443039387,
      "learning_rate": 9.786453161824625e-06,
      "loss": 0.0319,
      "step": 2059
    },
    {
      "epoch": 0.55242692410834,
      "grad_norm": 0.47340406708614097,
      "learning_rate": 9.786001836005397e-06,
      "loss": 0.0462,
      "step": 2060
    },
    {
      "epoch": 0.5526950925181013,
      "grad_norm": 0.48626099584110305,
      "learning_rate": 9.785550044185877e-06,
      "loss": 0.0433,
      "step": 2061
    },
    {
      "epoch": 0.5529632609278627,
      "grad_norm": 0.3167950381107025,
      "learning_rate": 9.785097786410055e-06,
      "loss": 0.0312,
      "step": 2062
    },
    {
      "epoch": 0.553231429337624,
      "grad_norm": 0.6022408890720156,
      "learning_rate": 9.784645062721962e-06,
      "loss": 0.0547,
      "step": 2063
    },
    {
      "epoch": 0.5534995977473853,
      "grad_norm": 0.35470841267040665,
      "learning_rate": 9.784191873165686e-06,
      "loss": 0.0367,
      "step": 2064
    },
    {
      "epoch": 0.5537677661571467,
      "grad_norm": 0.424495311964531,
      "learning_rate": 9.783738217785349e-06,
      "loss": 0.0325,
      "step": 2065
    },
    {
      "epoch": 0.554035934566908,
      "grad_norm": 0.36146023364684426,
      "learning_rate": 9.78328409662512e-06,
      "loss": 0.0409,
      "step": 2066
    },
    {
      "epoch": 0.5543041029766693,
      "grad_norm": 0.5301766890345284,
      "learning_rate": 9.78282950972922e-06,
      "loss": 0.053,
      "step": 2067
    },
    {
      "epoch": 0.5545722713864307,
      "grad_norm": 0.5017354808164451,
      "learning_rate": 9.782374457141908e-06,
      "loss": 0.0544,
      "step": 2068
    },
    {
      "epoch": 0.554840439796192,
      "grad_norm": 0.5374461758811344,
      "learning_rate": 9.781918938907494e-06,
      "loss": 0.0506,
      "step": 2069
    },
    {
      "epoch": 0.5551086082059533,
      "grad_norm": 0.43022143865597107,
      "learning_rate": 9.781462955070326e-06,
      "loss": 0.0339,
      "step": 2070
    },
    {
      "epoch": 0.5553767766157147,
      "grad_norm": 0.5169420354861711,
      "learning_rate": 9.781006505674807e-06,
      "loss": 0.0387,
      "step": 2071
    },
    {
      "epoch": 0.555644945025476,
      "grad_norm": 0.4581194811481187,
      "learning_rate": 9.780549590765375e-06,
      "loss": 0.05,
      "step": 2072
    },
    {
      "epoch": 0.5559131134352373,
      "grad_norm": 0.3565631334832981,
      "learning_rate": 9.780092210386522e-06,
      "loss": 0.0421,
      "step": 2073
    },
    {
      "epoch": 0.5561812818449987,
      "grad_norm": 0.5216370615825768,
      "learning_rate": 9.779634364582784e-06,
      "loss": 0.0518,
      "step": 2074
    },
    {
      "epoch": 0.55644945025476,
      "grad_norm": 0.3122264947393512,
      "learning_rate": 9.779176053398736e-06,
      "loss": 0.0213,
      "step": 2075
    },
    {
      "epoch": 0.5567176186645213,
      "grad_norm": 1.579424762978187,
      "learning_rate": 9.778717276879004e-06,
      "loss": 0.0754,
      "step": 2076
    },
    {
      "epoch": 0.5569857870742827,
      "grad_norm": 0.3053673373031889,
      "learning_rate": 9.778258035068258e-06,
      "loss": 0.0281,
      "step": 2077
    },
    {
      "epoch": 0.557253955484044,
      "grad_norm": 0.5593653090136709,
      "learning_rate": 9.777798328011213e-06,
      "loss": 0.0513,
      "step": 2078
    },
    {
      "epoch": 0.5575221238938053,
      "grad_norm": 0.5753492890240071,
      "learning_rate": 9.77733815575263e-06,
      "loss": 0.0358,
      "step": 2079
    },
    {
      "epoch": 0.5577902923035667,
      "grad_norm": 0.4172494707571354,
      "learning_rate": 9.776877518337313e-06,
      "loss": 0.0382,
      "step": 2080
    },
    {
      "epoch": 0.558058460713328,
      "grad_norm": 0.4039088483929679,
      "learning_rate": 9.776416415810116e-06,
      "loss": 0.0309,
      "step": 2081
    },
    {
      "epoch": 0.5583266291230893,
      "grad_norm": 0.2979210038766568,
      "learning_rate": 9.775954848215934e-06,
      "loss": 0.0284,
      "step": 2082
    },
    {
      "epoch": 0.5585947975328507,
      "grad_norm": 0.3216565089918019,
      "learning_rate": 9.775492815599708e-06,
      "loss": 0.0308,
      "step": 2083
    },
    {
      "epoch": 0.558862965942612,
      "grad_norm": 0.4367729345593559,
      "learning_rate": 9.775030318006424e-06,
      "loss": 0.0467,
      "step": 2084
    },
    {
      "epoch": 0.5591311343523733,
      "grad_norm": 0.3212750754072431,
      "learning_rate": 9.774567355481116e-06,
      "loss": 0.035,
      "step": 2085
    },
    {
      "epoch": 0.5593993027621347,
      "grad_norm": 0.42474297370472514,
      "learning_rate": 9.77410392806886e-06,
      "loss": 0.0441,
      "step": 2086
    },
    {
      "epoch": 0.559667471171896,
      "grad_norm": 0.29477129658167295,
      "learning_rate": 9.773640035814783e-06,
      "loss": 0.0216,
      "step": 2087
    },
    {
      "epoch": 0.5599356395816573,
      "grad_norm": 0.3762342046061444,
      "learning_rate": 9.773175678764047e-06,
      "loss": 0.0355,
      "step": 2088
    },
    {
      "epoch": 0.5602038079914186,
      "grad_norm": 0.43503288944396545,
      "learning_rate": 9.772710856961867e-06,
      "loss": 0.0424,
      "step": 2089
    },
    {
      "epoch": 0.56047197640118,
      "grad_norm": 0.35264929411733437,
      "learning_rate": 9.772245570453502e-06,
      "loss": 0.0417,
      "step": 2090
    },
    {
      "epoch": 0.5607401448109413,
      "grad_norm": 0.5080284736308984,
      "learning_rate": 9.771779819284257e-06,
      "loss": 0.0464,
      "step": 2091
    },
    {
      "epoch": 0.5610083132207025,
      "grad_norm": 0.39316649675191195,
      "learning_rate": 9.77131360349948e-06,
      "loss": 0.0352,
      "step": 2092
    },
    {
      "epoch": 0.561276481630464,
      "grad_norm": 0.38818583247822613,
      "learning_rate": 9.770846923144565e-06,
      "loss": 0.047,
      "step": 2093
    },
    {
      "epoch": 0.5615446500402252,
      "grad_norm": 0.5377635772988183,
      "learning_rate": 9.77037977826495e-06,
      "loss": 0.051,
      "step": 2094
    },
    {
      "epoch": 0.5618128184499865,
      "grad_norm": 0.44354737977605535,
      "learning_rate": 9.769912168906124e-06,
      "loss": 0.0484,
      "step": 2095
    },
    {
      "epoch": 0.5620809868597479,
      "grad_norm": 0.45223325705951223,
      "learning_rate": 9.769444095113611e-06,
      "loss": 0.041,
      "step": 2096
    },
    {
      "epoch": 0.5623491552695092,
      "grad_norm": 0.4838352388714414,
      "learning_rate": 9.768975556932991e-06,
      "loss": 0.0442,
      "step": 2097
    },
    {
      "epoch": 0.5626173236792705,
      "grad_norm": 0.40456575148472995,
      "learning_rate": 9.768506554409882e-06,
      "loss": 0.0362,
      "step": 2098
    },
    {
      "epoch": 0.5628854920890319,
      "grad_norm": 0.5774510742223516,
      "learning_rate": 9.76803708758995e-06,
      "loss": 0.0483,
      "step": 2099
    },
    {
      "epoch": 0.5631536604987932,
      "grad_norm": 0.4171752223131799,
      "learning_rate": 9.767567156518904e-06,
      "loss": 0.0447,
      "step": 2100
    },
    {
      "epoch": 0.5634218289085545,
      "grad_norm": 0.4476448468888343,
      "learning_rate": 9.767096761242503e-06,
      "loss": 0.0463,
      "step": 2101
    },
    {
      "epoch": 0.5636899973183159,
      "grad_norm": 0.45290939533458785,
      "learning_rate": 9.766625901806549e-06,
      "loss": 0.0389,
      "step": 2102
    },
    {
      "epoch": 0.5639581657280772,
      "grad_norm": 0.7304028395868565,
      "learning_rate": 9.766154578256883e-06,
      "loss": 0.0443,
      "step": 2103
    },
    {
      "epoch": 0.5642263341378385,
      "grad_norm": 0.4765604495993669,
      "learning_rate": 9.765682790639399e-06,
      "loss": 0.0365,
      "step": 2104
    },
    {
      "epoch": 0.5644945025475999,
      "grad_norm": 0.7336690976584431,
      "learning_rate": 9.765210539000036e-06,
      "loss": 0.0438,
      "step": 2105
    },
    {
      "epoch": 0.5647626709573612,
      "grad_norm": 0.3560323954515163,
      "learning_rate": 9.764737823384774e-06,
      "loss": 0.0414,
      "step": 2106
    },
    {
      "epoch": 0.5650308393671225,
      "grad_norm": 0.45287341106154827,
      "learning_rate": 9.76426464383964e-06,
      "loss": 0.0384,
      "step": 2107
    },
    {
      "epoch": 0.5652990077768839,
      "grad_norm": 0.41152190747814515,
      "learning_rate": 9.763791000410708e-06,
      "loss": 0.0348,
      "step": 2108
    },
    {
      "epoch": 0.5655671761866452,
      "grad_norm": 0.285392034075322,
      "learning_rate": 9.763316893144092e-06,
      "loss": 0.0278,
      "step": 2109
    },
    {
      "epoch": 0.5658353445964065,
      "grad_norm": 0.3535447222610612,
      "learning_rate": 9.762842322085958e-06,
      "loss": 0.037,
      "step": 2110
    },
    {
      "epoch": 0.5661035130061679,
      "grad_norm": 0.4524967452957079,
      "learning_rate": 9.76236728728251e-06,
      "loss": 0.0427,
      "step": 2111
    },
    {
      "epoch": 0.5663716814159292,
      "grad_norm": 0.5218421110550597,
      "learning_rate": 9.761891788780005e-06,
      "loss": 0.037,
      "step": 2112
    },
    {
      "epoch": 0.5666398498256905,
      "grad_norm": 1.0863645734291743,
      "learning_rate": 9.761415826624737e-06,
      "loss": 0.0629,
      "step": 2113
    },
    {
      "epoch": 0.5669080182354519,
      "grad_norm": 0.325169515071217,
      "learning_rate": 9.760939400863054e-06,
      "loss": 0.0349,
      "step": 2114
    },
    {
      "epoch": 0.5671761866452132,
      "grad_norm": 0.36504534353636786,
      "learning_rate": 9.76046251154134e-06,
      "loss": 0.0397,
      "step": 2115
    },
    {
      "epoch": 0.5674443550549745,
      "grad_norm": 0.2962652617410509,
      "learning_rate": 9.759985158706031e-06,
      "loss": 0.0279,
      "step": 2116
    },
    {
      "epoch": 0.5677125234647359,
      "grad_norm": 0.4082815963454779,
      "learning_rate": 9.759507342403604e-06,
      "loss": 0.0305,
      "step": 2117
    },
    {
      "epoch": 0.5679806918744972,
      "grad_norm": 0.40428969729654063,
      "learning_rate": 9.759029062680582e-06,
      "loss": 0.0462,
      "step": 2118
    },
    {
      "epoch": 0.5682488602842585,
      "grad_norm": 0.3753737286300627,
      "learning_rate": 9.758550319583537e-06,
      "loss": 0.0421,
      "step": 2119
    },
    {
      "epoch": 0.5685170286940199,
      "grad_norm": 0.43614003710445937,
      "learning_rate": 9.758071113159082e-06,
      "loss": 0.0413,
      "step": 2120
    },
    {
      "epoch": 0.5687851971037812,
      "grad_norm": 0.31396105168948363,
      "learning_rate": 9.757591443453872e-06,
      "loss": 0.0446,
      "step": 2121
    },
    {
      "epoch": 0.5690533655135425,
      "grad_norm": 0.3462105378169937,
      "learning_rate": 9.757111310514616e-06,
      "loss": 0.0395,
      "step": 2122
    },
    {
      "epoch": 0.5693215339233039,
      "grad_norm": 0.8338544591156548,
      "learning_rate": 9.756630714388062e-06,
      "loss": 0.0456,
      "step": 2123
    },
    {
      "epoch": 0.5695897023330652,
      "grad_norm": 0.5701323783020232,
      "learning_rate": 9.756149655121003e-06,
      "loss": 0.0508,
      "step": 2124
    },
    {
      "epoch": 0.5698578707428265,
      "grad_norm": 0.34827120079679913,
      "learning_rate": 9.755668132760281e-06,
      "loss": 0.0306,
      "step": 2125
    },
    {
      "epoch": 0.5701260391525879,
      "grad_norm": 0.38347429166130564,
      "learning_rate": 9.755186147352778e-06,
      "loss": 0.0263,
      "step": 2126
    },
    {
      "epoch": 0.5703942075623492,
      "grad_norm": 0.588490483151567,
      "learning_rate": 9.754703698945425e-06,
      "loss": 0.0383,
      "step": 2127
    },
    {
      "epoch": 0.5706623759721104,
      "grad_norm": 0.27296474551765554,
      "learning_rate": 9.754220787585195e-06,
      "loss": 0.0289,
      "step": 2128
    },
    {
      "epoch": 0.5709305443818719,
      "grad_norm": 0.5444461516683837,
      "learning_rate": 9.753737413319109e-06,
      "loss": 0.0527,
      "step": 2129
    },
    {
      "epoch": 0.5711987127916331,
      "grad_norm": 0.6572488267723118,
      "learning_rate": 9.753253576194234e-06,
      "loss": 0.0491,
      "step": 2130
    },
    {
      "epoch": 0.5714668812013944,
      "grad_norm": 0.48119970459292355,
      "learning_rate": 9.752769276257676e-06,
      "loss": 0.0448,
      "step": 2131
    },
    {
      "epoch": 0.5717350496111558,
      "grad_norm": 0.38797727277170296,
      "learning_rate": 9.752284513556594e-06,
      "loss": 0.0372,
      "step": 2132
    },
    {
      "epoch": 0.5720032180209171,
      "grad_norm": 0.4276980331548487,
      "learning_rate": 9.751799288138184e-06,
      "loss": 0.0454,
      "step": 2133
    },
    {
      "epoch": 0.5722713864306784,
      "grad_norm": 0.43575611990419155,
      "learning_rate": 9.751313600049695e-06,
      "loss": 0.0544,
      "step": 2134
    },
    {
      "epoch": 0.5725395548404398,
      "grad_norm": 0.491247967210608,
      "learning_rate": 9.750827449338414e-06,
      "loss": 0.038,
      "step": 2135
    },
    {
      "epoch": 0.5728077232502011,
      "grad_norm": 0.5282192663398635,
      "learning_rate": 9.750340836051677e-06,
      "loss": 0.0615,
      "step": 2136
    },
    {
      "epoch": 0.5730758916599624,
      "grad_norm": 0.37085949609241015,
      "learning_rate": 9.749853760236866e-06,
      "loss": 0.0349,
      "step": 2137
    },
    {
      "epoch": 0.5733440600697238,
      "grad_norm": 0.44860013365532597,
      "learning_rate": 9.749366221941403e-06,
      "loss": 0.0384,
      "step": 2138
    },
    {
      "epoch": 0.5736122284794851,
      "grad_norm": 0.4542119002071712,
      "learning_rate": 9.748878221212763e-06,
      "loss": 0.0377,
      "step": 2139
    },
    {
      "epoch": 0.5738803968892464,
      "grad_norm": 0.5271416867196699,
      "learning_rate": 9.748389758098457e-06,
      "loss": 0.0396,
      "step": 2140
    },
    {
      "epoch": 0.5741485652990078,
      "grad_norm": 0.34991141371662426,
      "learning_rate": 9.747900832646047e-06,
      "loss": 0.0378,
      "step": 2141
    },
    {
      "epoch": 0.5744167337087691,
      "grad_norm": 0.3447358137381132,
      "learning_rate": 9.747411444903138e-06,
      "loss": 0.0264,
      "step": 2142
    },
    {
      "epoch": 0.5746849021185304,
      "grad_norm": 0.3728515318322614,
      "learning_rate": 9.746921594917382e-06,
      "loss": 0.0332,
      "step": 2143
    },
    {
      "epoch": 0.5749530705282918,
      "grad_norm": 0.5298190664493202,
      "learning_rate": 9.746431282736472e-06,
      "loss": 0.0413,
      "step": 2144
    },
    {
      "epoch": 0.5752212389380531,
      "grad_norm": 0.44507945510516694,
      "learning_rate": 9.74594050840815e-06,
      "loss": 0.0464,
      "step": 2145
    },
    {
      "epoch": 0.5754894073478144,
      "grad_norm": 0.49002151818755324,
      "learning_rate": 9.745449271980201e-06,
      "loss": 0.0552,
      "step": 2146
    },
    {
      "epoch": 0.5757575757575758,
      "grad_norm": 0.4441600986230594,
      "learning_rate": 9.744957573500455e-06,
      "loss": 0.0506,
      "step": 2147
    },
    {
      "epoch": 0.5760257441673371,
      "grad_norm": 0.5764105257002559,
      "learning_rate": 9.744465413016788e-06,
      "loss": 0.0397,
      "step": 2148
    },
    {
      "epoch": 0.5762939125770984,
      "grad_norm": 0.2855476292579802,
      "learning_rate": 9.74397279057712e-06,
      "loss": 0.0287,
      "step": 2149
    },
    {
      "epoch": 0.5765620809868598,
      "grad_norm": 0.390194945143855,
      "learning_rate": 9.743479706229417e-06,
      "loss": 0.0394,
      "step": 2150
    },
    {
      "epoch": 0.5768302493966211,
      "grad_norm": 0.5366491903500469,
      "learning_rate": 9.742986160021688e-06,
      "loss": 0.0443,
      "step": 2151
    },
    {
      "epoch": 0.5770984178063824,
      "grad_norm": 0.3072158253636778,
      "learning_rate": 9.742492152001989e-06,
      "loss": 0.0285,
      "step": 2152
    },
    {
      "epoch": 0.5773665862161438,
      "grad_norm": 0.467524127366552,
      "learning_rate": 9.741997682218421e-06,
      "loss": 0.0446,
      "step": 2153
    },
    {
      "epoch": 0.5776347546259051,
      "grad_norm": 0.48528206664672324,
      "learning_rate": 9.741502750719127e-06,
      "loss": 0.0452,
      "step": 2154
    },
    {
      "epoch": 0.5779029230356664,
      "grad_norm": 0.3451612756623971,
      "learning_rate": 9.741007357552298e-06,
      "loss": 0.0363,
      "step": 2155
    },
    {
      "epoch": 0.5781710914454278,
      "grad_norm": 0.43844220443100984,
      "learning_rate": 9.740511502766172e-06,
      "loss": 0.0413,
      "step": 2156
    },
    {
      "epoch": 0.5784392598551891,
      "grad_norm": 0.4696624759029428,
      "learning_rate": 9.740015186409027e-06,
      "loss": 0.0368,
      "step": 2157
    },
    {
      "epoch": 0.5787074282649504,
      "grad_norm": 0.4185091162817286,
      "learning_rate": 9.739518408529184e-06,
      "loss": 0.0359,
      "step": 2158
    },
    {
      "epoch": 0.5789755966747118,
      "grad_norm": 0.3941101034875836,
      "learning_rate": 9.739021169175021e-06,
      "loss": 0.0456,
      "step": 2159
    },
    {
      "epoch": 0.5792437650844731,
      "grad_norm": 0.8054524078002959,
      "learning_rate": 9.738523468394947e-06,
      "loss": 0.0406,
      "step": 2160
    },
    {
      "epoch": 0.5795119334942344,
      "grad_norm": 0.454688565025983,
      "learning_rate": 9.738025306237424e-06,
      "loss": 0.0382,
      "step": 2161
    },
    {
      "epoch": 0.5797801019039958,
      "grad_norm": 0.4507206044020744,
      "learning_rate": 9.737526682750955e-06,
      "loss": 0.043,
      "step": 2162
    },
    {
      "epoch": 0.580048270313757,
      "grad_norm": 0.460623106564885,
      "learning_rate": 9.73702759798409e-06,
      "loss": 0.0402,
      "step": 2163
    },
    {
      "epoch": 0.5803164387235183,
      "grad_norm": 0.47627129821737607,
      "learning_rate": 9.736528051985428e-06,
      "loss": 0.0458,
      "step": 2164
    },
    {
      "epoch": 0.5805846071332798,
      "grad_norm": 0.3166286003648832,
      "learning_rate": 9.736028044803602e-06,
      "loss": 0.0348,
      "step": 2165
    },
    {
      "epoch": 0.580852775543041,
      "grad_norm": 0.37749916470278194,
      "learning_rate": 9.7355275764873e-06,
      "loss": 0.0344,
      "step": 2166
    },
    {
      "epoch": 0.5811209439528023,
      "grad_norm": 0.6153093354304093,
      "learning_rate": 9.73502664708525e-06,
      "loss": 0.0418,
      "step": 2167
    },
    {
      "epoch": 0.5813891123625637,
      "grad_norm": 0.5107566190026994,
      "learning_rate": 9.734525256646226e-06,
      "loss": 0.034,
      "step": 2168
    },
    {
      "epoch": 0.581657280772325,
      "grad_norm": 0.4121734087539575,
      "learning_rate": 9.734023405219049e-06,
      "loss": 0.032,
      "step": 2169
    },
    {
      "epoch": 0.5819254491820863,
      "grad_norm": 0.43666703629430087,
      "learning_rate": 9.73352109285258e-06,
      "loss": 0.048,
      "step": 2170
    },
    {
      "epoch": 0.5821936175918477,
      "grad_norm": 0.3134956978568538,
      "learning_rate": 9.73301831959573e-06,
      "loss": 0.0383,
      "step": 2171
    },
    {
      "epoch": 0.582461786001609,
      "grad_norm": 0.4474145598439904,
      "learning_rate": 9.732515085497454e-06,
      "loss": 0.0372,
      "step": 2172
    },
    {
      "epoch": 0.5827299544113703,
      "grad_norm": 0.5343621344166201,
      "learning_rate": 9.732011390606748e-06,
      "loss": 0.0437,
      "step": 2173
    },
    {
      "epoch": 0.5829981228211317,
      "grad_norm": 0.5079619718393555,
      "learning_rate": 9.731507234972654e-06,
      "loss": 0.0453,
      "step": 2174
    },
    {
      "epoch": 0.583266291230893,
      "grad_norm": 0.9096753109290278,
      "learning_rate": 9.731002618644265e-06,
      "loss": 0.0359,
      "step": 2175
    },
    {
      "epoch": 0.5835344596406543,
      "grad_norm": 0.4280373228754013,
      "learning_rate": 9.73049754167071e-06,
      "loss": 0.0388,
      "step": 2176
    },
    {
      "epoch": 0.5838026280504156,
      "grad_norm": 0.43719637585384824,
      "learning_rate": 9.729992004101169e-06,
      "loss": 0.033,
      "step": 2177
    },
    {
      "epoch": 0.584070796460177,
      "grad_norm": 0.48033573099564386,
      "learning_rate": 9.729486005984864e-06,
      "loss": 0.0574,
      "step": 2178
    },
    {
      "epoch": 0.5843389648699383,
      "grad_norm": 1.0212384826506733,
      "learning_rate": 9.728979547371065e-06,
      "loss": 0.0527,
      "step": 2179
    },
    {
      "epoch": 0.5846071332796996,
      "grad_norm": 1.0403255456417992,
      "learning_rate": 9.728472628309081e-06,
      "loss": 0.0477,
      "step": 2180
    },
    {
      "epoch": 0.584875301689461,
      "grad_norm": 0.5244813618126417,
      "learning_rate": 9.727965248848273e-06,
      "loss": 0.0392,
      "step": 2181
    },
    {
      "epoch": 0.5851434700992223,
      "grad_norm": 0.3572887513832535,
      "learning_rate": 9.727457409038038e-06,
      "loss": 0.0356,
      "step": 2182
    },
    {
      "epoch": 0.5854116385089836,
      "grad_norm": 0.36221084727651665,
      "learning_rate": 9.726949108927828e-06,
      "loss": 0.0334,
      "step": 2183
    },
    {
      "epoch": 0.585679806918745,
      "grad_norm": 0.4966681747415277,
      "learning_rate": 9.726440348567133e-06,
      "loss": 0.04,
      "step": 2184
    },
    {
      "epoch": 0.5859479753285063,
      "grad_norm": 0.44885949706985573,
      "learning_rate": 9.72593112800549e-06,
      "loss": 0.0449,
      "step": 2185
    },
    {
      "epoch": 0.5862161437382676,
      "grad_norm": 0.3802783362425603,
      "learning_rate": 9.725421447292481e-06,
      "loss": 0.0354,
      "step": 2186
    },
    {
      "epoch": 0.586484312148029,
      "grad_norm": 0.41588836895012404,
      "learning_rate": 9.724911306477729e-06,
      "loss": 0.0363,
      "step": 2187
    },
    {
      "epoch": 0.5867524805577903,
      "grad_norm": 0.6468364941165152,
      "learning_rate": 9.72440070561091e-06,
      "loss": 0.05,
      "step": 2188
    },
    {
      "epoch": 0.5870206489675516,
      "grad_norm": 0.8598087938002239,
      "learning_rate": 9.723889644741735e-06,
      "loss": 0.0635,
      "step": 2189
    },
    {
      "epoch": 0.587288817377313,
      "grad_norm": 0.3545117807646672,
      "learning_rate": 9.723378123919969e-06,
      "loss": 0.0436,
      "step": 2190
    },
    {
      "epoch": 0.5875569857870743,
      "grad_norm": 0.47836329028824426,
      "learning_rate": 9.722866143195412e-06,
      "loss": 0.0413,
      "step": 2191
    },
    {
      "epoch": 0.5878251541968356,
      "grad_norm": 0.34704835376357335,
      "learning_rate": 9.722353702617921e-06,
      "loss": 0.036,
      "step": 2192
    },
    {
      "epoch": 0.588093322606597,
      "grad_norm": 0.4768755230370726,
      "learning_rate": 9.721840802237388e-06,
      "loss": 0.0395,
      "step": 2193
    },
    {
      "epoch": 0.5883614910163583,
      "grad_norm": 0.5389125022348678,
      "learning_rate": 9.72132744210375e-06,
      "loss": 0.0437,
      "step": 2194
    },
    {
      "epoch": 0.5886296594261196,
      "grad_norm": 0.41198521475630223,
      "learning_rate": 9.720813622266996e-06,
      "loss": 0.0485,
      "step": 2195
    },
    {
      "epoch": 0.588897827835881,
      "grad_norm": 0.470765776709888,
      "learning_rate": 9.72029934277715e-06,
      "loss": 0.0434,
      "step": 2196
    },
    {
      "epoch": 0.5891659962456423,
      "grad_norm": 0.5384187008665143,
      "learning_rate": 9.719784603684292e-06,
      "loss": 0.0446,
      "step": 2197
    },
    {
      "epoch": 0.5894341646554035,
      "grad_norm": 0.3969894997788225,
      "learning_rate": 9.719269405038537e-06,
      "loss": 0.033,
      "step": 2198
    },
    {
      "epoch": 0.589702333065165,
      "grad_norm": 0.5838373076782158,
      "learning_rate": 9.71875374689005e-06,
      "loss": 0.047,
      "step": 2199
    },
    {
      "epoch": 0.5899705014749262,
      "grad_norm": 0.4063369736458305,
      "learning_rate": 9.718237629289038e-06,
      "loss": 0.0401,
      "step": 2200
    },
    {
      "epoch": 0.5902386698846875,
      "grad_norm": 0.41504110835775054,
      "learning_rate": 9.717721052285754e-06,
      "loss": 0.0413,
      "step": 2201
    },
    {
      "epoch": 0.5905068382944489,
      "grad_norm": 0.39772463748917164,
      "learning_rate": 9.717204015930498e-06,
      "loss": 0.0427,
      "step": 2202
    },
    {
      "epoch": 0.5907750067042102,
      "grad_norm": 0.3346179999534236,
      "learning_rate": 9.716686520273611e-06,
      "loss": 0.0404,
      "step": 2203
    },
    {
      "epoch": 0.5910431751139715,
      "grad_norm": 0.4103858188853257,
      "learning_rate": 9.716168565365477e-06,
      "loss": 0.0437,
      "step": 2204
    },
    {
      "epoch": 0.5913113435237329,
      "grad_norm": 0.3527285845368918,
      "learning_rate": 9.715650151256534e-06,
      "loss": 0.0286,
      "step": 2205
    },
    {
      "epoch": 0.5915795119334942,
      "grad_norm": 0.3807566350942936,
      "learning_rate": 9.715131277997256e-06,
      "loss": 0.0359,
      "step": 2206
    },
    {
      "epoch": 0.5918476803432555,
      "grad_norm": 0.4000130132722489,
      "learning_rate": 9.714611945638162e-06,
      "loss": 0.0436,
      "step": 2207
    },
    {
      "epoch": 0.5921158487530169,
      "grad_norm": 0.47314414916876274,
      "learning_rate": 9.714092154229821e-06,
      "loss": 0.0379,
      "step": 2208
    },
    {
      "epoch": 0.5923840171627782,
      "grad_norm": 0.24046492937674482,
      "learning_rate": 9.713571903822842e-06,
      "loss": 0.0265,
      "step": 2209
    },
    {
      "epoch": 0.5926521855725395,
      "grad_norm": 0.6198286297498614,
      "learning_rate": 9.713051194467882e-06,
      "loss": 0.035,
      "step": 2210
    },
    {
      "epoch": 0.5929203539823009,
      "grad_norm": 0.4993532950411199,
      "learning_rate": 9.71253002621564e-06,
      "loss": 0.0432,
      "step": 2211
    },
    {
      "epoch": 0.5931885223920622,
      "grad_norm": 0.3145678969510866,
      "learning_rate": 9.712008399116861e-06,
      "loss": 0.032,
      "step": 2212
    },
    {
      "epoch": 0.5934566908018235,
      "grad_norm": 0.32512092579945895,
      "learning_rate": 9.711486313222332e-06,
      "loss": 0.0265,
      "step": 2213
    },
    {
      "epoch": 0.5937248592115849,
      "grad_norm": 0.4596823544846739,
      "learning_rate": 9.710963768582891e-06,
      "loss": 0.0392,
      "step": 2214
    },
    {
      "epoch": 0.5939930276213462,
      "grad_norm": 0.3044444403104068,
      "learning_rate": 9.710440765249416e-06,
      "loss": 0.0245,
      "step": 2215
    },
    {
      "epoch": 0.5942611960311075,
      "grad_norm": 0.5105664294956231,
      "learning_rate": 9.709917303272828e-06,
      "loss": 0.0382,
      "step": 2216
    },
    {
      "epoch": 0.5945293644408689,
      "grad_norm": 0.3442700076277357,
      "learning_rate": 9.709393382704098e-06,
      "loss": 0.0358,
      "step": 2217
    },
    {
      "epoch": 0.5947975328506302,
      "grad_norm": 0.6048828723248941,
      "learning_rate": 9.708869003594238e-06,
      "loss": 0.0563,
      "step": 2218
    },
    {
      "epoch": 0.5950657012603915,
      "grad_norm": 0.4216081745123837,
      "learning_rate": 9.708344165994304e-06,
      "loss": 0.0477,
      "step": 2219
    },
    {
      "epoch": 0.5953338696701529,
      "grad_norm": 0.29586953307035435,
      "learning_rate": 9.707818869955398e-06,
      "loss": 0.0347,
      "step": 2220
    },
    {
      "epoch": 0.5956020380799142,
      "grad_norm": 0.469085796136734,
      "learning_rate": 9.70729311552867e-06,
      "loss": 0.0475,
      "step": 2221
    },
    {
      "epoch": 0.5958702064896755,
      "grad_norm": 0.34934826031390986,
      "learning_rate": 9.706766902765308e-06,
      "loss": 0.0337,
      "step": 2222
    },
    {
      "epoch": 0.5961383748994369,
      "grad_norm": 0.4507792359312918,
      "learning_rate": 9.706240231716549e-06,
      "loss": 0.0386,
      "step": 2223
    },
    {
      "epoch": 0.5964065433091982,
      "grad_norm": 0.43696323476873283,
      "learning_rate": 9.705713102433673e-06,
      "loss": 0.0421,
      "step": 2224
    },
    {
      "epoch": 0.5966747117189595,
      "grad_norm": 0.4822237783343997,
      "learning_rate": 9.705185514968006e-06,
      "loss": 0.0563,
      "step": 2225
    },
    {
      "epoch": 0.5969428801287209,
      "grad_norm": 0.3733374702583346,
      "learning_rate": 9.704657469370917e-06,
      "loss": 0.0356,
      "step": 2226
    },
    {
      "epoch": 0.5972110485384822,
      "grad_norm": 0.39314583754393495,
      "learning_rate": 9.70412896569382e-06,
      "loss": 0.0365,
      "step": 2227
    },
    {
      "epoch": 0.5974792169482435,
      "grad_norm": 0.38806069218131134,
      "learning_rate": 9.703600003988176e-06,
      "loss": 0.0338,
      "step": 2228
    },
    {
      "epoch": 0.5977473853580049,
      "grad_norm": 0.3754476111476056,
      "learning_rate": 9.703070584305487e-06,
      "loss": 0.0302,
      "step": 2229
    },
    {
      "epoch": 0.5980155537677662,
      "grad_norm": 0.5970809307581462,
      "learning_rate": 9.702540706697303e-06,
      "loss": 0.0601,
      "step": 2230
    },
    {
      "epoch": 0.5982837221775275,
      "grad_norm": 0.6346509039914741,
      "learning_rate": 9.702010371215213e-06,
      "loss": 0.0335,
      "step": 2231
    },
    {
      "epoch": 0.5985518905872889,
      "grad_norm": 0.4036233679298006,
      "learning_rate": 9.701479577910859e-06,
      "loss": 0.0477,
      "step": 2232
    },
    {
      "epoch": 0.5988200589970502,
      "grad_norm": 0.4439968651954235,
      "learning_rate": 9.70094832683592e-06,
      "loss": 0.0407,
      "step": 2233
    },
    {
      "epoch": 0.5990882274068114,
      "grad_norm": 0.37095619925690465,
      "learning_rate": 9.700416618042123e-06,
      "loss": 0.0412,
      "step": 2234
    },
    {
      "epoch": 0.5993563958165729,
      "grad_norm": 0.41861742846672156,
      "learning_rate": 9.699884451581238e-06,
      "loss": 0.0314,
      "step": 2235
    },
    {
      "epoch": 0.5996245642263341,
      "grad_norm": 0.44463773300016574,
      "learning_rate": 9.699351827505084e-06,
      "loss": 0.0456,
      "step": 2236
    },
    {
      "epoch": 0.5998927326360954,
      "grad_norm": 0.3693821852752085,
      "learning_rate": 9.698818745865518e-06,
      "loss": 0.0296,
      "step": 2237
    },
    {
      "epoch": 0.6001609010458568,
      "grad_norm": 0.28920306561067777,
      "learning_rate": 9.698285206714446e-06,
      "loss": 0.0279,
      "step": 2238
    },
    {
      "epoch": 0.6004290694556181,
      "grad_norm": 0.5730761223565658,
      "learning_rate": 9.697751210103817e-06,
      "loss": 0.0418,
      "step": 2239
    },
    {
      "epoch": 0.6006972378653794,
      "grad_norm": 0.42848234973724597,
      "learning_rate": 9.697216756085625e-06,
      "loss": 0.0361,
      "step": 2240
    },
    {
      "epoch": 0.6009654062751408,
      "grad_norm": 0.3231561664789962,
      "learning_rate": 9.696681844711909e-06,
      "loss": 0.0304,
      "step": 2241
    },
    {
      "epoch": 0.6012335746849021,
      "grad_norm": 0.5239270405028139,
      "learning_rate": 9.696146476034752e-06,
      "loss": 0.035,
      "step": 2242
    },
    {
      "epoch": 0.6015017430946634,
      "grad_norm": 0.7148736205821443,
      "learning_rate": 9.695610650106278e-06,
      "loss": 0.0666,
      "step": 2243
    },
    {
      "epoch": 0.6017699115044248,
      "grad_norm": 0.2620489523356466,
      "learning_rate": 9.695074366978663e-06,
      "loss": 0.0269,
      "step": 2244
    },
    {
      "epoch": 0.6020380799141861,
      "grad_norm": 0.43710790784223125,
      "learning_rate": 9.69453762670412e-06,
      "loss": 0.0477,
      "step": 2245
    },
    {
      "epoch": 0.6023062483239474,
      "grad_norm": 0.4288880193649563,
      "learning_rate": 9.694000429334915e-06,
      "loss": 0.0426,
      "step": 2246
    },
    {
      "epoch": 0.6025744167337088,
      "grad_norm": 0.5896139340292553,
      "learning_rate": 9.693462774923351e-06,
      "loss": 0.0476,
      "step": 2247
    },
    {
      "epoch": 0.6028425851434701,
      "grad_norm": 0.3110674263352271,
      "learning_rate": 9.692924663521776e-06,
      "loss": 0.0339,
      "step": 2248
    },
    {
      "epoch": 0.6031107535532314,
      "grad_norm": 0.5414130537588965,
      "learning_rate": 9.692386095182585e-06,
      "loss": 0.0414,
      "step": 2249
    },
    {
      "epoch": 0.6033789219629928,
      "grad_norm": 1.1355091141009677,
      "learning_rate": 9.691847069958221e-06,
      "loss": 0.0366,
      "step": 2250
    },
    {
      "epoch": 0.6036470903727541,
      "grad_norm": 0.5677678615109082,
      "learning_rate": 9.691307587901161e-06,
      "loss": 0.0361,
      "step": 2251
    },
    {
      "epoch": 0.6039152587825154,
      "grad_norm": 0.3910756127631531,
      "learning_rate": 9.690767649063938e-06,
      "loss": 0.0437,
      "step": 2252
    },
    {
      "epoch": 0.6041834271922768,
      "grad_norm": 0.3594163077118075,
      "learning_rate": 9.690227253499122e-06,
      "loss": 0.0333,
      "step": 2253
    },
    {
      "epoch": 0.6044515956020381,
      "grad_norm": 0.3372276024627769,
      "learning_rate": 9.689686401259331e-06,
      "loss": 0.0325,
      "step": 2254
    },
    {
      "epoch": 0.6047197640117994,
      "grad_norm": 0.2571848951828705,
      "learning_rate": 9.689145092397227e-06,
      "loss": 0.0268,
      "step": 2255
    },
    {
      "epoch": 0.6049879324215608,
      "grad_norm": 0.5801177810993481,
      "learning_rate": 9.688603326965513e-06,
      "loss": 0.0489,
      "step": 2256
    },
    {
      "epoch": 0.6052561008313221,
      "grad_norm": 0.3058781308045926,
      "learning_rate": 9.688061105016942e-06,
      "loss": 0.0288,
      "step": 2257
    },
    {
      "epoch": 0.6055242692410834,
      "grad_norm": 0.4697661910113497,
      "learning_rate": 9.687518426604308e-06,
      "loss": 0.0331,
      "step": 2258
    },
    {
      "epoch": 0.6057924376508448,
      "grad_norm": 0.5441314074030762,
      "learning_rate": 9.686975291780449e-06,
      "loss": 0.0503,
      "step": 2259
    },
    {
      "epoch": 0.6060606060606061,
      "grad_norm": 0.5737967022001436,
      "learning_rate": 9.686431700598251e-06,
      "loss": 0.0492,
      "step": 2260
    },
    {
      "epoch": 0.6063287744703674,
      "grad_norm": 0.6483603804329836,
      "learning_rate": 9.68588765311064e-06,
      "loss": 0.0541,
      "step": 2261
    },
    {
      "epoch": 0.6065969428801288,
      "grad_norm": 0.3304615518027629,
      "learning_rate": 9.68534314937059e-06,
      "loss": 0.0332,
      "step": 2262
    },
    {
      "epoch": 0.6068651112898901,
      "grad_norm": 0.42468857667842247,
      "learning_rate": 9.684798189431114e-06,
      "loss": 0.0449,
      "step": 2263
    },
    {
      "epoch": 0.6071332796996514,
      "grad_norm": 0.4399345744885095,
      "learning_rate": 9.684252773345279e-06,
      "loss": 0.0451,
      "step": 2264
    },
    {
      "epoch": 0.6074014481094127,
      "grad_norm": 1.1710580615219952,
      "learning_rate": 9.683706901166186e-06,
      "loss": 0.0482,
      "step": 2265
    },
    {
      "epoch": 0.6076696165191741,
      "grad_norm": 0.5256092821455982,
      "learning_rate": 9.683160572946988e-06,
      "loss": 0.0628,
      "step": 2266
    },
    {
      "epoch": 0.6079377849289354,
      "grad_norm": 0.42811551374286155,
      "learning_rate": 9.682613788740878e-06,
      "loss": 0.0444,
      "step": 2267
    },
    {
      "epoch": 0.6082059533386966,
      "grad_norm": 0.5378938284618875,
      "learning_rate": 9.682066548601097e-06,
      "loss": 0.046,
      "step": 2268
    },
    {
      "epoch": 0.608474121748458,
      "grad_norm": 0.3245020121384951,
      "learning_rate": 9.681518852580925e-06,
      "loss": 0.033,
      "step": 2269
    },
    {
      "epoch": 0.6087422901582193,
      "grad_norm": 0.4649481392805461,
      "learning_rate": 9.680970700733692e-06,
      "loss": 0.0374,
      "step": 2270
    },
    {
      "epoch": 0.6090104585679806,
      "grad_norm": 0.6778754461735137,
      "learning_rate": 9.68042209311277e-06,
      "loss": 0.0403,
      "step": 2271
    },
    {
      "epoch": 0.609278626977742,
      "grad_norm": 0.46378905290317013,
      "learning_rate": 9.679873029771575e-06,
      "loss": 0.0403,
      "step": 2272
    },
    {
      "epoch": 0.6095467953875033,
      "grad_norm": 0.4552230860160537,
      "learning_rate": 9.679323510763568e-06,
      "loss": 0.0419,
      "step": 2273
    },
    {
      "epoch": 0.6098149637972646,
      "grad_norm": 0.5471325484675843,
      "learning_rate": 9.678773536142254e-06,
      "loss": 0.047,
      "step": 2274
    },
    {
      "epoch": 0.610083132207026,
      "grad_norm": 0.5233573422025999,
      "learning_rate": 9.678223105961182e-06,
      "loss": 0.0599,
      "step": 2275
    },
    {
      "epoch": 0.6103513006167873,
      "grad_norm": 0.2911582456311716,
      "learning_rate": 9.677672220273948e-06,
      "loss": 0.0227,
      "step": 2276
    },
    {
      "epoch": 0.6106194690265486,
      "grad_norm": 0.3375568328231506,
      "learning_rate": 9.677120879134187e-06,
      "loss": 0.0454,
      "step": 2277
    },
    {
      "epoch": 0.61088763743631,
      "grad_norm": 0.4487487839547596,
      "learning_rate": 9.676569082595585e-06,
      "loss": 0.0361,
      "step": 2278
    },
    {
      "epoch": 0.6111558058460713,
      "grad_norm": 0.6009020436948649,
      "learning_rate": 9.676016830711868e-06,
      "loss": 0.0529,
      "step": 2279
    },
    {
      "epoch": 0.6114239742558326,
      "grad_norm": 0.5289667287982411,
      "learning_rate": 9.675464123536803e-06,
      "loss": 0.0477,
      "step": 2280
    },
    {
      "epoch": 0.611692142665594,
      "grad_norm": 0.46202018559487407,
      "learning_rate": 9.674910961124215e-06,
      "loss": 0.0454,
      "step": 2281
    },
    {
      "epoch": 0.6119603110753553,
      "grad_norm": 0.408030503770942,
      "learning_rate": 9.674357343527955e-06,
      "loss": 0.0394,
      "step": 2282
    },
    {
      "epoch": 0.6122284794851166,
      "grad_norm": 0.478802591725159,
      "learning_rate": 9.67380327080193e-06,
      "loss": 0.0398,
      "step": 2283
    },
    {
      "epoch": 0.612496647894878,
      "grad_norm": 0.5665621214222397,
      "learning_rate": 9.673248743000092e-06,
      "loss": 0.0562,
      "step": 2284
    },
    {
      "epoch": 0.6127648163046393,
      "grad_norm": 0.3295307071238235,
      "learning_rate": 9.67269376017643e-06,
      "loss": 0.0439,
      "step": 2285
    },
    {
      "epoch": 0.6130329847144006,
      "grad_norm": 0.5155569113910213,
      "learning_rate": 9.67213832238498e-06,
      "loss": 0.0353,
      "step": 2286
    },
    {
      "epoch": 0.613301153124162,
      "grad_norm": 0.495477585095254,
      "learning_rate": 9.671582429679829e-06,
      "loss": 0.0363,
      "step": 2287
    },
    {
      "epoch": 0.6135693215339233,
      "grad_norm": 0.41075967198482366,
      "learning_rate": 9.671026082115098e-06,
      "loss": 0.0335,
      "step": 2288
    },
    {
      "epoch": 0.6138374899436846,
      "grad_norm": 0.3036951598571639,
      "learning_rate": 9.670469279744959e-06,
      "loss": 0.0273,
      "step": 2289
    },
    {
      "epoch": 0.614105658353446,
      "grad_norm": 0.4304101878353346,
      "learning_rate": 9.669912022623626e-06,
      "loss": 0.0332,
      "step": 2290
    },
    {
      "epoch": 0.6143738267632073,
      "grad_norm": 0.47724570133383293,
      "learning_rate": 9.669354310805359e-06,
      "loss": 0.0372,
      "step": 2291
    },
    {
      "epoch": 0.6146419951729686,
      "grad_norm": 0.501098182017171,
      "learning_rate": 9.668796144344459e-06,
      "loss": 0.0332,
      "step": 2292
    },
    {
      "epoch": 0.61491016358273,
      "grad_norm": 0.4038034464914963,
      "learning_rate": 9.668237523295272e-06,
      "loss": 0.0374,
      "step": 2293
    },
    {
      "epoch": 0.6151783319924913,
      "grad_norm": 0.9153982800595213,
      "learning_rate": 9.667678447712194e-06,
      "loss": 0.0453,
      "step": 2294
    },
    {
      "epoch": 0.6154465004022526,
      "grad_norm": 0.5609581816682323,
      "learning_rate": 9.667118917649656e-06,
      "loss": 0.0532,
      "step": 2295
    },
    {
      "epoch": 0.615714668812014,
      "grad_norm": 0.3613140997722086,
      "learning_rate": 9.66655893316214e-06,
      "loss": 0.025,
      "step": 2296
    },
    {
      "epoch": 0.6159828372217753,
      "grad_norm": 0.5666033023724834,
      "learning_rate": 9.665998494304172e-06,
      "loss": 0.0434,
      "step": 2297
    },
    {
      "epoch": 0.6162510056315366,
      "grad_norm": 0.7711228743063872,
      "learning_rate": 9.665437601130318e-06,
      "loss": 0.0433,
      "step": 2298
    },
    {
      "epoch": 0.616519174041298,
      "grad_norm": 0.6214107150343847,
      "learning_rate": 9.66487625369519e-06,
      "loss": 0.0393,
      "step": 2299
    },
    {
      "epoch": 0.6167873424510593,
      "grad_norm": 0.5390563830457505,
      "learning_rate": 9.664314452053447e-06,
      "loss": 0.062,
      "step": 2300
    },
    {
      "epoch": 0.6170555108608206,
      "grad_norm": 0.3856033220879983,
      "learning_rate": 9.66375219625979e-06,
      "loss": 0.0391,
      "step": 2301
    },
    {
      "epoch": 0.617323679270582,
      "grad_norm": 0.5878638617531717,
      "learning_rate": 9.663189486368962e-06,
      "loss": 0.0601,
      "step": 2302
    },
    {
      "epoch": 0.6175918476803433,
      "grad_norm": 0.4563286618497218,
      "learning_rate": 9.662626322435755e-06,
      "loss": 0.0331,
      "step": 2303
    },
    {
      "epoch": 0.6178600160901045,
      "grad_norm": 0.7963332856275146,
      "learning_rate": 9.662062704515003e-06,
      "loss": 0.0453,
      "step": 2304
    },
    {
      "epoch": 0.618128184499866,
      "grad_norm": 0.3389645185938771,
      "learning_rate": 9.661498632661582e-06,
      "loss": 0.0279,
      "step": 2305
    },
    {
      "epoch": 0.6183963529096272,
      "grad_norm": 0.4686813651433305,
      "learning_rate": 9.660934106930416e-06,
      "loss": 0.0473,
      "step": 2306
    },
    {
      "epoch": 0.6186645213193885,
      "grad_norm": 0.46239118270460333,
      "learning_rate": 9.660369127376469e-06,
      "loss": 0.0418,
      "step": 2307
    },
    {
      "epoch": 0.6189326897291499,
      "grad_norm": 0.3264683836244514,
      "learning_rate": 9.659803694054753e-06,
      "loss": 0.0394,
      "step": 2308
    },
    {
      "epoch": 0.6192008581389112,
      "grad_norm": 0.43244190268315214,
      "learning_rate": 9.659237807020326e-06,
      "loss": 0.039,
      "step": 2309
    },
    {
      "epoch": 0.6194690265486725,
      "grad_norm": 0.47067277918301836,
      "learning_rate": 9.65867146632828e-06,
      "loss": 0.0468,
      "step": 2310
    },
    {
      "epoch": 0.6197371949584339,
      "grad_norm": 0.5989890643605386,
      "learning_rate": 9.658104672033763e-06,
      "loss": 0.0425,
      "step": 2311
    },
    {
      "epoch": 0.6200053633681952,
      "grad_norm": 0.293211081035094,
      "learning_rate": 9.657537424191964e-06,
      "loss": 0.0327,
      "step": 2312
    },
    {
      "epoch": 0.6202735317779565,
      "grad_norm": 0.4091336108375734,
      "learning_rate": 9.656969722858108e-06,
      "loss": 0.0367,
      "step": 2313
    },
    {
      "epoch": 0.6205417001877179,
      "grad_norm": 0.46586720171971074,
      "learning_rate": 9.656401568087475e-06,
      "loss": 0.0461,
      "step": 2314
    },
    {
      "epoch": 0.6208098685974792,
      "grad_norm": 0.49094714545753565,
      "learning_rate": 9.655832959935382e-06,
      "loss": 0.0485,
      "step": 2315
    },
    {
      "epoch": 0.6210780370072405,
      "grad_norm": 0.43078934717974526,
      "learning_rate": 9.6552638984572e-06,
      "loss": 0.0398,
      "step": 2316
    },
    {
      "epoch": 0.6213462054170019,
      "grad_norm": 0.46902333369633836,
      "learning_rate": 9.654694383708327e-06,
      "loss": 0.0559,
      "step": 2317
    },
    {
      "epoch": 0.6216143738267632,
      "grad_norm": 0.3883938870513487,
      "learning_rate": 9.654124415744222e-06,
      "loss": 0.0353,
      "step": 2318
    },
    {
      "epoch": 0.6218825422365245,
      "grad_norm": 0.5478775503864985,
      "learning_rate": 9.653553994620378e-06,
      "loss": 0.0468,
      "step": 2319
    },
    {
      "epoch": 0.6221507106462859,
      "grad_norm": 0.44792210367963803,
      "learning_rate": 9.652983120392336e-06,
      "loss": 0.035,
      "step": 2320
    },
    {
      "epoch": 0.6224188790560472,
      "grad_norm": 0.5006488537645952,
      "learning_rate": 9.652411793115681e-06,
      "loss": 0.0448,
      "step": 2321
    },
    {
      "epoch": 0.6226870474658085,
      "grad_norm": 0.3595507458608492,
      "learning_rate": 9.651840012846043e-06,
      "loss": 0.0322,
      "step": 2322
    },
    {
      "epoch": 0.6229552158755699,
      "grad_norm": 0.30493458510324334,
      "learning_rate": 9.651267779639093e-06,
      "loss": 0.0295,
      "step": 2323
    },
    {
      "epoch": 0.6232233842853312,
      "grad_norm": 0.2718145354244287,
      "learning_rate": 9.650695093550549e-06,
      "loss": 0.0322,
      "step": 2324
    },
    {
      "epoch": 0.6234915526950925,
      "grad_norm": 0.4115545924262125,
      "learning_rate": 9.650121954636169e-06,
      "loss": 0.0376,
      "step": 2325
    },
    {
      "epoch": 0.6237597211048539,
      "grad_norm": 0.48162397063309037,
      "learning_rate": 9.649548362951762e-06,
      "loss": 0.0389,
      "step": 2326
    },
    {
      "epoch": 0.6240278895146152,
      "grad_norm": 0.2898629326587395,
      "learning_rate": 9.648974318553173e-06,
      "loss": 0.0256,
      "step": 2327
    },
    {
      "epoch": 0.6242960579243765,
      "grad_norm": 0.3617439375559836,
      "learning_rate": 9.6483998214963e-06,
      "loss": 0.0338,
      "step": 2328
    },
    {
      "epoch": 0.6245642263341379,
      "grad_norm": 0.3574358550245054,
      "learning_rate": 9.647824871837076e-06,
      "loss": 0.0385,
      "step": 2329
    },
    {
      "epoch": 0.6248323947438992,
      "grad_norm": 0.38540020261967123,
      "learning_rate": 9.647249469631484e-06,
      "loss": 0.0456,
      "step": 2330
    },
    {
      "epoch": 0.6251005631536605,
      "grad_norm": 0.3896947160111264,
      "learning_rate": 9.64667361493555e-06,
      "loss": 0.0423,
      "step": 2331
    },
    {
      "epoch": 0.6253687315634219,
      "grad_norm": 0.31638908917071545,
      "learning_rate": 9.646097307805341e-06,
      "loss": 0.0307,
      "step": 2332
    },
    {
      "epoch": 0.6256368999731832,
      "grad_norm": 0.6441690799074042,
      "learning_rate": 9.645520548296975e-06,
      "loss": 0.043,
      "step": 2333
    },
    {
      "epoch": 0.6259050683829445,
      "grad_norm": 0.47773262502237623,
      "learning_rate": 9.644943336466605e-06,
      "loss": 0.0502,
      "step": 2334
    },
    {
      "epoch": 0.6261732367927059,
      "grad_norm": 0.42448614977277965,
      "learning_rate": 9.644365672370437e-06,
      "loss": 0.0481,
      "step": 2335
    },
    {
      "epoch": 0.6264414052024672,
      "grad_norm": 0.3902042113230199,
      "learning_rate": 9.643787556064713e-06,
      "loss": 0.0392,
      "step": 2336
    },
    {
      "epoch": 0.6267095736122285,
      "grad_norm": 0.4046381474320068,
      "learning_rate": 9.643208987605722e-06,
      "loss": 0.0364,
      "step": 2337
    },
    {
      "epoch": 0.6269777420219899,
      "grad_norm": 0.5577086271643042,
      "learning_rate": 9.642629967049799e-06,
      "loss": 0.0595,
      "step": 2338
    },
    {
      "epoch": 0.6272459104317512,
      "grad_norm": 0.4059179032915396,
      "learning_rate": 9.642050494453323e-06,
      "loss": 0.0377,
      "step": 2339
    },
    {
      "epoch": 0.6275140788415124,
      "grad_norm": 0.3274461308138133,
      "learning_rate": 9.641470569872715e-06,
      "loss": 0.0334,
      "step": 2340
    },
    {
      "epoch": 0.6277822472512739,
      "grad_norm": 0.4533377206356322,
      "learning_rate": 9.64089019336444e-06,
      "loss": 0.0379,
      "step": 2341
    },
    {
      "epoch": 0.6280504156610351,
      "grad_norm": 0.41142488777789143,
      "learning_rate": 9.64030936498501e-06,
      "loss": 0.0417,
      "step": 2342
    },
    {
      "epoch": 0.6283185840707964,
      "grad_norm": 0.31353325891357314,
      "learning_rate": 9.639728084790976e-06,
      "loss": 0.031,
      "step": 2343
    },
    {
      "epoch": 0.6285867524805578,
      "grad_norm": 0.43536443553142146,
      "learning_rate": 9.639146352838935e-06,
      "loss": 0.0533,
      "step": 2344
    },
    {
      "epoch": 0.6288549208903191,
      "grad_norm": 0.4717141453954006,
      "learning_rate": 9.638564169185532e-06,
      "loss": 0.0461,
      "step": 2345
    },
    {
      "epoch": 0.6291230893000804,
      "grad_norm": 0.3298703550734682,
      "learning_rate": 9.63798153388745e-06,
      "loss": 0.0295,
      "step": 2346
    },
    {
      "epoch": 0.6293912577098418,
      "grad_norm": 0.44540558681983444,
      "learning_rate": 9.63739844700142e-06,
      "loss": 0.0378,
      "step": 2347
    },
    {
      "epoch": 0.6296594261196031,
      "grad_norm": 0.35193675516462675,
      "learning_rate": 9.636814908584215e-06,
      "loss": 0.0448,
      "step": 2348
    },
    {
      "epoch": 0.6299275945293644,
      "grad_norm": 0.5073069239342581,
      "learning_rate": 9.636230918692654e-06,
      "loss": 0.0565,
      "step": 2349
    },
    {
      "epoch": 0.6301957629391258,
      "grad_norm": 0.41555359739675146,
      "learning_rate": 9.635646477383597e-06,
      "loss": 0.0304,
      "step": 2350
    },
    {
      "epoch": 0.6304639313488871,
      "grad_norm": 0.5570800204635162,
      "learning_rate": 9.635061584713949e-06,
      "loss": 0.0433,
      "step": 2351
    },
    {
      "epoch": 0.6307320997586484,
      "grad_norm": 0.333977143755449,
      "learning_rate": 9.634476240740662e-06,
      "loss": 0.0305,
      "step": 2352
    },
    {
      "epoch": 0.6310002681684097,
      "grad_norm": 0.4597261511904927,
      "learning_rate": 9.633890445520726e-06,
      "loss": 0.0322,
      "step": 2353
    },
    {
      "epoch": 0.6312684365781711,
      "grad_norm": 0.34734635337566555,
      "learning_rate": 9.63330419911118e-06,
      "loss": 0.0407,
      "step": 2354
    },
    {
      "epoch": 0.6315366049879324,
      "grad_norm": 0.37833481662675106,
      "learning_rate": 9.632717501569106e-06,
      "loss": 0.0442,
      "step": 2355
    },
    {
      "epoch": 0.6318047733976937,
      "grad_norm": 0.38784359330636803,
      "learning_rate": 9.63213035295163e-06,
      "loss": 0.0359,
      "step": 2356
    },
    {
      "epoch": 0.6320729418074551,
      "grad_norm": 0.30656933615245535,
      "learning_rate": 9.631542753315917e-06,
      "loss": 0.0297,
      "step": 2357
    },
    {
      "epoch": 0.6323411102172164,
      "grad_norm": 0.4015790915141212,
      "learning_rate": 9.630954702719184e-06,
      "loss": 0.0418,
      "step": 2358
    },
    {
      "epoch": 0.6326092786269777,
      "grad_norm": 0.4126321499975713,
      "learning_rate": 9.630366201218686e-06,
      "loss": 0.0443,
      "step": 2359
    },
    {
      "epoch": 0.6328774470367391,
      "grad_norm": 0.48963380843404836,
      "learning_rate": 9.629777248871725e-06,
      "loss": 0.0345,
      "step": 2360
    },
    {
      "epoch": 0.6331456154465004,
      "grad_norm": 0.46095669428663266,
      "learning_rate": 9.629187845735645e-06,
      "loss": 0.0335,
      "step": 2361
    },
    {
      "epoch": 0.6334137838562617,
      "grad_norm": 0.5838893428481031,
      "learning_rate": 9.628597991867834e-06,
      "loss": 0.0575,
      "step": 2362
    },
    {
      "epoch": 0.6336819522660231,
      "grad_norm": 0.325218286175406,
      "learning_rate": 9.628007687325726e-06,
      "loss": 0.0372,
      "step": 2363
    },
    {
      "epoch": 0.6339501206757844,
      "grad_norm": 0.9756261661285838,
      "learning_rate": 9.627416932166795e-06,
      "loss": 0.0524,
      "step": 2364
    },
    {
      "epoch": 0.6342182890855457,
      "grad_norm": 0.36053462758323435,
      "learning_rate": 9.626825726448565e-06,
      "loss": 0.0361,
      "step": 2365
    },
    {
      "epoch": 0.6344864574953071,
      "grad_norm": 0.3325114396441304,
      "learning_rate": 9.626234070228597e-06,
      "loss": 0.0325,
      "step": 2366
    },
    {
      "epoch": 0.6347546259050684,
      "grad_norm": 0.46119397884936003,
      "learning_rate": 9.6256419635645e-06,
      "loss": 0.0376,
      "step": 2367
    },
    {
      "epoch": 0.6350227943148297,
      "grad_norm": 0.4507052840812779,
      "learning_rate": 9.625049406513926e-06,
      "loss": 0.0444,
      "step": 2368
    },
    {
      "epoch": 0.6352909627245911,
      "grad_norm": 0.3856328897999926,
      "learning_rate": 9.624456399134571e-06,
      "loss": 0.0396,
      "step": 2369
    },
    {
      "epoch": 0.6355591311343524,
      "grad_norm": 0.3756979633762363,
      "learning_rate": 9.623862941484174e-06,
      "loss": 0.039,
      "step": 2370
    },
    {
      "epoch": 0.6358272995441137,
      "grad_norm": 0.2936440553209829,
      "learning_rate": 9.623269033620518e-06,
      "loss": 0.0219,
      "step": 2371
    },
    {
      "epoch": 0.6360954679538751,
      "grad_norm": 0.8162721629582562,
      "learning_rate": 9.622674675601431e-06,
      "loss": 0.0483,
      "step": 2372
    },
    {
      "epoch": 0.6363636363636364,
      "grad_norm": 0.3916503536588522,
      "learning_rate": 9.622079867484785e-06,
      "loss": 0.032,
      "step": 2373
    },
    {
      "epoch": 0.6366318047733976,
      "grad_norm": 0.5746142914165883,
      "learning_rate": 9.621484609328491e-06,
      "loss": 0.0417,
      "step": 2374
    },
    {
      "epoch": 0.636899973183159,
      "grad_norm": 0.4322147575797061,
      "learning_rate": 9.620888901190513e-06,
      "loss": 0.0442,
      "step": 2375
    },
    {
      "epoch": 0.6371681415929203,
      "grad_norm": 0.3742842450561866,
      "learning_rate": 9.62029274312885e-06,
      "loss": 0.0447,
      "step": 2376
    },
    {
      "epoch": 0.6374363100026816,
      "grad_norm": 0.3680071447793632,
      "learning_rate": 9.619696135201549e-06,
      "loss": 0.0313,
      "step": 2377
    },
    {
      "epoch": 0.637704478412443,
      "grad_norm": 0.3939743367736881,
      "learning_rate": 9.619099077466699e-06,
      "loss": 0.0406,
      "step": 2378
    },
    {
      "epoch": 0.6379726468222043,
      "grad_norm": 0.40012194755094194,
      "learning_rate": 9.618501569982437e-06,
      "loss": 0.0445,
      "step": 2379
    },
    {
      "epoch": 0.6382408152319656,
      "grad_norm": 0.2973338726410971,
      "learning_rate": 9.617903612806938e-06,
      "loss": 0.0328,
      "step": 2380
    },
    {
      "epoch": 0.638508983641727,
      "grad_norm": 0.6488117433440108,
      "learning_rate": 9.617305205998427e-06,
      "loss": 0.0536,
      "step": 2381
    },
    {
      "epoch": 0.6387771520514883,
      "grad_norm": 0.49315862534546395,
      "learning_rate": 9.616706349615163e-06,
      "loss": 0.0461,
      "step": 2382
    },
    {
      "epoch": 0.6390453204612496,
      "grad_norm": 0.45762092630839146,
      "learning_rate": 9.616107043715462e-06,
      "loss": 0.0364,
      "step": 2383
    },
    {
      "epoch": 0.639313488871011,
      "grad_norm": 0.5471956231718901,
      "learning_rate": 9.615507288357671e-06,
      "loss": 0.0438,
      "step": 2384
    },
    {
      "epoch": 0.6395816572807723,
      "grad_norm": 0.5165387190817277,
      "learning_rate": 9.614907083600191e-06,
      "loss": 0.0594,
      "step": 2385
    },
    {
      "epoch": 0.6398498256905336,
      "grad_norm": 0.2879088575027302,
      "learning_rate": 9.614306429501461e-06,
      "loss": 0.026,
      "step": 2386
    },
    {
      "epoch": 0.640117994100295,
      "grad_norm": 0.33367566019930045,
      "learning_rate": 9.613705326119964e-06,
      "loss": 0.0354,
      "step": 2387
    },
    {
      "epoch": 0.6403861625100563,
      "grad_norm": 0.47695440721622656,
      "learning_rate": 9.613103773514229e-06,
      "loss": 0.0505,
      "step": 2388
    },
    {
      "epoch": 0.6406543309198176,
      "grad_norm": 0.39381038539484803,
      "learning_rate": 9.612501771742825e-06,
      "loss": 0.04,
      "step": 2389
    },
    {
      "epoch": 0.640922499329579,
      "grad_norm": 0.30907986913105073,
      "learning_rate": 9.611899320864374e-06,
      "loss": 0.0294,
      "step": 2390
    },
    {
      "epoch": 0.6411906677393403,
      "grad_norm": 0.39142191760284184,
      "learning_rate": 9.611296420937526e-06,
      "loss": 0.0461,
      "step": 2391
    },
    {
      "epoch": 0.6414588361491016,
      "grad_norm": 0.6549380522117229,
      "learning_rate": 9.61069307202099e-06,
      "loss": 0.0641,
      "step": 2392
    },
    {
      "epoch": 0.641727004558863,
      "grad_norm": 0.4241414654650714,
      "learning_rate": 9.61008927417351e-06,
      "loss": 0.0418,
      "step": 2393
    },
    {
      "epoch": 0.6419951729686243,
      "grad_norm": 0.6335894993721941,
      "learning_rate": 9.609485027453878e-06,
      "loss": 0.0361,
      "step": 2394
    },
    {
      "epoch": 0.6422633413783856,
      "grad_norm": 0.3288830427939089,
      "learning_rate": 9.608880331920927e-06,
      "loss": 0.0347,
      "step": 2395
    },
    {
      "epoch": 0.642531509788147,
      "grad_norm": 0.5051519359924269,
      "learning_rate": 9.608275187633533e-06,
      "loss": 0.0322,
      "step": 2396
    },
    {
      "epoch": 0.6427996781979083,
      "grad_norm": 0.47400700635569226,
      "learning_rate": 9.60766959465062e-06,
      "loss": 0.0334,
      "step": 2397
    },
    {
      "epoch": 0.6430678466076696,
      "grad_norm": 0.3763986125445419,
      "learning_rate": 9.607063553031152e-06,
      "loss": 0.0339,
      "step": 2398
    },
    {
      "epoch": 0.643336015017431,
      "grad_norm": 0.5680003205967475,
      "learning_rate": 9.606457062834137e-06,
      "loss": 0.0495,
      "step": 2399
    },
    {
      "epoch": 0.6436041834271923,
      "grad_norm": 0.4451727428090002,
      "learning_rate": 9.605850124118627e-06,
      "loss": 0.034,
      "step": 2400
    },
    {
      "epoch": 0.6438723518369536,
      "grad_norm": 0.9832018030631693,
      "learning_rate": 9.605242736943718e-06,
      "loss": 0.037,
      "step": 2401
    },
    {
      "epoch": 0.644140520246715,
      "grad_norm": 0.3035305558820535,
      "learning_rate": 9.604634901368553e-06,
      "loss": 0.0343,
      "step": 2402
    },
    {
      "epoch": 0.6444086886564763,
      "grad_norm": 0.6354074868499225,
      "learning_rate": 9.60402661745231e-06,
      "loss": 0.0471,
      "step": 2403
    },
    {
      "epoch": 0.6446768570662376,
      "grad_norm": 0.34136416753254806,
      "learning_rate": 9.60341788525422e-06,
      "loss": 0.049,
      "step": 2404
    },
    {
      "epoch": 0.644945025475999,
      "grad_norm": 0.626553087598181,
      "learning_rate": 9.602808704833553e-06,
      "loss": 0.0467,
      "step": 2405
    },
    {
      "epoch": 0.6452131938857603,
      "grad_norm": 0.5093310108556991,
      "learning_rate": 9.602199076249621e-06,
      "loss": 0.0457,
      "step": 2406
    },
    {
      "epoch": 0.6454813622955216,
      "grad_norm": 0.33155382181522636,
      "learning_rate": 9.601588999561784e-06,
      "loss": 0.0368,
      "step": 2407
    },
    {
      "epoch": 0.645749530705283,
      "grad_norm": 0.6191788648658846,
      "learning_rate": 9.600978474829443e-06,
      "loss": 0.0336,
      "step": 2408
    },
    {
      "epoch": 0.6460176991150443,
      "grad_norm": 0.33634917035568584,
      "learning_rate": 9.600367502112044e-06,
      "loss": 0.0296,
      "step": 2409
    },
    {
      "epoch": 0.6462858675248055,
      "grad_norm": 0.5203711763493594,
      "learning_rate": 9.599756081469076e-06,
      "loss": 0.0415,
      "step": 2410
    },
    {
      "epoch": 0.646554035934567,
      "grad_norm": 0.4494497834554676,
      "learning_rate": 9.59914421296007e-06,
      "loss": 0.0398,
      "step": 2411
    },
    {
      "epoch": 0.6468222043443282,
      "grad_norm": 0.42143423700300287,
      "learning_rate": 9.5985318966446e-06,
      "loss": 0.0389,
      "step": 2412
    },
    {
      "epoch": 0.6470903727540895,
      "grad_norm": 0.42023154114947653,
      "learning_rate": 9.59791913258229e-06,
      "loss": 0.0462,
      "step": 2413
    },
    {
      "epoch": 0.6473585411638509,
      "grad_norm": 0.35897589537607527,
      "learning_rate": 9.597305920832801e-06,
      "loss": 0.0313,
      "step": 2414
    },
    {
      "epoch": 0.6476267095736122,
      "grad_norm": 0.3570337438373847,
      "learning_rate": 9.59669226145584e-06,
      "loss": 0.0304,
      "step": 2415
    },
    {
      "epoch": 0.6478948779833735,
      "grad_norm": 0.510795561905092,
      "learning_rate": 9.596078154511159e-06,
      "loss": 0.0473,
      "step": 2416
    },
    {
      "epoch": 0.6481630463931349,
      "grad_norm": 1.0685936705529246,
      "learning_rate": 9.595463600058547e-06,
      "loss": 0.0446,
      "step": 2417
    },
    {
      "epoch": 0.6484312148028962,
      "grad_norm": 0.32259851236522114,
      "learning_rate": 9.594848598157848e-06,
      "loss": 0.0376,
      "step": 2418
    },
    {
      "epoch": 0.6486993832126575,
      "grad_norm": 0.6224002037008342,
      "learning_rate": 9.59423314886894e-06,
      "loss": 0.0524,
      "step": 2419
    },
    {
      "epoch": 0.6489675516224189,
      "grad_norm": 0.43694412247903197,
      "learning_rate": 9.593617252251747e-06,
      "loss": 0.0465,
      "step": 2420
    },
    {
      "epoch": 0.6492357200321802,
      "grad_norm": 0.5181856402056153,
      "learning_rate": 9.593000908366238e-06,
      "loss": 0.0516,
      "step": 2421
    },
    {
      "epoch": 0.6495038884419415,
      "grad_norm": 0.36826459683129004,
      "learning_rate": 9.592384117272424e-06,
      "loss": 0.0291,
      "step": 2422
    },
    {
      "epoch": 0.6497720568517029,
      "grad_norm": 0.43393607236531845,
      "learning_rate": 9.591766879030363e-06,
      "loss": 0.041,
      "step": 2423
    },
    {
      "epoch": 0.6500402252614642,
      "grad_norm": 0.3962791427376134,
      "learning_rate": 9.59114919370015e-06,
      "loss": 0.0341,
      "step": 2424
    },
    {
      "epoch": 0.6503083936712255,
      "grad_norm": 0.6295076291848843,
      "learning_rate": 9.59053106134193e-06,
      "loss": 0.053,
      "step": 2425
    },
    {
      "epoch": 0.6505765620809869,
      "grad_norm": 0.39380011469837295,
      "learning_rate": 9.589912482015888e-06,
      "loss": 0.0383,
      "step": 2426
    },
    {
      "epoch": 0.6508447304907482,
      "grad_norm": 0.2598954492813354,
      "learning_rate": 9.589293455782253e-06,
      "loss": 0.0239,
      "step": 2427
    },
    {
      "epoch": 0.6511128989005095,
      "grad_norm": 0.817868949073019,
      "learning_rate": 9.5886739827013e-06,
      "loss": 0.0425,
      "step": 2428
    },
    {
      "epoch": 0.6513810673102709,
      "grad_norm": 0.44105717206507045,
      "learning_rate": 9.588054062833343e-06,
      "loss": 0.0447,
      "step": 2429
    },
    {
      "epoch": 0.6516492357200322,
      "grad_norm": 0.42351482218457953,
      "learning_rate": 9.587433696238742e-06,
      "loss": 0.0361,
      "step": 2430
    },
    {
      "epoch": 0.6519174041297935,
      "grad_norm": 0.3201253240795514,
      "learning_rate": 9.586812882977902e-06,
      "loss": 0.035,
      "step": 2431
    },
    {
      "epoch": 0.6521855725395549,
      "grad_norm": 0.31174656582466975,
      "learning_rate": 9.586191623111272e-06,
      "loss": 0.0301,
      "step": 2432
    },
    {
      "epoch": 0.6524537409493162,
      "grad_norm": 0.2379491745497701,
      "learning_rate": 9.585569916699337e-06,
      "loss": 0.0258,
      "step": 2433
    },
    {
      "epoch": 0.6527219093590775,
      "grad_norm": 0.3597768311397914,
      "learning_rate": 9.584947763802635e-06,
      "loss": 0.0375,
      "step": 2434
    },
    {
      "epoch": 0.6529900777688389,
      "grad_norm": 0.8436306951206857,
      "learning_rate": 9.58432516448174e-06,
      "loss": 0.0551,
      "step": 2435
    },
    {
      "epoch": 0.6532582461786002,
      "grad_norm": 0.4146036138808878,
      "learning_rate": 9.583702118797277e-06,
      "loss": 0.0432,
      "step": 2436
    },
    {
      "epoch": 0.6535264145883615,
      "grad_norm": 0.5496255201024303,
      "learning_rate": 9.583078626809907e-06,
      "loss": 0.0525,
      "step": 2437
    },
    {
      "epoch": 0.6537945829981229,
      "grad_norm": 0.3729896976666874,
      "learning_rate": 9.58245468858034e-06,
      "loss": 0.0284,
      "step": 2438
    },
    {
      "epoch": 0.6540627514078842,
      "grad_norm": 1.127738702096665,
      "learning_rate": 9.581830304169325e-06,
      "loss": 0.038,
      "step": 2439
    },
    {
      "epoch": 0.6543309198176455,
      "grad_norm": 0.3253717241871863,
      "learning_rate": 9.581205473637659e-06,
      "loss": 0.0343,
      "step": 2440
    },
    {
      "epoch": 0.6545990882274069,
      "grad_norm": 1.0452832246653085,
      "learning_rate": 9.580580197046177e-06,
      "loss": 0.0434,
      "step": 2441
    },
    {
      "epoch": 0.6548672566371682,
      "grad_norm": 0.35309634759056413,
      "learning_rate": 9.579954474455763e-06,
      "loss": 0.0361,
      "step": 2442
    },
    {
      "epoch": 0.6551354250469295,
      "grad_norm": 0.47579145737988227,
      "learning_rate": 9.579328305927343e-06,
      "loss": 0.0384,
      "step": 2443
    },
    {
      "epoch": 0.6554035934566907,
      "grad_norm": 0.8231036525009132,
      "learning_rate": 9.578701691521884e-06,
      "loss": 0.0434,
      "step": 2444
    },
    {
      "epoch": 0.6556717618664522,
      "grad_norm": 0.4654866188529236,
      "learning_rate": 9.578074631300396e-06,
      "loss": 0.0361,
      "step": 2445
    },
    {
      "epoch": 0.6559399302762134,
      "grad_norm": 0.36001860117085316,
      "learning_rate": 9.577447125323935e-06,
      "loss": 0.031,
      "step": 2446
    },
    {
      "epoch": 0.6562080986859747,
      "grad_norm": 0.2925098776649432,
      "learning_rate": 9.576819173653602e-06,
      "loss": 0.0295,
      "step": 2447
    },
    {
      "epoch": 0.6564762670957361,
      "grad_norm": 0.3427065794250314,
      "learning_rate": 9.576190776350536e-06,
      "loss": 0.0261,
      "step": 2448
    },
    {
      "epoch": 0.6567444355054974,
      "grad_norm": 0.9873337287243593,
      "learning_rate": 9.575561933475926e-06,
      "loss": 0.0388,
      "step": 2449
    },
    {
      "epoch": 0.6570126039152587,
      "grad_norm": 0.4922538569310218,
      "learning_rate": 9.574932645090995e-06,
      "loss": 0.0472,
      "step": 2450
    },
    {
      "epoch": 0.6572807723250201,
      "grad_norm": 0.5722429065472527,
      "learning_rate": 9.574302911257021e-06,
      "loss": 0.04,
      "step": 2451
    },
    {
      "epoch": 0.6575489407347814,
      "grad_norm": 0.64442187215386,
      "learning_rate": 9.573672732035315e-06,
      "loss": 0.0429,
      "step": 2452
    },
    {
      "epoch": 0.6578171091445427,
      "grad_norm": 0.43402451605292736,
      "learning_rate": 9.57304210748724e-06,
      "loss": 0.0388,
      "step": 2453
    },
    {
      "epoch": 0.6580852775543041,
      "grad_norm": 0.3536492387021235,
      "learning_rate": 9.572411037674196e-06,
      "loss": 0.0304,
      "step": 2454
    },
    {
      "epoch": 0.6583534459640654,
      "grad_norm": 0.4402933639571913,
      "learning_rate": 9.571779522657628e-06,
      "loss": 0.0378,
      "step": 2455
    },
    {
      "epoch": 0.6586216143738267,
      "grad_norm": 0.3869416724247881,
      "learning_rate": 9.571147562499026e-06,
      "loss": 0.0429,
      "step": 2456
    },
    {
      "epoch": 0.6588897827835881,
      "grad_norm": 0.49193432809463833,
      "learning_rate": 9.57051515725992e-06,
      "loss": 0.057,
      "step": 2457
    },
    {
      "epoch": 0.6591579511933494,
      "grad_norm": 0.5806472762725997,
      "learning_rate": 9.569882307001891e-06,
      "loss": 0.0451,
      "step": 2458
    },
    {
      "epoch": 0.6594261196031107,
      "grad_norm": 0.46980681800654517,
      "learning_rate": 9.569249011786552e-06,
      "loss": 0.0455,
      "step": 2459
    },
    {
      "epoch": 0.6596942880128721,
      "grad_norm": 0.44967370251054656,
      "learning_rate": 9.56861527167557e-06,
      "loss": 0.0456,
      "step": 2460
    },
    {
      "epoch": 0.6599624564226334,
      "grad_norm": 0.5928011679265865,
      "learning_rate": 9.567981086730646e-06,
      "loss": 0.0416,
      "step": 2461
    },
    {
      "epoch": 0.6602306248323947,
      "grad_norm": 0.6376397959798353,
      "learning_rate": 9.567346457013533e-06,
      "loss": 0.0569,
      "step": 2462
    },
    {
      "epoch": 0.6604987932421561,
      "grad_norm": 0.21152571402982312,
      "learning_rate": 9.566711382586022e-06,
      "loss": 0.0237,
      "step": 2463
    },
    {
      "epoch": 0.6607669616519174,
      "grad_norm": 0.4689455073504569,
      "learning_rate": 9.566075863509947e-06,
      "loss": 0.0394,
      "step": 2464
    },
    {
      "epoch": 0.6610351300616787,
      "grad_norm": 0.4584965534072919,
      "learning_rate": 9.565439899847188e-06,
      "loss": 0.0519,
      "step": 2465
    },
    {
      "epoch": 0.6613032984714401,
      "grad_norm": 0.49901060559888427,
      "learning_rate": 9.564803491659668e-06,
      "loss": 0.0446,
      "step": 2466
    },
    {
      "epoch": 0.6615714668812014,
      "grad_norm": 0.34127402888108216,
      "learning_rate": 9.564166639009351e-06,
      "loss": 0.0357,
      "step": 2467
    },
    {
      "epoch": 0.6618396352909627,
      "grad_norm": 0.31564006057320915,
      "learning_rate": 9.563529341958245e-06,
      "loss": 0.0334,
      "step": 2468
    },
    {
      "epoch": 0.6621078037007241,
      "grad_norm": 0.2805183913596248,
      "learning_rate": 9.562891600568403e-06,
      "loss": 0.0298,
      "step": 2469
    },
    {
      "epoch": 0.6623759721104854,
      "grad_norm": 0.2861276794650215,
      "learning_rate": 9.562253414901921e-06,
      "loss": 0.0302,
      "step": 2470
    },
    {
      "epoch": 0.6626441405202467,
      "grad_norm": 0.4707432008524316,
      "learning_rate": 9.561614785020937e-06,
      "loss": 0.0467,
      "step": 2471
    },
    {
      "epoch": 0.6629123089300081,
      "grad_norm": 0.38525907337681276,
      "learning_rate": 9.56097571098763e-06,
      "loss": 0.0478,
      "step": 2472
    },
    {
      "epoch": 0.6631804773397694,
      "grad_norm": 0.455098753017548,
      "learning_rate": 9.560336192864227e-06,
      "loss": 0.0396,
      "step": 2473
    },
    {
      "epoch": 0.6634486457495307,
      "grad_norm": 0.4910643876533477,
      "learning_rate": 9.559696230713e-06,
      "loss": 0.0338,
      "step": 2474
    },
    {
      "epoch": 0.6637168141592921,
      "grad_norm": 0.3463796325396375,
      "learning_rate": 9.559055824596252e-06,
      "loss": 0.025,
      "step": 2475
    },
    {
      "epoch": 0.6639849825690534,
      "grad_norm": 0.3507585873572776,
      "learning_rate": 9.558414974576346e-06,
      "loss": 0.0351,
      "step": 2476
    },
    {
      "epoch": 0.6642531509788147,
      "grad_norm": 0.3371623164827828,
      "learning_rate": 9.557773680715676e-06,
      "loss": 0.0315,
      "step": 2477
    },
    {
      "epoch": 0.6645213193885761,
      "grad_norm": 0.3495124463857507,
      "learning_rate": 9.557131943076683e-06,
      "loss": 0.0342,
      "step": 2478
    },
    {
      "epoch": 0.6647894877983374,
      "grad_norm": 0.591054972580119,
      "learning_rate": 9.55648976172185e-06,
      "loss": 0.0467,
      "step": 2479
    },
    {
      "epoch": 0.6650576562080986,
      "grad_norm": 0.4820148096233884,
      "learning_rate": 9.555847136713707e-06,
      "loss": 0.0512,
      "step": 2480
    },
    {
      "epoch": 0.66532582461786,
      "grad_norm": 0.27898432702346293,
      "learning_rate": 9.555204068114826e-06,
      "loss": 0.0321,
      "step": 2481
    },
    {
      "epoch": 0.6655939930276213,
      "grad_norm": 1.0538362878660053,
      "learning_rate": 9.554560555987816e-06,
      "loss": 0.0505,
      "step": 2482
    },
    {
      "epoch": 0.6658621614373826,
      "grad_norm": 0.3751456903140579,
      "learning_rate": 9.55391660039534e-06,
      "loss": 0.0309,
      "step": 2483
    },
    {
      "epoch": 0.666130329847144,
      "grad_norm": 0.38022316589501687,
      "learning_rate": 9.55327220140009e-06,
      "loss": 0.0367,
      "step": 2484
    },
    {
      "epoch": 0.6663984982569053,
      "grad_norm": 0.44799752919908253,
      "learning_rate": 9.55262735906482e-06,
      "loss": 0.0455,
      "step": 2485
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 0.5767863237603938,
      "learning_rate": 9.551982073452305e-06,
      "loss": 0.0463,
      "step": 2486
    },
    {
      "epoch": 0.666934835076428,
      "grad_norm": 0.25303247890909913,
      "learning_rate": 9.551336344625387e-06,
      "loss": 0.0286,
      "step": 2487
    },
    {
      "epoch": 0.6672030034861893,
      "grad_norm": 0.29930611475809443,
      "learning_rate": 9.550690172646928e-06,
      "loss": 0.0284,
      "step": 2488
    },
    {
      "epoch": 0.6674711718959506,
      "grad_norm": 0.502576244389439,
      "learning_rate": 9.55004355757985e-06,
      "loss": 0.0388,
      "step": 2489
    },
    {
      "epoch": 0.667739340305712,
      "grad_norm": 0.5509057566369913,
      "learning_rate": 9.549396499487112e-06,
      "loss": 0.0502,
      "step": 2490
    },
    {
      "epoch": 0.6680075087154733,
      "grad_norm": 0.4595956673690557,
      "learning_rate": 9.548748998431715e-06,
      "loss": 0.0479,
      "step": 2491
    },
    {
      "epoch": 0.6682756771252346,
      "grad_norm": 0.792796775270736,
      "learning_rate": 9.548101054476704e-06,
      "loss": 0.0645,
      "step": 2492
    },
    {
      "epoch": 0.668543845534996,
      "grad_norm": 0.3158969278603121,
      "learning_rate": 9.54745266768517e-06,
      "loss": 0.038,
      "step": 2493
    },
    {
      "epoch": 0.6688120139447573,
      "grad_norm": 0.40803514424811943,
      "learning_rate": 9.546803838120241e-06,
      "loss": 0.039,
      "step": 2494
    },
    {
      "epoch": 0.6690801823545186,
      "grad_norm": 0.5313483773031777,
      "learning_rate": 9.546154565845095e-06,
      "loss": 0.042,
      "step": 2495
    },
    {
      "epoch": 0.66934835076428,
      "grad_norm": 0.4605502204898016,
      "learning_rate": 9.545504850922947e-06,
      "loss": 0.0415,
      "step": 2496
    },
    {
      "epoch": 0.6696165191740413,
      "grad_norm": 0.45128250519507584,
      "learning_rate": 9.544854693417061e-06,
      "loss": 0.0364,
      "step": 2497
    },
    {
      "epoch": 0.6698846875838026,
      "grad_norm": 0.47598216038976343,
      "learning_rate": 9.544204093390741e-06,
      "loss": 0.0495,
      "step": 2498
    },
    {
      "epoch": 0.670152855993564,
      "grad_norm": 0.4048113685517227,
      "learning_rate": 9.543553050907332e-06,
      "loss": 0.0369,
      "step": 2499
    },
    {
      "epoch": 0.6704210244033253,
      "grad_norm": 0.35790476949815786,
      "learning_rate": 9.542901566030225e-06,
      "loss": 0.0333,
      "step": 2500
    },
    {
      "epoch": 0.6706891928130866,
      "grad_norm": 0.35151156933088024,
      "learning_rate": 9.542249638822854e-06,
      "loss": 0.0283,
      "step": 2501
    },
    {
      "epoch": 0.670957361222848,
      "grad_norm": 0.8957338686836287,
      "learning_rate": 9.541597269348695e-06,
      "loss": 0.0649,
      "step": 2502
    },
    {
      "epoch": 0.6712255296326093,
      "grad_norm": 0.3778752702840763,
      "learning_rate": 9.540944457671267e-06,
      "loss": 0.0339,
      "step": 2503
    },
    {
      "epoch": 0.6714936980423706,
      "grad_norm": 0.4435685329985762,
      "learning_rate": 9.540291203854135e-06,
      "loss": 0.0342,
      "step": 2504
    },
    {
      "epoch": 0.671761866452132,
      "grad_norm": 0.3457214893318813,
      "learning_rate": 9.5396375079609e-06,
      "loss": 0.0354,
      "step": 2505
    },
    {
      "epoch": 0.6720300348618933,
      "grad_norm": 0.582011908779313,
      "learning_rate": 9.538983370055215e-06,
      "loss": 0.0331,
      "step": 2506
    },
    {
      "epoch": 0.6722982032716546,
      "grad_norm": 0.4103006624082808,
      "learning_rate": 9.53832879020077e-06,
      "loss": 0.0459,
      "step": 2507
    },
    {
      "epoch": 0.672566371681416,
      "grad_norm": 0.2693647014282593,
      "learning_rate": 9.5376737684613e-06,
      "loss": 0.0247,
      "step": 2508
    },
    {
      "epoch": 0.6728345400911773,
      "grad_norm": 0.4183588272621983,
      "learning_rate": 9.537018304900583e-06,
      "loss": 0.0393,
      "step": 2509
    },
    {
      "epoch": 0.6731027085009386,
      "grad_norm": 0.5369666746817175,
      "learning_rate": 9.536362399582438e-06,
      "loss": 0.0633,
      "step": 2510
    },
    {
      "epoch": 0.6733708769107,
      "grad_norm": 0.36759454430928884,
      "learning_rate": 9.53570605257073e-06,
      "loss": 0.0367,
      "step": 2511
    },
    {
      "epoch": 0.6736390453204613,
      "grad_norm": 0.3846405856360957,
      "learning_rate": 9.535049263929366e-06,
      "loss": 0.043,
      "step": 2512
    },
    {
      "epoch": 0.6739072137302226,
      "grad_norm": 0.41733956583088194,
      "learning_rate": 9.534392033722295e-06,
      "loss": 0.0409,
      "step": 2513
    },
    {
      "epoch": 0.674175382139984,
      "grad_norm": 0.6748139333251381,
      "learning_rate": 9.53373436201351e-06,
      "loss": 0.0425,
      "step": 2514
    },
    {
      "epoch": 0.6744435505497453,
      "grad_norm": 0.406760941920907,
      "learning_rate": 9.533076248867048e-06,
      "loss": 0.0331,
      "step": 2515
    },
    {
      "epoch": 0.6747117189595065,
      "grad_norm": 0.6370082990594801,
      "learning_rate": 9.532417694346987e-06,
      "loss": 0.0427,
      "step": 2516
    },
    {
      "epoch": 0.674979887369268,
      "grad_norm": 0.40993807671236726,
      "learning_rate": 9.531758698517449e-06,
      "loss": 0.0395,
      "step": 2517
    },
    {
      "epoch": 0.6752480557790292,
      "grad_norm": 0.43755432883047446,
      "learning_rate": 9.531099261442596e-06,
      "loss": 0.0328,
      "step": 2518
    },
    {
      "epoch": 0.6755162241887905,
      "grad_norm": 0.690390566912873,
      "learning_rate": 9.530439383186637e-06,
      "loss": 0.0687,
      "step": 2519
    },
    {
      "epoch": 0.6757843925985519,
      "grad_norm": 0.5754759549920266,
      "learning_rate": 9.529779063813826e-06,
      "loss": 0.0339,
      "step": 2520
    },
    {
      "epoch": 0.6760525610083132,
      "grad_norm": 0.7550415783642387,
      "learning_rate": 9.529118303388454e-06,
      "loss": 0.039,
      "step": 2521
    },
    {
      "epoch": 0.6763207294180745,
      "grad_norm": 0.35245417283232505,
      "learning_rate": 9.528457101974855e-06,
      "loss": 0.034,
      "step": 2522
    },
    {
      "epoch": 0.6765888978278359,
      "grad_norm": 0.7118386755008984,
      "learning_rate": 9.527795459637413e-06,
      "loss": 0.0458,
      "step": 2523
    },
    {
      "epoch": 0.6768570662375972,
      "grad_norm": 0.49006175605253444,
      "learning_rate": 9.527133376440547e-06,
      "loss": 0.0333,
      "step": 2524
    },
    {
      "epoch": 0.6771252346473585,
      "grad_norm": 0.2789279502086122,
      "learning_rate": 9.526470852448724e-06,
      "loss": 0.0243,
      "step": 2525
    },
    {
      "epoch": 0.6773934030571199,
      "grad_norm": 0.39351651684997646,
      "learning_rate": 9.52580788772645e-06,
      "loss": 0.0411,
      "step": 2526
    },
    {
      "epoch": 0.6776615714668812,
      "grad_norm": 0.4855281613138422,
      "learning_rate": 9.52514448233828e-06,
      "loss": 0.0293,
      "step": 2527
    },
    {
      "epoch": 0.6779297398766425,
      "grad_norm": 0.44212379171255034,
      "learning_rate": 9.524480636348804e-06,
      "loss": 0.0369,
      "step": 2528
    },
    {
      "epoch": 0.6781979082864039,
      "grad_norm": 0.44747876160053385,
      "learning_rate": 9.523816349822663e-06,
      "loss": 0.042,
      "step": 2529
    },
    {
      "epoch": 0.6784660766961652,
      "grad_norm": 0.6072242268782345,
      "learning_rate": 9.523151622824534e-06,
      "loss": 0.0462,
      "step": 2530
    },
    {
      "epoch": 0.6787342451059265,
      "grad_norm": 0.3222438521891234,
      "learning_rate": 9.522486455419137e-06,
      "loss": 0.0356,
      "step": 2531
    },
    {
      "epoch": 0.6790024135156878,
      "grad_norm": 0.3472575731295562,
      "learning_rate": 9.521820847671243e-06,
      "loss": 0.0355,
      "step": 2532
    },
    {
      "epoch": 0.6792705819254492,
      "grad_norm": 0.35892499406226547,
      "learning_rate": 9.52115479964566e-06,
      "loss": 0.0317,
      "step": 2533
    },
    {
      "epoch": 0.6795387503352105,
      "grad_norm": 0.4112503167761095,
      "learning_rate": 9.520488311407237e-06,
      "loss": 0.0458,
      "step": 2534
    },
    {
      "epoch": 0.6798069187449718,
      "grad_norm": 0.5881966467078341,
      "learning_rate": 9.519821383020866e-06,
      "loss": 0.0479,
      "step": 2535
    },
    {
      "epoch": 0.6800750871547332,
      "grad_norm": 0.4325103237279451,
      "learning_rate": 9.51915401455149e-06,
      "loss": 0.0387,
      "step": 2536
    },
    {
      "epoch": 0.6803432555644945,
      "grad_norm": 0.2996825749001986,
      "learning_rate": 9.518486206064084e-06,
      "loss": 0.0327,
      "step": 2537
    },
    {
      "epoch": 0.6806114239742558,
      "grad_norm": 0.6776460505679925,
      "learning_rate": 9.517817957623673e-06,
      "loss": 0.0326,
      "step": 2538
    },
    {
      "epoch": 0.6808795923840172,
      "grad_norm": 0.39865282048201456,
      "learning_rate": 9.517149269295321e-06,
      "loss": 0.0385,
      "step": 2539
    },
    {
      "epoch": 0.6811477607937785,
      "grad_norm": 0.40569190739321653,
      "learning_rate": 9.516480141144139e-06,
      "loss": 0.0403,
      "step": 2540
    },
    {
      "epoch": 0.6814159292035398,
      "grad_norm": 0.3684866639240183,
      "learning_rate": 9.515810573235275e-06,
      "loss": 0.0393,
      "step": 2541
    },
    {
      "epoch": 0.6816840976133012,
      "grad_norm": 0.3154899506669576,
      "learning_rate": 9.515140565633927e-06,
      "loss": 0.039,
      "step": 2542
    },
    {
      "epoch": 0.6819522660230625,
      "grad_norm": 0.3803868759522678,
      "learning_rate": 9.514470118405328e-06,
      "loss": 0.0381,
      "step": 2543
    },
    {
      "epoch": 0.6822204344328238,
      "grad_norm": 0.3177215285600402,
      "learning_rate": 9.51379923161476e-06,
      "loss": 0.039,
      "step": 2544
    },
    {
      "epoch": 0.6824886028425852,
      "grad_norm": 0.29728491779977984,
      "learning_rate": 9.513127905327545e-06,
      "loss": 0.0254,
      "step": 2545
    },
    {
      "epoch": 0.6827567712523465,
      "grad_norm": 0.4799023322003075,
      "learning_rate": 9.512456139609045e-06,
      "loss": 0.0449,
      "step": 2546
    },
    {
      "epoch": 0.6830249396621078,
      "grad_norm": 0.6179982179369042,
      "learning_rate": 9.511783934524674e-06,
      "loss": 0.0486,
      "step": 2547
    },
    {
      "epoch": 0.6832931080718692,
      "grad_norm": 0.3190895603249532,
      "learning_rate": 9.51111129013988e-06,
      "loss": 0.0353,
      "step": 2548
    },
    {
      "epoch": 0.6835612764816305,
      "grad_norm": 0.28988687997146284,
      "learning_rate": 9.510438206520155e-06,
      "loss": 0.0258,
      "step": 2549
    },
    {
      "epoch": 0.6838294448913917,
      "grad_norm": 0.4381121592562018,
      "learning_rate": 9.509764683731038e-06,
      "loss": 0.0324,
      "step": 2550
    },
    {
      "epoch": 0.6840976133011532,
      "grad_norm": 0.5883931011333319,
      "learning_rate": 9.509090721838106e-06,
      "loss": 0.0434,
      "step": 2551
    },
    {
      "epoch": 0.6843657817109144,
      "grad_norm": 0.4591180277738051,
      "learning_rate": 9.508416320906981e-06,
      "loss": 0.0376,
      "step": 2552
    },
    {
      "epoch": 0.6846339501206757,
      "grad_norm": 0.2808601787284627,
      "learning_rate": 9.50774148100333e-06,
      "loss": 0.0279,
      "step": 2553
    },
    {
      "epoch": 0.6849021185304371,
      "grad_norm": 0.33379721928923084,
      "learning_rate": 9.507066202192858e-06,
      "loss": 0.0374,
      "step": 2554
    },
    {
      "epoch": 0.6851702869401984,
      "grad_norm": 0.333905454460441,
      "learning_rate": 9.506390484541317e-06,
      "loss": 0.0326,
      "step": 2555
    },
    {
      "epoch": 0.6854384553499597,
      "grad_norm": 0.351033039229521,
      "learning_rate": 9.5057143281145e-06,
      "loss": 0.0503,
      "step": 2556
    },
    {
      "epoch": 0.6857066237597211,
      "grad_norm": 0.3489231349528481,
      "learning_rate": 9.505037732978237e-06,
      "loss": 0.0295,
      "step": 2557
    },
    {
      "epoch": 0.6859747921694824,
      "grad_norm": 0.38039790413061136,
      "learning_rate": 9.504360699198414e-06,
      "loss": 0.0512,
      "step": 2558
    },
    {
      "epoch": 0.6862429605792437,
      "grad_norm": 0.4325365751610192,
      "learning_rate": 9.503683226840948e-06,
      "loss": 0.048,
      "step": 2559
    },
    {
      "epoch": 0.6865111289890051,
      "grad_norm": 0.3766428564210721,
      "learning_rate": 9.503005315971803e-06,
      "loss": 0.0444,
      "step": 2560
    },
    {
      "epoch": 0.6867792973987664,
      "grad_norm": 0.267400591958884,
      "learning_rate": 9.502326966656984e-06,
      "loss": 0.0345,
      "step": 2561
    },
    {
      "epoch": 0.6870474658085277,
      "grad_norm": 0.36314328183183203,
      "learning_rate": 9.501648178962544e-06,
      "loss": 0.0396,
      "step": 2562
    },
    {
      "epoch": 0.6873156342182891,
      "grad_norm": 0.4466959766130045,
      "learning_rate": 9.500968952954572e-06,
      "loss": 0.0357,
      "step": 2563
    },
    {
      "epoch": 0.6875838026280504,
      "grad_norm": 0.5250677968669969,
      "learning_rate": 9.500289288699202e-06,
      "loss": 0.0372,
      "step": 2564
    },
    {
      "epoch": 0.6878519710378117,
      "grad_norm": 0.3719551829815669,
      "learning_rate": 9.499609186262612e-06,
      "loss": 0.0254,
      "step": 2565
    },
    {
      "epoch": 0.6881201394475731,
      "grad_norm": 0.5587475972831016,
      "learning_rate": 9.498928645711023e-06,
      "loss": 0.0392,
      "step": 2566
    },
    {
      "epoch": 0.6883883078573344,
      "grad_norm": 0.5663351284849398,
      "learning_rate": 9.498247667110696e-06,
      "loss": 0.0551,
      "step": 2567
    },
    {
      "epoch": 0.6886564762670957,
      "grad_norm": 0.30181982505256044,
      "learning_rate": 9.497566250527935e-06,
      "loss": 0.0312,
      "step": 2568
    },
    {
      "epoch": 0.6889246446768571,
      "grad_norm": 0.32836183299663185,
      "learning_rate": 9.49688439602909e-06,
      "loss": 0.0277,
      "step": 2569
    },
    {
      "epoch": 0.6891928130866184,
      "grad_norm": 0.45476675360163143,
      "learning_rate": 9.496202103680549e-06,
      "loss": 0.036,
      "step": 2570
    },
    {
      "epoch": 0.6894609814963797,
      "grad_norm": 0.3346050811291732,
      "learning_rate": 9.495519373548748e-06,
      "loss": 0.0361,
      "step": 2571
    },
    {
      "epoch": 0.6897291499061411,
      "grad_norm": 0.430181423539497,
      "learning_rate": 9.49483620570016e-06,
      "loss": 0.0379,
      "step": 2572
    },
    {
      "epoch": 0.6899973183159024,
      "grad_norm": 0.4251714909760362,
      "learning_rate": 9.494152600201306e-06,
      "loss": 0.0321,
      "step": 2573
    },
    {
      "epoch": 0.6902654867256637,
      "grad_norm": 0.5993517567355114,
      "learning_rate": 9.493468557118744e-06,
      "loss": 0.0458,
      "step": 2574
    },
    {
      "epoch": 0.6905336551354251,
      "grad_norm": 0.3365545433534593,
      "learning_rate": 9.492784076519077e-06,
      "loss": 0.0377,
      "step": 2575
    },
    {
      "epoch": 0.6908018235451864,
      "grad_norm": 0.4712980170300561,
      "learning_rate": 9.492099158468955e-06,
      "loss": 0.0532,
      "step": 2576
    },
    {
      "epoch": 0.6910699919549477,
      "grad_norm": 0.3653586012199385,
      "learning_rate": 9.491413803035062e-06,
      "loss": 0.0381,
      "step": 2577
    },
    {
      "epoch": 0.6913381603647091,
      "grad_norm": 0.42926558411140125,
      "learning_rate": 9.490728010284132e-06,
      "loss": 0.0418,
      "step": 2578
    },
    {
      "epoch": 0.6916063287744704,
      "grad_norm": 0.4791043152210209,
      "learning_rate": 9.490041780282939e-06,
      "loss": 0.0321,
      "step": 2579
    },
    {
      "epoch": 0.6918744971842317,
      "grad_norm": 0.45313370537527153,
      "learning_rate": 9.489355113098298e-06,
      "loss": 0.0516,
      "step": 2580
    },
    {
      "epoch": 0.6921426655939931,
      "grad_norm": 0.435819599578046,
      "learning_rate": 9.488668008797071e-06,
      "loss": 0.0415,
      "step": 2581
    },
    {
      "epoch": 0.6924108340037544,
      "grad_norm": 0.3933139340371445,
      "learning_rate": 9.487980467446153e-06,
      "loss": 0.0428,
      "step": 2582
    },
    {
      "epoch": 0.6926790024135157,
      "grad_norm": 0.4604090914208074,
      "learning_rate": 9.487292489112497e-06,
      "loss": 0.047,
      "step": 2583
    },
    {
      "epoch": 0.6929471708232771,
      "grad_norm": 0.28180255787866837,
      "learning_rate": 9.486604073863081e-06,
      "loss": 0.0338,
      "step": 2584
    },
    {
      "epoch": 0.6932153392330384,
      "grad_norm": 0.459239400187049,
      "learning_rate": 9.48591522176494e-06,
      "loss": 0.0413,
      "step": 2585
    },
    {
      "epoch": 0.6934835076427996,
      "grad_norm": 0.4225224386795257,
      "learning_rate": 9.485225932885143e-06,
      "loss": 0.0343,
      "step": 2586
    },
    {
      "epoch": 0.693751676052561,
      "grad_norm": 0.42121334035161756,
      "learning_rate": 9.484536207290804e-06,
      "loss": 0.0406,
      "step": 2587
    },
    {
      "epoch": 0.6940198444623223,
      "grad_norm": 0.3290552017152997,
      "learning_rate": 9.483846045049083e-06,
      "loss": 0.0406,
      "step": 2588
    },
    {
      "epoch": 0.6942880128720836,
      "grad_norm": 0.344510815942509,
      "learning_rate": 9.483155446227176e-06,
      "loss": 0.0283,
      "step": 2589
    },
    {
      "epoch": 0.694556181281845,
      "grad_norm": 0.38528408477262127,
      "learning_rate": 9.482464410892324e-06,
      "loss": 0.03,
      "step": 2590
    },
    {
      "epoch": 0.6948243496916063,
      "grad_norm": 0.30466968099348896,
      "learning_rate": 9.481772939111815e-06,
      "loss": 0.0294,
      "step": 2591
    },
    {
      "epoch": 0.6950925181013676,
      "grad_norm": 0.4508906122509676,
      "learning_rate": 9.481081030952974e-06,
      "loss": 0.0394,
      "step": 2592
    },
    {
      "epoch": 0.695360686511129,
      "grad_norm": 0.3462477316840472,
      "learning_rate": 9.48038868648317e-06,
      "loss": 0.0351,
      "step": 2593
    },
    {
      "epoch": 0.6956288549208903,
      "grad_norm": 0.4053748803448256,
      "learning_rate": 9.479695905769813e-06,
      "loss": 0.0321,
      "step": 2594
    },
    {
      "epoch": 0.6958970233306516,
      "grad_norm": 0.6797731725806727,
      "learning_rate": 9.479002688880362e-06,
      "loss": 0.0286,
      "step": 2595
    },
    {
      "epoch": 0.696165191740413,
      "grad_norm": 0.4001893269467368,
      "learning_rate": 9.478309035882308e-06,
      "loss": 0.0435,
      "step": 2596
    },
    {
      "epoch": 0.6964333601501743,
      "grad_norm": 0.5297141001294885,
      "learning_rate": 9.477614946843194e-06,
      "loss": 0.0394,
      "step": 2597
    },
    {
      "epoch": 0.6967015285599356,
      "grad_norm": 0.5739944481572751,
      "learning_rate": 9.4769204218306e-06,
      "loss": 0.0376,
      "step": 2598
    },
    {
      "epoch": 0.696969696969697,
      "grad_norm": 0.37926399791468157,
      "learning_rate": 9.47622546091215e-06,
      "loss": 0.0385,
      "step": 2599
    },
    {
      "epoch": 0.6972378653794583,
      "grad_norm": 0.33394294276152053,
      "learning_rate": 9.475530064155512e-06,
      "loss": 0.0293,
      "step": 2600
    },
    {
      "epoch": 0.6975060337892196,
      "grad_norm": 0.4481808270233917,
      "learning_rate": 9.474834231628394e-06,
      "loss": 0.0437,
      "step": 2601
    },
    {
      "epoch": 0.697774202198981,
      "grad_norm": 0.4214258811414788,
      "learning_rate": 9.474137963398546e-06,
      "loss": 0.0348,
      "step": 2602
    },
    {
      "epoch": 0.6980423706087423,
      "grad_norm": 0.7041863271410196,
      "learning_rate": 9.473441259533765e-06,
      "loss": 0.0533,
      "step": 2603
    },
    {
      "epoch": 0.6983105390185036,
      "grad_norm": 0.45993421469224777,
      "learning_rate": 9.472744120101884e-06,
      "loss": 0.0574,
      "step": 2604
    },
    {
      "epoch": 0.698578707428265,
      "grad_norm": 0.48370580541406893,
      "learning_rate": 9.472046545170784e-06,
      "loss": 0.0362,
      "step": 2605
    },
    {
      "epoch": 0.6988468758380263,
      "grad_norm": 0.5008560265974089,
      "learning_rate": 9.471348534808385e-06,
      "loss": 0.0408,
      "step": 2606
    },
    {
      "epoch": 0.6991150442477876,
      "grad_norm": 0.4458965308879168,
      "learning_rate": 9.470650089082649e-06,
      "loss": 0.029,
      "step": 2607
    },
    {
      "epoch": 0.699383212657549,
      "grad_norm": 0.3500305270414859,
      "learning_rate": 9.469951208061584e-06,
      "loss": 0.0386,
      "step": 2608
    },
    {
      "epoch": 0.6996513810673103,
      "grad_norm": 0.36297712015513944,
      "learning_rate": 9.469251891813238e-06,
      "loss": 0.0332,
      "step": 2609
    },
    {
      "epoch": 0.6999195494770716,
      "grad_norm": 0.4247977316549169,
      "learning_rate": 9.4685521404057e-06,
      "loss": 0.0324,
      "step": 2610
    },
    {
      "epoch": 0.700187717886833,
      "grad_norm": 0.6117997400671085,
      "learning_rate": 9.467851953907103e-06,
      "loss": 0.043,
      "step": 2611
    },
    {
      "epoch": 0.7004558862965943,
      "grad_norm": 0.3059310892043086,
      "learning_rate": 9.467151332385624e-06,
      "loss": 0.0312,
      "step": 2612
    },
    {
      "epoch": 0.7007240547063556,
      "grad_norm": 0.31896964285004903,
      "learning_rate": 9.466450275909481e-06,
      "loss": 0.0261,
      "step": 2613
    },
    {
      "epoch": 0.700992223116117,
      "grad_norm": 0.5542071309897022,
      "learning_rate": 9.465748784546932e-06,
      "loss": 0.043,
      "step": 2614
    },
    {
      "epoch": 0.7012603915258783,
      "grad_norm": 0.2990252918943228,
      "learning_rate": 9.46504685836628e-06,
      "loss": 0.0289,
      "step": 2615
    },
    {
      "epoch": 0.7015285599356396,
      "grad_norm": 0.44788446770739,
      "learning_rate": 9.46434449743587e-06,
      "loss": 0.0283,
      "step": 2616
    },
    {
      "epoch": 0.701796728345401,
      "grad_norm": 0.30622799118200966,
      "learning_rate": 9.46364170182409e-06,
      "loss": 0.0385,
      "step": 2617
    },
    {
      "epoch": 0.7020648967551623,
      "grad_norm": 0.3486997893903188,
      "learning_rate": 9.462938471599366e-06,
      "loss": 0.0429,
      "step": 2618
    },
    {
      "epoch": 0.7023330651649236,
      "grad_norm": 0.42453664983518774,
      "learning_rate": 9.462234806830172e-06,
      "loss": 0.0398,
      "step": 2619
    },
    {
      "epoch": 0.7026012335746848,
      "grad_norm": 0.5372338962115267,
      "learning_rate": 9.461530707585023e-06,
      "loss": 0.0435,
      "step": 2620
    },
    {
      "epoch": 0.7028694019844463,
      "grad_norm": 0.46297375941689933,
      "learning_rate": 9.460826173932475e-06,
      "loss": 0.057,
      "step": 2621
    },
    {
      "epoch": 0.7031375703942075,
      "grad_norm": 0.3846557121024482,
      "learning_rate": 9.460121205941125e-06,
      "loss": 0.0437,
      "step": 2622
    },
    {
      "epoch": 0.7034057388039688,
      "grad_norm": 0.289497351076226,
      "learning_rate": 9.459415803679613e-06,
      "loss": 0.0271,
      "step": 2623
    },
    {
      "epoch": 0.7036739072137302,
      "grad_norm": 0.3689500869142178,
      "learning_rate": 9.458709967216625e-06,
      "loss": 0.0393,
      "step": 2624
    },
    {
      "epoch": 0.7039420756234915,
      "grad_norm": 0.4058195670260156,
      "learning_rate": 9.458003696620885e-06,
      "loss": 0.0329,
      "step": 2625
    },
    {
      "epoch": 0.7042102440332528,
      "grad_norm": 0.37015538351102995,
      "learning_rate": 9.457296991961163e-06,
      "loss": 0.0357,
      "step": 2626
    },
    {
      "epoch": 0.7044784124430142,
      "grad_norm": 0.3943897197560825,
      "learning_rate": 9.456589853306263e-06,
      "loss": 0.031,
      "step": 2627
    },
    {
      "epoch": 0.7047465808527755,
      "grad_norm": 0.3557154907479563,
      "learning_rate": 9.455882280725045e-06,
      "loss": 0.0282,
      "step": 2628
    },
    {
      "epoch": 0.7050147492625368,
      "grad_norm": 0.31914531507869204,
      "learning_rate": 9.455174274286396e-06,
      "loss": 0.0344,
      "step": 2629
    },
    {
      "epoch": 0.7052829176722982,
      "grad_norm": 0.4410708893157704,
      "learning_rate": 9.454465834059257e-06,
      "loss": 0.0581,
      "step": 2630
    },
    {
      "epoch": 0.7055510860820595,
      "grad_norm": 0.3467163725262633,
      "learning_rate": 9.453756960112605e-06,
      "loss": 0.0347,
      "step": 2631
    },
    {
      "epoch": 0.7058192544918208,
      "grad_norm": 0.33880922694137566,
      "learning_rate": 9.453047652515466e-06,
      "loss": 0.034,
      "step": 2632
    },
    {
      "epoch": 0.7060874229015822,
      "grad_norm": 0.365933099742316,
      "learning_rate": 9.452337911336896e-06,
      "loss": 0.0338,
      "step": 2633
    },
    {
      "epoch": 0.7063555913113435,
      "grad_norm": 0.32131246842023076,
      "learning_rate": 9.451627736646006e-06,
      "loss": 0.0352,
      "step": 2634
    },
    {
      "epoch": 0.7066237597211048,
      "grad_norm": 0.3735523107297418,
      "learning_rate": 9.450917128511942e-06,
      "loss": 0.0275,
      "step": 2635
    },
    {
      "epoch": 0.7068919281308662,
      "grad_norm": 0.4353006311922661,
      "learning_rate": 9.450206087003895e-06,
      "loss": 0.0397,
      "step": 2636
    },
    {
      "epoch": 0.7071600965406275,
      "grad_norm": 0.5263977724241677,
      "learning_rate": 9.449494612191094e-06,
      "loss": 0.044,
      "step": 2637
    },
    {
      "epoch": 0.7074282649503888,
      "grad_norm": 0.6886391063866727,
      "learning_rate": 9.448782704142818e-06,
      "loss": 0.0623,
      "step": 2638
    },
    {
      "epoch": 0.7076964333601502,
      "grad_norm": 0.24988749434997662,
      "learning_rate": 9.44807036292838e-06,
      "loss": 0.0255,
      "step": 2639
    },
    {
      "epoch": 0.7079646017699115,
      "grad_norm": 0.563986417897695,
      "learning_rate": 9.447357588617142e-06,
      "loss": 0.0387,
      "step": 2640
    },
    {
      "epoch": 0.7082327701796728,
      "grad_norm": 0.38306834556808345,
      "learning_rate": 9.4466443812785e-06,
      "loss": 0.0453,
      "step": 2641
    },
    {
      "epoch": 0.7085009385894342,
      "grad_norm": 0.3083146974228588,
      "learning_rate": 9.445930740981904e-06,
      "loss": 0.0358,
      "step": 2642
    },
    {
      "epoch": 0.7087691069991955,
      "grad_norm": 0.31657131910368547,
      "learning_rate": 9.445216667796833e-06,
      "loss": 0.042,
      "step": 2643
    },
    {
      "epoch": 0.7090372754089568,
      "grad_norm": 0.3983298516828245,
      "learning_rate": 9.444502161792817e-06,
      "loss": 0.0382,
      "step": 2644
    },
    {
      "epoch": 0.7093054438187182,
      "grad_norm": 0.257400812724172,
      "learning_rate": 9.443787223039425e-06,
      "loss": 0.0208,
      "step": 2645
    },
    {
      "epoch": 0.7095736122284795,
      "grad_norm": 0.4662173454064492,
      "learning_rate": 9.443071851606271e-06,
      "loss": 0.0307,
      "step": 2646
    },
    {
      "epoch": 0.7098417806382408,
      "grad_norm": 0.3823289015294413,
      "learning_rate": 9.442356047563007e-06,
      "loss": 0.0322,
      "step": 2647
    },
    {
      "epoch": 0.7101099490480022,
      "grad_norm": 0.45729757992449044,
      "learning_rate": 9.441639810979327e-06,
      "loss": 0.0396,
      "step": 2648
    },
    {
      "epoch": 0.7103781174577635,
      "grad_norm": 0.4920484398909539,
      "learning_rate": 9.44092314192497e-06,
      "loss": 0.0368,
      "step": 2649
    },
    {
      "epoch": 0.7106462858675248,
      "grad_norm": 0.42011224625710414,
      "learning_rate": 9.44020604046972e-06,
      "loss": 0.0409,
      "step": 2650
    },
    {
      "epoch": 0.7109144542772862,
      "grad_norm": 0.4732314867817539,
      "learning_rate": 9.439488506683393e-06,
      "loss": 0.0423,
      "step": 2651
    },
    {
      "epoch": 0.7111826226870475,
      "grad_norm": 0.4227496017019246,
      "learning_rate": 9.43877054063586e-06,
      "loss": 0.0448,
      "step": 2652
    },
    {
      "epoch": 0.7114507910968088,
      "grad_norm": 0.4386308568902799,
      "learning_rate": 9.438052142397024e-06,
      "loss": 0.0336,
      "step": 2653
    },
    {
      "epoch": 0.7117189595065702,
      "grad_norm": 0.4675238932937317,
      "learning_rate": 9.437333312036832e-06,
      "loss": 0.0434,
      "step": 2654
    },
    {
      "epoch": 0.7119871279163315,
      "grad_norm": 0.30064764201401006,
      "learning_rate": 9.436614049625277e-06,
      "loss": 0.034,
      "step": 2655
    },
    {
      "epoch": 0.7122552963260927,
      "grad_norm": 0.3873299831428463,
      "learning_rate": 9.435894355232392e-06,
      "loss": 0.0476,
      "step": 2656
    },
    {
      "epoch": 0.7125234647358542,
      "grad_norm": 0.3059157820894288,
      "learning_rate": 9.435174228928248e-06,
      "loss": 0.0312,
      "step": 2657
    },
    {
      "epoch": 0.7127916331456154,
      "grad_norm": 0.6277293936392747,
      "learning_rate": 9.434453670782968e-06,
      "loss": 0.0529,
      "step": 2658
    },
    {
      "epoch": 0.7130598015553767,
      "grad_norm": 0.29561977553876284,
      "learning_rate": 9.433732680866704e-06,
      "loss": 0.0237,
      "step": 2659
    },
    {
      "epoch": 0.7133279699651381,
      "grad_norm": 0.40827700251984134,
      "learning_rate": 9.433011259249663e-06,
      "loss": 0.0462,
      "step": 2660
    },
    {
      "epoch": 0.7135961383748994,
      "grad_norm": 0.39495441024244804,
      "learning_rate": 9.432289406002085e-06,
      "loss": 0.0413,
      "step": 2661
    },
    {
      "epoch": 0.7138643067846607,
      "grad_norm": 0.5363938821784745,
      "learning_rate": 9.431567121194251e-06,
      "loss": 0.0403,
      "step": 2662
    },
    {
      "epoch": 0.7141324751944221,
      "grad_norm": 0.31257287055808247,
      "learning_rate": 9.430844404896496e-06,
      "loss": 0.0248,
      "step": 2663
    },
    {
      "epoch": 0.7144006436041834,
      "grad_norm": 0.832348316463663,
      "learning_rate": 9.430121257179183e-06,
      "loss": 0.0526,
      "step": 2664
    },
    {
      "epoch": 0.7146688120139447,
      "grad_norm": 0.506238721951936,
      "learning_rate": 9.429397678112726e-06,
      "loss": 0.0421,
      "step": 2665
    },
    {
      "epoch": 0.7149369804237061,
      "grad_norm": 0.387817490729793,
      "learning_rate": 9.428673667767577e-06,
      "loss": 0.0368,
      "step": 2666
    },
    {
      "epoch": 0.7152051488334674,
      "grad_norm": 0.3816904342879657,
      "learning_rate": 9.42794922621423e-06,
      "loss": 0.0344,
      "step": 2667
    },
    {
      "epoch": 0.7154733172432287,
      "grad_norm": 0.32147696479265375,
      "learning_rate": 9.427224353523223e-06,
      "loss": 0.0248,
      "step": 2668
    },
    {
      "epoch": 0.7157414856529901,
      "grad_norm": 0.36970052678357146,
      "learning_rate": 9.426499049765136e-06,
      "loss": 0.0358,
      "step": 2669
    },
    {
      "epoch": 0.7160096540627514,
      "grad_norm": 0.3789213281709949,
      "learning_rate": 9.425773315010587e-06,
      "loss": 0.0331,
      "step": 2670
    },
    {
      "epoch": 0.7162778224725127,
      "grad_norm": 0.5757118573519422,
      "learning_rate": 9.425047149330242e-06,
      "loss": 0.0581,
      "step": 2671
    },
    {
      "epoch": 0.7165459908822741,
      "grad_norm": 0.3777858458474212,
      "learning_rate": 9.424320552794803e-06,
      "loss": 0.0475,
      "step": 2672
    },
    {
      "epoch": 0.7168141592920354,
      "grad_norm": 0.32835603840106786,
      "learning_rate": 9.423593525475019e-06,
      "loss": 0.0329,
      "step": 2673
    },
    {
      "epoch": 0.7170823277017967,
      "grad_norm": 0.43375540459549766,
      "learning_rate": 9.422866067441678e-06,
      "loss": 0.0335,
      "step": 2674
    },
    {
      "epoch": 0.7173504961115581,
      "grad_norm": 0.5776245092895724,
      "learning_rate": 9.42213817876561e-06,
      "loss": 0.0457,
      "step": 2675
    },
    {
      "epoch": 0.7176186645213194,
      "grad_norm": 0.3716675522805873,
      "learning_rate": 9.421409859517688e-06,
      "loss": 0.0347,
      "step": 2676
    },
    {
      "epoch": 0.7178868329310807,
      "grad_norm": 0.32693660346552744,
      "learning_rate": 9.420681109768827e-06,
      "loss": 0.0306,
      "step": 2677
    },
    {
      "epoch": 0.7181550013408421,
      "grad_norm": 0.54620231822749,
      "learning_rate": 9.419951929589983e-06,
      "loss": 0.036,
      "step": 2678
    },
    {
      "epoch": 0.7184231697506034,
      "grad_norm": 0.34161193262352446,
      "learning_rate": 9.419222319052154e-06,
      "loss": 0.035,
      "step": 2679
    },
    {
      "epoch": 0.7186913381603647,
      "grad_norm": 0.49071864076508265,
      "learning_rate": 9.41849227822638e-06,
      "loss": 0.0469,
      "step": 2680
    },
    {
      "epoch": 0.7189595065701261,
      "grad_norm": 0.47220287395251737,
      "learning_rate": 9.417761807183745e-06,
      "loss": 0.0351,
      "step": 2681
    },
    {
      "epoch": 0.7192276749798874,
      "grad_norm": 0.346906855876665,
      "learning_rate": 9.417030905995372e-06,
      "loss": 0.0371,
      "step": 2682
    },
    {
      "epoch": 0.7194958433896487,
      "grad_norm": 0.36879488049620507,
      "learning_rate": 9.416299574732425e-06,
      "loss": 0.0388,
      "step": 2683
    },
    {
      "epoch": 0.7197640117994101,
      "grad_norm": 0.41172394744974267,
      "learning_rate": 9.415567813466114e-06,
      "loss": 0.0361,
      "step": 2684
    },
    {
      "epoch": 0.7200321802091714,
      "grad_norm": 0.4304818958839077,
      "learning_rate": 9.414835622267689e-06,
      "loss": 0.0278,
      "step": 2685
    },
    {
      "epoch": 0.7203003486189327,
      "grad_norm": 0.2956703087356682,
      "learning_rate": 9.41410300120844e-06,
      "loss": 0.0348,
      "step": 2686
    },
    {
      "epoch": 0.7205685170286941,
      "grad_norm": 0.28843451323726627,
      "learning_rate": 9.4133699503597e-06,
      "loss": 0.029,
      "step": 2687
    },
    {
      "epoch": 0.7208366854384554,
      "grad_norm": 0.46630114862955235,
      "learning_rate": 9.412636469792847e-06,
      "loss": 0.0347,
      "step": 2688
    },
    {
      "epoch": 0.7211048538482167,
      "grad_norm": 0.3173721013535111,
      "learning_rate": 9.411902559579295e-06,
      "loss": 0.0287,
      "step": 2689
    },
    {
      "epoch": 0.7213730222579781,
      "grad_norm": 0.5303554957669326,
      "learning_rate": 9.411168219790506e-06,
      "loss": 0.0403,
      "step": 2690
    },
    {
      "epoch": 0.7216411906677394,
      "grad_norm": 0.27369982832533374,
      "learning_rate": 9.410433450497977e-06,
      "loss": 0.0335,
      "step": 2691
    },
    {
      "epoch": 0.7219093590775006,
      "grad_norm": 0.5513954865903511,
      "learning_rate": 9.409698251773254e-06,
      "loss": 0.038,
      "step": 2692
    },
    {
      "epoch": 0.722177527487262,
      "grad_norm": 0.3215182094325023,
      "learning_rate": 9.408962623687918e-06,
      "loss": 0.0293,
      "step": 2693
    },
    {
      "epoch": 0.7224456958970233,
      "grad_norm": 0.4354235663623081,
      "learning_rate": 9.408226566313598e-06,
      "loss": 0.034,
      "step": 2694
    },
    {
      "epoch": 0.7227138643067846,
      "grad_norm": 0.45467759149282094,
      "learning_rate": 9.407490079721959e-06,
      "loss": 0.0387,
      "step": 2695
    },
    {
      "epoch": 0.722982032716546,
      "grad_norm": 0.3572642249436466,
      "learning_rate": 9.406753163984714e-06,
      "loss": 0.0359,
      "step": 2696
    },
    {
      "epoch": 0.7232502011263073,
      "grad_norm": 0.43167678135881854,
      "learning_rate": 9.406015819173614e-06,
      "loss": 0.0266,
      "step": 2697
    },
    {
      "epoch": 0.7235183695360686,
      "grad_norm": 0.22929703376388086,
      "learning_rate": 9.40527804536045e-06,
      "loss": 0.0271,
      "step": 2698
    },
    {
      "epoch": 0.72378653794583,
      "grad_norm": 0.4762434226247083,
      "learning_rate": 9.404539842617059e-06,
      "loss": 0.0353,
      "step": 2699
    },
    {
      "epoch": 0.7240547063555913,
      "grad_norm": 0.43229494341064917,
      "learning_rate": 9.403801211015317e-06,
      "loss": 0.0342,
      "step": 2700
    },
    {
      "epoch": 0.7243228747653526,
      "grad_norm": 0.5582270287177813,
      "learning_rate": 9.403062150627145e-06,
      "loss": 0.0429,
      "step": 2701
    },
    {
      "epoch": 0.724591043175114,
      "grad_norm": 0.43457849315388136,
      "learning_rate": 9.402322661524499e-06,
      "loss": 0.0333,
      "step": 2702
    },
    {
      "epoch": 0.7248592115848753,
      "grad_norm": 0.47674628828301996,
      "learning_rate": 9.401582743779384e-06,
      "loss": 0.0384,
      "step": 2703
    },
    {
      "epoch": 0.7251273799946366,
      "grad_norm": 0.4836546715157746,
      "learning_rate": 9.400842397463842e-06,
      "loss": 0.0311,
      "step": 2704
    },
    {
      "epoch": 0.725395548404398,
      "grad_norm": 0.35911375326587447,
      "learning_rate": 9.400101622649962e-06,
      "loss": 0.0445,
      "step": 2705
    },
    {
      "epoch": 0.7256637168141593,
      "grad_norm": 0.4726544443293054,
      "learning_rate": 9.399360419409869e-06,
      "loss": 0.0358,
      "step": 2706
    },
    {
      "epoch": 0.7259318852239206,
      "grad_norm": 0.4743025316039749,
      "learning_rate": 9.398618787815731e-06,
      "loss": 0.0387,
      "step": 2707
    },
    {
      "epoch": 0.7262000536336819,
      "grad_norm": 0.6972954834740794,
      "learning_rate": 9.39787672793976e-06,
      "loss": 0.0411,
      "step": 2708
    },
    {
      "epoch": 0.7264682220434433,
      "grad_norm": 1.2269764745242044,
      "learning_rate": 9.39713423985421e-06,
      "loss": 0.0327,
      "step": 2709
    },
    {
      "epoch": 0.7267363904532046,
      "grad_norm": 0.37503187748723465,
      "learning_rate": 9.39639132363137e-06,
      "loss": 0.0326,
      "step": 2710
    },
    {
      "epoch": 0.7270045588629659,
      "grad_norm": 0.5397512962600839,
      "learning_rate": 9.39564797934358e-06,
      "loss": 0.0535,
      "step": 2711
    },
    {
      "epoch": 0.7272727272727273,
      "grad_norm": 0.31694373979239815,
      "learning_rate": 9.394904207063216e-06,
      "loss": 0.0289,
      "step": 2712
    },
    {
      "epoch": 0.7275408956824886,
      "grad_norm": 0.5704152848423665,
      "learning_rate": 9.394160006862698e-06,
      "loss": 0.0351,
      "step": 2713
    },
    {
      "epoch": 0.7278090640922499,
      "grad_norm": 0.3729889250766108,
      "learning_rate": 9.393415378814488e-06,
      "loss": 0.0376,
      "step": 2714
    },
    {
      "epoch": 0.7280772325020113,
      "grad_norm": 0.4198426567186146,
      "learning_rate": 9.392670322991085e-06,
      "loss": 0.0388,
      "step": 2715
    },
    {
      "epoch": 0.7283454009117726,
      "grad_norm": 0.3223314182164308,
      "learning_rate": 9.391924839465036e-06,
      "loss": 0.0296,
      "step": 2716
    },
    {
      "epoch": 0.7286135693215339,
      "grad_norm": 0.7235493059670693,
      "learning_rate": 9.391178928308925e-06,
      "loss": 0.0302,
      "step": 2717
    },
    {
      "epoch": 0.7288817377312953,
      "grad_norm": 0.3672472666522542,
      "learning_rate": 9.390432589595379e-06,
      "loss": 0.0407,
      "step": 2718
    },
    {
      "epoch": 0.7291499061410566,
      "grad_norm": 0.20448967956359898,
      "learning_rate": 9.38968582339707e-06,
      "loss": 0.0253,
      "step": 2719
    },
    {
      "epoch": 0.7294180745508179,
      "grad_norm": 0.3373537464993408,
      "learning_rate": 9.388938629786708e-06,
      "loss": 0.0405,
      "step": 2720
    },
    {
      "epoch": 0.7296862429605793,
      "grad_norm": 0.3108173757712986,
      "learning_rate": 9.388191008837042e-06,
      "loss": 0.0349,
      "step": 2721
    },
    {
      "epoch": 0.7299544113703406,
      "grad_norm": 0.518795718010647,
      "learning_rate": 9.387442960620868e-06,
      "loss": 0.0353,
      "step": 2722
    },
    {
      "epoch": 0.7302225797801019,
      "grad_norm": 0.4120483815959848,
      "learning_rate": 9.386694485211021e-06,
      "loss": 0.0343,
      "step": 2723
    },
    {
      "epoch": 0.7304907481898633,
      "grad_norm": 0.6108624511819352,
      "learning_rate": 9.385945582680379e-06,
      "loss": 0.0408,
      "step": 2724
    },
    {
      "epoch": 0.7307589165996246,
      "grad_norm": 0.4656012118755314,
      "learning_rate": 9.385196253101861e-06,
      "loss": 0.0462,
      "step": 2725
    },
    {
      "epoch": 0.7310270850093858,
      "grad_norm": 0.28342164680600157,
      "learning_rate": 9.384446496548425e-06,
      "loss": 0.0267,
      "step": 2726
    },
    {
      "epoch": 0.7312952534191473,
      "grad_norm": 0.264307583114337,
      "learning_rate": 9.383696313093073e-06,
      "loss": 0.027,
      "step": 2727
    },
    {
      "epoch": 0.7315634218289085,
      "grad_norm": 0.6074153635124294,
      "learning_rate": 9.382945702808852e-06,
      "loss": 0.0429,
      "step": 2728
    },
    {
      "epoch": 0.7318315902386698,
      "grad_norm": 0.5174273502949047,
      "learning_rate": 9.382194665768844e-06,
      "loss": 0.051,
      "step": 2729
    },
    {
      "epoch": 0.7320997586484312,
      "grad_norm": 0.6265495632451057,
      "learning_rate": 9.381443202046175e-06,
      "loss": 0.0358,
      "step": 2730
    },
    {
      "epoch": 0.7323679270581925,
      "grad_norm": 0.32620497314101726,
      "learning_rate": 9.380691311714013e-06,
      "loss": 0.0336,
      "step": 2731
    },
    {
      "epoch": 0.7326360954679538,
      "grad_norm": 0.35064993515706444,
      "learning_rate": 9.37993899484557e-06,
      "loss": 0.0257,
      "step": 2732
    },
    {
      "epoch": 0.7329042638777152,
      "grad_norm": 0.4175418038425128,
      "learning_rate": 9.379186251514096e-06,
      "loss": 0.039,
      "step": 2733
    },
    {
      "epoch": 0.7331724322874765,
      "grad_norm": 0.6179551406430485,
      "learning_rate": 9.378433081792883e-06,
      "loss": 0.0471,
      "step": 2734
    },
    {
      "epoch": 0.7334406006972378,
      "grad_norm": 0.4189199048181353,
      "learning_rate": 9.377679485755265e-06,
      "loss": 0.03,
      "step": 2735
    },
    {
      "epoch": 0.7337087691069992,
      "grad_norm": 0.334562590479166,
      "learning_rate": 9.376925463474618e-06,
      "loss": 0.0312,
      "step": 2736
    },
    {
      "epoch": 0.7339769375167605,
      "grad_norm": 0.33658092472066803,
      "learning_rate": 9.376171015024358e-06,
      "loss": 0.0358,
      "step": 2737
    },
    {
      "epoch": 0.7342451059265218,
      "grad_norm": 0.292881163076582,
      "learning_rate": 9.375416140477945e-06,
      "loss": 0.0353,
      "step": 2738
    },
    {
      "epoch": 0.7345132743362832,
      "grad_norm": 0.4485549853790626,
      "learning_rate": 9.374660839908881e-06,
      "loss": 0.0434,
      "step": 2739
    },
    {
      "epoch": 0.7347814427460445,
      "grad_norm": 0.4348298214180735,
      "learning_rate": 9.373905113390704e-06,
      "loss": 0.0377,
      "step": 2740
    },
    {
      "epoch": 0.7350496111558058,
      "grad_norm": 0.34086553822572935,
      "learning_rate": 9.373148960996998e-06,
      "loss": 0.0316,
      "step": 2741
    },
    {
      "epoch": 0.7353177795655672,
      "grad_norm": 0.45340390928171354,
      "learning_rate": 9.372392382801389e-06,
      "loss": 0.042,
      "step": 2742
    },
    {
      "epoch": 0.7355859479753285,
      "grad_norm": 0.41290457730171065,
      "learning_rate": 9.371635378877542e-06,
      "loss": 0.0387,
      "step": 2743
    },
    {
      "epoch": 0.7358541163850898,
      "grad_norm": 0.6464675423412379,
      "learning_rate": 9.370877949299164e-06,
      "loss": 0.0339,
      "step": 2744
    },
    {
      "epoch": 0.7361222847948512,
      "grad_norm": 0.34479109473592995,
      "learning_rate": 9.370120094140006e-06,
      "loss": 0.0278,
      "step": 2745
    },
    {
      "epoch": 0.7363904532046125,
      "grad_norm": 0.3870503407858689,
      "learning_rate": 9.369361813473856e-06,
      "loss": 0.0437,
      "step": 2746
    },
    {
      "epoch": 0.7366586216143738,
      "grad_norm": 0.38155919561310914,
      "learning_rate": 9.368603107374547e-06,
      "loss": 0.0394,
      "step": 2747
    },
    {
      "epoch": 0.7369267900241352,
      "grad_norm": 0.26727591386995,
      "learning_rate": 9.36784397591595e-06,
      "loss": 0.0359,
      "step": 2748
    },
    {
      "epoch": 0.7371949584338965,
      "grad_norm": 0.5343887998039686,
      "learning_rate": 9.367084419171984e-06,
      "loss": 0.0477,
      "step": 2749
    },
    {
      "epoch": 0.7374631268436578,
      "grad_norm": 0.3355150329619468,
      "learning_rate": 9.366324437216602e-06,
      "loss": 0.0309,
      "step": 2750
    },
    {
      "epoch": 0.7377312952534192,
      "grad_norm": 0.5286632485501943,
      "learning_rate": 9.365564030123802e-06,
      "loss": 0.0492,
      "step": 2751
    },
    {
      "epoch": 0.7379994636631805,
      "grad_norm": 0.314639637029734,
      "learning_rate": 9.364803197967623e-06,
      "loss": 0.0311,
      "step": 2752
    },
    {
      "epoch": 0.7382676320729418,
      "grad_norm": 0.4540144994912781,
      "learning_rate": 9.364041940822143e-06,
      "loss": 0.0383,
      "step": 2753
    },
    {
      "epoch": 0.7385358004827032,
      "grad_norm": 0.4032548200009073,
      "learning_rate": 9.363280258761487e-06,
      "loss": 0.0382,
      "step": 2754
    },
    {
      "epoch": 0.7388039688924645,
      "grad_norm": 0.5072680538948661,
      "learning_rate": 9.362518151859816e-06,
      "loss": 0.0418,
      "step": 2755
    },
    {
      "epoch": 0.7390721373022258,
      "grad_norm": 0.3879804826519922,
      "learning_rate": 9.361755620191335e-06,
      "loss": 0.031,
      "step": 2756
    },
    {
      "epoch": 0.7393403057119872,
      "grad_norm": 0.34329708817200605,
      "learning_rate": 9.360992663830289e-06,
      "loss": 0.0335,
      "step": 2757
    },
    {
      "epoch": 0.7396084741217485,
      "grad_norm": 0.35043694286614124,
      "learning_rate": 9.360229282850966e-06,
      "loss": 0.0374,
      "step": 2758
    },
    {
      "epoch": 0.7398766425315098,
      "grad_norm": 0.328290188830268,
      "learning_rate": 9.359465477327694e-06,
      "loss": 0.0373,
      "step": 2759
    },
    {
      "epoch": 0.7401448109412712,
      "grad_norm": 0.2892081286213167,
      "learning_rate": 9.358701247334842e-06,
      "loss": 0.037,
      "step": 2760
    },
    {
      "epoch": 0.7404129793510325,
      "grad_norm": 0.4259384745948188,
      "learning_rate": 9.357936592946823e-06,
      "loss": 0.0403,
      "step": 2761
    },
    {
      "epoch": 0.7406811477607937,
      "grad_norm": 0.33539921631282366,
      "learning_rate": 9.357171514238086e-06,
      "loss": 0.0348,
      "step": 2762
    },
    {
      "epoch": 0.7409493161705552,
      "grad_norm": 0.4534777268325014,
      "learning_rate": 9.356406011283128e-06,
      "loss": 0.0339,
      "step": 2763
    },
    {
      "epoch": 0.7412174845803164,
      "grad_norm": 0.4445702179297352,
      "learning_rate": 9.355640084156483e-06,
      "loss": 0.0346,
      "step": 2764
    },
    {
      "epoch": 0.7414856529900777,
      "grad_norm": 0.4973435423297838,
      "learning_rate": 9.354873732932727e-06,
      "loss": 0.0464,
      "step": 2765
    },
    {
      "epoch": 0.7417538213998391,
      "grad_norm": 0.46215667235157587,
      "learning_rate": 9.354106957686478e-06,
      "loss": 0.0325,
      "step": 2766
    },
    {
      "epoch": 0.7420219898096004,
      "grad_norm": 0.3586736475885427,
      "learning_rate": 9.353339758492394e-06,
      "loss": 0.0391,
      "step": 2767
    },
    {
      "epoch": 0.7422901582193617,
      "grad_norm": 0.35818587554594095,
      "learning_rate": 9.352572135425175e-06,
      "loss": 0.0289,
      "step": 2768
    },
    {
      "epoch": 0.7425583266291231,
      "grad_norm": 0.3472660612561635,
      "learning_rate": 9.351804088559565e-06,
      "loss": 0.0387,
      "step": 2769
    },
    {
      "epoch": 0.7428264950388844,
      "grad_norm": 0.38512994460302025,
      "learning_rate": 9.35103561797034e-06,
      "loss": 0.0315,
      "step": 2770
    },
    {
      "epoch": 0.7430946634486457,
      "grad_norm": 0.32312852819656157,
      "learning_rate": 9.350266723732333e-06,
      "loss": 0.0377,
      "step": 2771
    },
    {
      "epoch": 0.7433628318584071,
      "grad_norm": 0.3025608674208243,
      "learning_rate": 9.349497405920405e-06,
      "loss": 0.0297,
      "step": 2772
    },
    {
      "epoch": 0.7436310002681684,
      "grad_norm": 0.3983460176911991,
      "learning_rate": 9.34872766460946e-06,
      "loss": 0.0483,
      "step": 2773
    },
    {
      "epoch": 0.7438991686779297,
      "grad_norm": 0.36466344321223576,
      "learning_rate": 9.347957499874449e-06,
      "loss": 0.0463,
      "step": 2774
    },
    {
      "epoch": 0.7441673370876911,
      "grad_norm": 0.35713408458877244,
      "learning_rate": 9.34718691179036e-06,
      "loss": 0.0324,
      "step": 2775
    },
    {
      "epoch": 0.7444355054974524,
      "grad_norm": 0.45480867301502875,
      "learning_rate": 9.346415900432222e-06,
      "loss": 0.039,
      "step": 2776
    },
    {
      "epoch": 0.7447036739072137,
      "grad_norm": 0.3379154057008416,
      "learning_rate": 9.345644465875108e-06,
      "loss": 0.0386,
      "step": 2777
    },
    {
      "epoch": 0.7449718423169751,
      "grad_norm": 0.3398445033643358,
      "learning_rate": 9.34487260819413e-06,
      "loss": 0.0436,
      "step": 2778
    },
    {
      "epoch": 0.7452400107267364,
      "grad_norm": 0.5005377681793879,
      "learning_rate": 9.344100327464441e-06,
      "loss": 0.0589,
      "step": 2779
    },
    {
      "epoch": 0.7455081791364977,
      "grad_norm": 0.31331370622173227,
      "learning_rate": 9.343327623761237e-06,
      "loss": 0.0354,
      "step": 2780
    },
    {
      "epoch": 0.7457763475462591,
      "grad_norm": 0.31095828425191324,
      "learning_rate": 9.342554497159753e-06,
      "loss": 0.039,
      "step": 2781
    },
    {
      "epoch": 0.7460445159560204,
      "grad_norm": 0.2672237072998254,
      "learning_rate": 9.34178094773527e-06,
      "loss": 0.0282,
      "step": 2782
    },
    {
      "epoch": 0.7463126843657817,
      "grad_norm": 0.34134367428360257,
      "learning_rate": 9.341006975563098e-06,
      "loss": 0.0309,
      "step": 2783
    },
    {
      "epoch": 0.7465808527755431,
      "grad_norm": 0.44398683839981,
      "learning_rate": 9.340232580718604e-06,
      "loss": 0.032,
      "step": 2784
    },
    {
      "epoch": 0.7468490211853044,
      "grad_norm": 0.29778815373743833,
      "learning_rate": 9.339457763277188e-06,
      "loss": 0.0317,
      "step": 2785
    },
    {
      "epoch": 0.7471171895950657,
      "grad_norm": 0.4981136466879834,
      "learning_rate": 9.33868252331429e-06,
      "loss": 0.0391,
      "step": 2786
    },
    {
      "epoch": 0.7473853580048271,
      "grad_norm": 0.6912211320297063,
      "learning_rate": 9.337906860905394e-06,
      "loss": 0.0389,
      "step": 2787
    },
    {
      "epoch": 0.7476535264145884,
      "grad_norm": 0.46057263005209936,
      "learning_rate": 9.337130776126021e-06,
      "loss": 0.0391,
      "step": 2788
    },
    {
      "epoch": 0.7479216948243497,
      "grad_norm": 0.240339964203105,
      "learning_rate": 9.336354269051744e-06,
      "loss": 0.0274,
      "step": 2789
    },
    {
      "epoch": 0.7481898632341111,
      "grad_norm": 0.34925017912672046,
      "learning_rate": 9.335577339758162e-06,
      "loss": 0.0322,
      "step": 2790
    },
    {
      "epoch": 0.7484580316438724,
      "grad_norm": 0.3467047405252644,
      "learning_rate": 9.334799988320926e-06,
      "loss": 0.0254,
      "step": 2791
    },
    {
      "epoch": 0.7487262000536337,
      "grad_norm": 0.4209284572408516,
      "learning_rate": 9.334022214815722e-06,
      "loss": 0.0309,
      "step": 2792
    },
    {
      "epoch": 0.7489943684633951,
      "grad_norm": 0.32821203004969746,
      "learning_rate": 9.333244019318283e-06,
      "loss": 0.0404,
      "step": 2793
    },
    {
      "epoch": 0.7492625368731564,
      "grad_norm": 0.28862734723977007,
      "learning_rate": 9.332465401904379e-06,
      "loss": 0.0255,
      "step": 2794
    },
    {
      "epoch": 0.7495307052829177,
      "grad_norm": 0.36497909507937576,
      "learning_rate": 9.331686362649819e-06,
      "loss": 0.04,
      "step": 2795
    },
    {
      "epoch": 0.7497988736926791,
      "grad_norm": 0.4421134454525723,
      "learning_rate": 9.33090690163046e-06,
      "loss": 0.0481,
      "step": 2796
    },
    {
      "epoch": 0.7500670421024404,
      "grad_norm": 0.40687626672383576,
      "learning_rate": 9.330127018922195e-06,
      "loss": 0.0422,
      "step": 2797
    },
    {
      "epoch": 0.7503352105122016,
      "grad_norm": 0.3488669750301508,
      "learning_rate": 9.329346714600957e-06,
      "loss": 0.0323,
      "step": 2798
    },
    {
      "epoch": 0.7506033789219629,
      "grad_norm": 0.4117735162237764,
      "learning_rate": 9.328565988742723e-06,
      "loss": 0.0429,
      "step": 2799
    },
    {
      "epoch": 0.7508715473317243,
      "grad_norm": 0.30377496881316224,
      "learning_rate": 9.327784841423512e-06,
      "loss": 0.0253,
      "step": 2800
    },
    {
      "epoch": 0.7511397157414856,
      "grad_norm": 0.38558397181756543,
      "learning_rate": 9.327003272719382e-06,
      "loss": 0.0385,
      "step": 2801
    },
    {
      "epoch": 0.7514078841512469,
      "grad_norm": 0.5658615480261258,
      "learning_rate": 9.32622128270643e-06,
      "loss": 0.0391,
      "step": 2802
    },
    {
      "epoch": 0.7516760525610083,
      "grad_norm": 0.38644785287456734,
      "learning_rate": 9.325438871460798e-06,
      "loss": 0.0402,
      "step": 2803
    },
    {
      "epoch": 0.7519442209707696,
      "grad_norm": 0.31169179554676274,
      "learning_rate": 9.324656039058666e-06,
      "loss": 0.0232,
      "step": 2804
    },
    {
      "epoch": 0.7522123893805309,
      "grad_norm": 0.4454915867544068,
      "learning_rate": 9.323872785576257e-06,
      "loss": 0.0421,
      "step": 2805
    },
    {
      "epoch": 0.7524805577902923,
      "grad_norm": 0.3671060804392625,
      "learning_rate": 9.323089111089835e-06,
      "loss": 0.0293,
      "step": 2806
    },
    {
      "epoch": 0.7527487262000536,
      "grad_norm": 0.2748163698812082,
      "learning_rate": 9.322305015675706e-06,
      "loss": 0.0308,
      "step": 2807
    },
    {
      "epoch": 0.7530168946098149,
      "grad_norm": 0.38387044672063336,
      "learning_rate": 9.32152049941021e-06,
      "loss": 0.0342,
      "step": 2808
    },
    {
      "epoch": 0.7532850630195763,
      "grad_norm": 0.35711667070284353,
      "learning_rate": 9.320735562369739e-06,
      "loss": 0.0386,
      "step": 2809
    },
    {
      "epoch": 0.7535532314293376,
      "grad_norm": 0.39585449874355755,
      "learning_rate": 9.319950204630716e-06,
      "loss": 0.0311,
      "step": 2810
    },
    {
      "epoch": 0.7538213998390989,
      "grad_norm": 0.2594219980914141,
      "learning_rate": 9.31916442626961e-06,
      "loss": 0.03,
      "step": 2811
    },
    {
      "epoch": 0.7540895682488603,
      "grad_norm": 0.29246613793869797,
      "learning_rate": 9.318378227362934e-06,
      "loss": 0.0295,
      "step": 2812
    },
    {
      "epoch": 0.7543577366586216,
      "grad_norm": 0.3386397756663265,
      "learning_rate": 9.317591607987233e-06,
      "loss": 0.0264,
      "step": 2813
    },
    {
      "epoch": 0.7546259050683829,
      "grad_norm": 0.5574377174150268,
      "learning_rate": 9.3168045682191e-06,
      "loss": 0.0345,
      "step": 2814
    },
    {
      "epoch": 0.7548940734781443,
      "grad_norm": 0.29644893730102756,
      "learning_rate": 9.316017108135167e-06,
      "loss": 0.0304,
      "step": 2815
    },
    {
      "epoch": 0.7551622418879056,
      "grad_norm": 0.30848909279110376,
      "learning_rate": 9.315229227812108e-06,
      "loss": 0.0323,
      "step": 2816
    },
    {
      "epoch": 0.7554304102976669,
      "grad_norm": 0.3686131786249299,
      "learning_rate": 9.314440927326635e-06,
      "loss": 0.0579,
      "step": 2817
    },
    {
      "epoch": 0.7556985787074283,
      "grad_norm": 0.3451032885483723,
      "learning_rate": 9.313652206755503e-06,
      "loss": 0.0387,
      "step": 2818
    },
    {
      "epoch": 0.7559667471171896,
      "grad_norm": 0.4316139119185835,
      "learning_rate": 9.31286306617551e-06,
      "loss": 0.0305,
      "step": 2819
    },
    {
      "epoch": 0.7562349155269509,
      "grad_norm": 0.3896054839897764,
      "learning_rate": 9.312073505663489e-06,
      "loss": 0.0331,
      "step": 2820
    },
    {
      "epoch": 0.7565030839367123,
      "grad_norm": 0.2630097832827242,
      "learning_rate": 9.31128352529632e-06,
      "loss": 0.0284,
      "step": 2821
    },
    {
      "epoch": 0.7567712523464736,
      "grad_norm": 0.4975251258421063,
      "learning_rate": 9.31049312515092e-06,
      "loss": 0.0375,
      "step": 2822
    },
    {
      "epoch": 0.7570394207562349,
      "grad_norm": 0.43725150047313127,
      "learning_rate": 9.30970230530425e-06,
      "loss": 0.0326,
      "step": 2823
    },
    {
      "epoch": 0.7573075891659963,
      "grad_norm": 0.4395012434604193,
      "learning_rate": 9.308911065833307e-06,
      "loss": 0.0349,
      "step": 2824
    },
    {
      "epoch": 0.7575757575757576,
      "grad_norm": 0.2948723105878521,
      "learning_rate": 9.308119406815132e-06,
      "loss": 0.0313,
      "step": 2825
    },
    {
      "epoch": 0.7578439259855189,
      "grad_norm": 0.32737556273358553,
      "learning_rate": 9.30732732832681e-06,
      "loss": 0.0306,
      "step": 2826
    },
    {
      "epoch": 0.7581120943952803,
      "grad_norm": 0.29765855399117036,
      "learning_rate": 9.306534830445462e-06,
      "loss": 0.0311,
      "step": 2827
    },
    {
      "epoch": 0.7583802628050416,
      "grad_norm": 0.43533171977478574,
      "learning_rate": 9.30574191324825e-06,
      "loss": 0.0392,
      "step": 2828
    },
    {
      "epoch": 0.7586484312148029,
      "grad_norm": 0.34123353046946786,
      "learning_rate": 9.30494857681238e-06,
      "loss": 0.0391,
      "step": 2829
    },
    {
      "epoch": 0.7589165996245643,
      "grad_norm": 0.28412177952417883,
      "learning_rate": 9.304154821215097e-06,
      "loss": 0.0256,
      "step": 2830
    },
    {
      "epoch": 0.7591847680343256,
      "grad_norm": 0.3952978871819059,
      "learning_rate": 9.303360646533686e-06,
      "loss": 0.0373,
      "step": 2831
    },
    {
      "epoch": 0.7594529364440868,
      "grad_norm": 0.31200664206565537,
      "learning_rate": 9.302566052845474e-06,
      "loss": 0.0347,
      "step": 2832
    },
    {
      "epoch": 0.7597211048538483,
      "grad_norm": 0.28754811845098593,
      "learning_rate": 9.30177104022783e-06,
      "loss": 0.0252,
      "step": 2833
    },
    {
      "epoch": 0.7599892732636095,
      "grad_norm": 0.3062300046971535,
      "learning_rate": 9.300975608758159e-06,
      "loss": 0.0258,
      "step": 2834
    },
    {
      "epoch": 0.7602574416733708,
      "grad_norm": 0.46971055749416557,
      "learning_rate": 9.300179758513912e-06,
      "loss": 0.0338,
      "step": 2835
    },
    {
      "epoch": 0.7605256100831322,
      "grad_norm": 0.44441042175321327,
      "learning_rate": 9.29938348957258e-06,
      "loss": 0.0441,
      "step": 2836
    },
    {
      "epoch": 0.7607937784928935,
      "grad_norm": 0.31396063959115716,
      "learning_rate": 9.29858680201169e-06,
      "loss": 0.0338,
      "step": 2837
    },
    {
      "epoch": 0.7610619469026548,
      "grad_norm": 0.4217592096793415,
      "learning_rate": 9.297789695908816e-06,
      "loss": 0.0251,
      "step": 2838
    },
    {
      "epoch": 0.7613301153124162,
      "grad_norm": 0.3895764660852341,
      "learning_rate": 9.296992171341573e-06,
      "loss": 0.0438,
      "step": 2839
    },
    {
      "epoch": 0.7615982837221775,
      "grad_norm": 0.5054486100633101,
      "learning_rate": 9.296194228387607e-06,
      "loss": 0.0303,
      "step": 2840
    },
    {
      "epoch": 0.7618664521319388,
      "grad_norm": 0.3170490172278557,
      "learning_rate": 9.295395867124617e-06,
      "loss": 0.0312,
      "step": 2841
    },
    {
      "epoch": 0.7621346205417002,
      "grad_norm": 0.3836661595493773,
      "learning_rate": 9.294597087630333e-06,
      "loss": 0.0505,
      "step": 2842
    },
    {
      "epoch": 0.7624027889514615,
      "grad_norm": 0.3457173638208979,
      "learning_rate": 9.293797889982535e-06,
      "loss": 0.028,
      "step": 2843
    },
    {
      "epoch": 0.7626709573612228,
      "grad_norm": 0.4499570372614781,
      "learning_rate": 9.292998274259035e-06,
      "loss": 0.0268,
      "step": 2844
    },
    {
      "epoch": 0.7629391257709842,
      "grad_norm": 0.3683629565418978,
      "learning_rate": 9.292198240537692e-06,
      "loss": 0.023,
      "step": 2845
    },
    {
      "epoch": 0.7632072941807455,
      "grad_norm": 0.2631422748647027,
      "learning_rate": 9.291397788896401e-06,
      "loss": 0.0211,
      "step": 2846
    },
    {
      "epoch": 0.7634754625905068,
      "grad_norm": 0.3721415811454178,
      "learning_rate": 9.290596919413101e-06,
      "loss": 0.0385,
      "step": 2847
    },
    {
      "epoch": 0.7637436310002682,
      "grad_norm": 0.38618573361904773,
      "learning_rate": 9.28979563216577e-06,
      "loss": 0.036,
      "step": 2848
    },
    {
      "epoch": 0.7640117994100295,
      "grad_norm": 0.32117983868967254,
      "learning_rate": 9.288993927232428e-06,
      "loss": 0.0308,
      "step": 2849
    },
    {
      "epoch": 0.7642799678197908,
      "grad_norm": 0.4286102768522441,
      "learning_rate": 9.288191804691133e-06,
      "loss": 0.0551,
      "step": 2850
    },
    {
      "epoch": 0.7645481362295522,
      "grad_norm": 0.27393345841022004,
      "learning_rate": 9.287389264619988e-06,
      "loss": 0.0324,
      "step": 2851
    },
    {
      "epoch": 0.7648163046393135,
      "grad_norm": 1.9631919697935445,
      "learning_rate": 9.286586307097133e-06,
      "loss": 0.0369,
      "step": 2852
    },
    {
      "epoch": 0.7650844730490748,
      "grad_norm": 0.47120762936962785,
      "learning_rate": 9.28578293220075e-06,
      "loss": 0.047,
      "step": 2853
    },
    {
      "epoch": 0.7653526414588362,
      "grad_norm": 0.35379517799937205,
      "learning_rate": 9.284979140009063e-06,
      "loss": 0.0302,
      "step": 2854
    },
    {
      "epoch": 0.7656208098685975,
      "grad_norm": 0.3077324276445034,
      "learning_rate": 9.28417493060033e-06,
      "loss": 0.0328,
      "step": 2855
    },
    {
      "epoch": 0.7658889782783588,
      "grad_norm": 0.4032159073727949,
      "learning_rate": 9.28337030405286e-06,
      "loss": 0.0349,
      "step": 2856
    },
    {
      "epoch": 0.7661571466881202,
      "grad_norm": 0.32622453273250657,
      "learning_rate": 9.282565260444997e-06,
      "loss": 0.0315,
      "step": 2857
    },
    {
      "epoch": 0.7664253150978815,
      "grad_norm": 0.24278069794108856,
      "learning_rate": 9.281759799855124e-06,
      "loss": 0.0215,
      "step": 2858
    },
    {
      "epoch": 0.7666934835076428,
      "grad_norm": 0.4207401007821895,
      "learning_rate": 9.280953922361667e-06,
      "loss": 0.0413,
      "step": 2859
    },
    {
      "epoch": 0.7669616519174042,
      "grad_norm": 0.37167423437554475,
      "learning_rate": 9.28014762804309e-06,
      "loss": 0.0447,
      "step": 2860
    },
    {
      "epoch": 0.7672298203271655,
      "grad_norm": 0.38849035610185767,
      "learning_rate": 9.279340916977905e-06,
      "loss": 0.0276,
      "step": 2861
    },
    {
      "epoch": 0.7674979887369268,
      "grad_norm": 0.33990446055648604,
      "learning_rate": 9.278533789244654e-06,
      "loss": 0.0411,
      "step": 2862
    },
    {
      "epoch": 0.7677661571466882,
      "grad_norm": 0.44579152880493045,
      "learning_rate": 9.27772624492193e-06,
      "loss": 0.03,
      "step": 2863
    },
    {
      "epoch": 0.7680343255564495,
      "grad_norm": 0.5594183542152563,
      "learning_rate": 9.276918284088357e-06,
      "loss": 0.0453,
      "step": 2864
    },
    {
      "epoch": 0.7683024939662108,
      "grad_norm": 0.28616054293230125,
      "learning_rate": 9.276109906822606e-06,
      "loss": 0.0278,
      "step": 2865
    },
    {
      "epoch": 0.7685706623759722,
      "grad_norm": 0.3998977683121425,
      "learning_rate": 9.275301113203385e-06,
      "loss": 0.0445,
      "step": 2866
    },
    {
      "epoch": 0.7688388307857335,
      "grad_norm": 0.2906678388234518,
      "learning_rate": 9.274491903309446e-06,
      "loss": 0.0177,
      "step": 2867
    },
    {
      "epoch": 0.7691069991954947,
      "grad_norm": 0.32125109508961996,
      "learning_rate": 9.27368227721958e-06,
      "loss": 0.0321,
      "step": 2868
    },
    {
      "epoch": 0.7693751676052561,
      "grad_norm": 0.29516684440327334,
      "learning_rate": 9.272872235012616e-06,
      "loss": 0.0184,
      "step": 2869
    },
    {
      "epoch": 0.7696433360150174,
      "grad_norm": 0.39368941622323006,
      "learning_rate": 9.272061776767426e-06,
      "loss": 0.0354,
      "step": 2870
    },
    {
      "epoch": 0.7699115044247787,
      "grad_norm": 0.30777440585080384,
      "learning_rate": 9.271250902562925e-06,
      "loss": 0.0391,
      "step": 2871
    },
    {
      "epoch": 0.7701796728345401,
      "grad_norm": 0.6774327998503527,
      "learning_rate": 9.270439612478062e-06,
      "loss": 0.0359,
      "step": 2872
    },
    {
      "epoch": 0.7704478412443014,
      "grad_norm": 0.4441090177897583,
      "learning_rate": 9.269627906591832e-06,
      "loss": 0.0316,
      "step": 2873
    },
    {
      "epoch": 0.7707160096540627,
      "grad_norm": 0.33288132099469747,
      "learning_rate": 9.268815784983269e-06,
      "loss": 0.0296,
      "step": 2874
    },
    {
      "epoch": 0.7709841780638241,
      "grad_norm": 0.4206354089356336,
      "learning_rate": 9.268003247731446e-06,
      "loss": 0.0306,
      "step": 2875
    },
    {
      "epoch": 0.7712523464735854,
      "grad_norm": 0.3168111800525939,
      "learning_rate": 9.267190294915477e-06,
      "loss": 0.0314,
      "step": 2876
    },
    {
      "epoch": 0.7715205148833467,
      "grad_norm": 0.3264549854941881,
      "learning_rate": 9.266376926614521e-06,
      "loss": 0.033,
      "step": 2877
    },
    {
      "epoch": 0.7717886832931081,
      "grad_norm": 0.3753908491147173,
      "learning_rate": 9.26556314290777e-06,
      "loss": 0.0336,
      "step": 2878
    },
    {
      "epoch": 0.7720568517028694,
      "grad_norm": 0.3618555930468025,
      "learning_rate": 9.264748943874459e-06,
      "loss": 0.0274,
      "step": 2879
    },
    {
      "epoch": 0.7723250201126307,
      "grad_norm": 0.40818383519936635,
      "learning_rate": 9.263934329593867e-06,
      "loss": 0.0305,
      "step": 2880
    },
    {
      "epoch": 0.7725931885223921,
      "grad_norm": 0.5633948549489048,
      "learning_rate": 9.26311930014531e-06,
      "loss": 0.0416,
      "step": 2881
    },
    {
      "epoch": 0.7728613569321534,
      "grad_norm": 0.4888058442782244,
      "learning_rate": 9.262303855608144e-06,
      "loss": 0.0325,
      "step": 2882
    },
    {
      "epoch": 0.7731295253419147,
      "grad_norm": 0.31288107955832184,
      "learning_rate": 9.26148799606177e-06,
      "loss": 0.0262,
      "step": 2883
    },
    {
      "epoch": 0.7733976937516761,
      "grad_norm": 0.2663125418483868,
      "learning_rate": 9.260671721585621e-06,
      "loss": 0.0259,
      "step": 2884
    },
    {
      "epoch": 0.7736658621614374,
      "grad_norm": 0.3425978119906183,
      "learning_rate": 9.25985503225918e-06,
      "loss": 0.0328,
      "step": 2885
    },
    {
      "epoch": 0.7739340305711987,
      "grad_norm": 0.2477388835190887,
      "learning_rate": 9.259037928161963e-06,
      "loss": 0.0242,
      "step": 2886
    },
    {
      "epoch": 0.77420219898096,
      "grad_norm": 0.47817972581460266,
      "learning_rate": 9.258220409373531e-06,
      "loss": 0.0376,
      "step": 2887
    },
    {
      "epoch": 0.7744703673907214,
      "grad_norm": 0.4591811733774961,
      "learning_rate": 9.257402475973484e-06,
      "loss": 0.0442,
      "step": 2888
    },
    {
      "epoch": 0.7747385358004827,
      "grad_norm": 0.36702684523285767,
      "learning_rate": 9.256584128041461e-06,
      "loss": 0.0301,
      "step": 2889
    },
    {
      "epoch": 0.775006704210244,
      "grad_norm": 0.2607986350619122,
      "learning_rate": 9.255765365657142e-06,
      "loss": 0.023,
      "step": 2890
    },
    {
      "epoch": 0.7752748726200054,
      "grad_norm": 0.38115942224921817,
      "learning_rate": 9.254946188900247e-06,
      "loss": 0.027,
      "step": 2891
    },
    {
      "epoch": 0.7755430410297667,
      "grad_norm": 0.4750067878282839,
      "learning_rate": 9.25412659785054e-06,
      "loss": 0.0469,
      "step": 2892
    },
    {
      "epoch": 0.775811209439528,
      "grad_norm": 0.38435579051399765,
      "learning_rate": 9.253306592587819e-06,
      "loss": 0.03,
      "step": 2893
    },
    {
      "epoch": 0.7760793778492894,
      "grad_norm": 0.42991710155314566,
      "learning_rate": 9.25248617319193e-06,
      "loss": 0.0369,
      "step": 2894
    },
    {
      "epoch": 0.7763475462590507,
      "grad_norm": 0.291274152886589,
      "learning_rate": 9.251665339742751e-06,
      "loss": 0.0274,
      "step": 2895
    },
    {
      "epoch": 0.776615714668812,
      "grad_norm": 0.3959025467719945,
      "learning_rate": 9.250844092320205e-06,
      "loss": 0.0343,
      "step": 2896
    },
    {
      "epoch": 0.7768838830785734,
      "grad_norm": 0.2963606879208201,
      "learning_rate": 9.250022431004256e-06,
      "loss": 0.0348,
      "step": 2897
    },
    {
      "epoch": 0.7771520514883347,
      "grad_norm": 0.32210228850944017,
      "learning_rate": 9.249200355874907e-06,
      "loss": 0.0253,
      "step": 2898
    },
    {
      "epoch": 0.777420219898096,
      "grad_norm": 0.3874885696876033,
      "learning_rate": 9.248377867012201e-06,
      "loss": 0.0317,
      "step": 2899
    },
    {
      "epoch": 0.7776883883078574,
      "grad_norm": 0.29586802768376946,
      "learning_rate": 9.247554964496219e-06,
      "loss": 0.035,
      "step": 2900
    },
    {
      "epoch": 0.7779565567176187,
      "grad_norm": 0.2857833803016495,
      "learning_rate": 9.24673164840709e-06,
      "loss": 0.0266,
      "step": 2901
    },
    {
      "epoch": 0.77822472512738,
      "grad_norm": 0.39239095191745055,
      "learning_rate": 9.245907918824975e-06,
      "loss": 0.0432,
      "step": 2902
    },
    {
      "epoch": 0.7784928935371414,
      "grad_norm": 0.7872422186360707,
      "learning_rate": 9.245083775830078e-06,
      "loss": 0.037,
      "step": 2903
    },
    {
      "epoch": 0.7787610619469026,
      "grad_norm": 0.35325558197097995,
      "learning_rate": 9.244259219502642e-06,
      "loss": 0.0467,
      "step": 2904
    },
    {
      "epoch": 0.7790292303566639,
      "grad_norm": 0.42416244258953445,
      "learning_rate": 9.243434249922956e-06,
      "loss": 0.0341,
      "step": 2905
    },
    {
      "epoch": 0.7792973987664253,
      "grad_norm": 0.2851745504245758,
      "learning_rate": 9.242608867171344e-06,
      "loss": 0.0337,
      "step": 2906
    },
    {
      "epoch": 0.7795655671761866,
      "grad_norm": 0.3285099770515234,
      "learning_rate": 9.24178307132817e-06,
      "loss": 0.0304,
      "step": 2907
    },
    {
      "epoch": 0.7798337355859479,
      "grad_norm": 0.3120850566213279,
      "learning_rate": 9.24095686247384e-06,
      "loss": 0.0252,
      "step": 2908
    },
    {
      "epoch": 0.7801019039957093,
      "grad_norm": 0.3715484083170873,
      "learning_rate": 9.2401302406888e-06,
      "loss": 0.0312,
      "step": 2909
    },
    {
      "epoch": 0.7803700724054706,
      "grad_norm": 0.4724172755880104,
      "learning_rate": 9.239303206053536e-06,
      "loss": 0.0393,
      "step": 2910
    },
    {
      "epoch": 0.7806382408152319,
      "grad_norm": 0.24490762360874477,
      "learning_rate": 9.238475758648572e-06,
      "loss": 0.0241,
      "step": 2911
    },
    {
      "epoch": 0.7809064092249933,
      "grad_norm": 0.43182627428023196,
      "learning_rate": 9.237647898554478e-06,
      "loss": 0.0364,
      "step": 2912
    },
    {
      "epoch": 0.7811745776347546,
      "grad_norm": 0.2983818565572143,
      "learning_rate": 9.236819625851858e-06,
      "loss": 0.0341,
      "step": 2913
    },
    {
      "epoch": 0.7814427460445159,
      "grad_norm": 0.3156349225491049,
      "learning_rate": 9.23599094062136e-06,
      "loss": 0.041,
      "step": 2914
    },
    {
      "epoch": 0.7817109144542773,
      "grad_norm": 0.32192952501813255,
      "learning_rate": 9.235161842943671e-06,
      "loss": 0.028,
      "step": 2915
    },
    {
      "epoch": 0.7819790828640386,
      "grad_norm": 0.31853734732326805,
      "learning_rate": 9.234332332899518e-06,
      "loss": 0.0279,
      "step": 2916
    },
    {
      "epoch": 0.7822472512737999,
      "grad_norm": 0.2877768595970916,
      "learning_rate": 9.233502410569666e-06,
      "loss": 0.0286,
      "step": 2917
    },
    {
      "epoch": 0.7825154196835613,
      "grad_norm": 0.6719638037032413,
      "learning_rate": 9.232672076034924e-06,
      "loss": 0.046,
      "step": 2918
    },
    {
      "epoch": 0.7827835880933226,
      "grad_norm": 0.42557318495665974,
      "learning_rate": 9.231841329376142e-06,
      "loss": 0.0292,
      "step": 2919
    },
    {
      "epoch": 0.7830517565030839,
      "grad_norm": 0.47262458431062365,
      "learning_rate": 9.231010170674202e-06,
      "loss": 0.0453,
      "step": 2920
    },
    {
      "epoch": 0.7833199249128453,
      "grad_norm": 0.25954311224914733,
      "learning_rate": 9.230178600010039e-06,
      "loss": 0.0284,
      "step": 2921
    },
    {
      "epoch": 0.7835880933226066,
      "grad_norm": 0.2921434926455236,
      "learning_rate": 9.229346617464612e-06,
      "loss": 0.03,
      "step": 2922
    },
    {
      "epoch": 0.7838562617323679,
      "grad_norm": 0.2462859847809662,
      "learning_rate": 9.228514223118936e-06,
      "loss": 0.022,
      "step": 2923
    },
    {
      "epoch": 0.7841244301421293,
      "grad_norm": 0.71145255409983,
      "learning_rate": 9.227681417054056e-06,
      "loss": 0.0266,
      "step": 2924
    },
    {
      "epoch": 0.7843925985518906,
      "grad_norm": 0.4334291185190317,
      "learning_rate": 9.22684819935106e-06,
      "loss": 0.0348,
      "step": 2925
    },
    {
      "epoch": 0.7846607669616519,
      "grad_norm": 0.3735825386362639,
      "learning_rate": 9.226014570091078e-06,
      "loss": 0.0352,
      "step": 2926
    },
    {
      "epoch": 0.7849289353714133,
      "grad_norm": 0.6750603457745463,
      "learning_rate": 9.225180529355275e-06,
      "loss": 0.038,
      "step": 2927
    },
    {
      "epoch": 0.7851971037811746,
      "grad_norm": 0.5312029882760894,
      "learning_rate": 9.224346077224864e-06,
      "loss": 0.0378,
      "step": 2928
    },
    {
      "epoch": 0.7854652721909359,
      "grad_norm": 0.34456646067107993,
      "learning_rate": 9.223511213781091e-06,
      "loss": 0.0325,
      "step": 2929
    },
    {
      "epoch": 0.7857334406006973,
      "grad_norm": 0.2864240529300436,
      "learning_rate": 9.222675939105243e-06,
      "loss": 0.0273,
      "step": 2930
    },
    {
      "epoch": 0.7860016090104586,
      "grad_norm": 0.5657466333074619,
      "learning_rate": 9.22184025327865e-06,
      "loss": 0.0299,
      "step": 2931
    },
    {
      "epoch": 0.7862697774202199,
      "grad_norm": 0.34960080141117295,
      "learning_rate": 9.22100415638268e-06,
      "loss": 0.0397,
      "step": 2932
    },
    {
      "epoch": 0.7865379458299813,
      "grad_norm": 0.4910558216783106,
      "learning_rate": 9.220167648498743e-06,
      "loss": 0.045,
      "step": 2933
    },
    {
      "epoch": 0.7868061142397426,
      "grad_norm": 0.39766018944907,
      "learning_rate": 9.219330729708286e-06,
      "loss": 0.0303,
      "step": 2934
    },
    {
      "epoch": 0.7870742826495039,
      "grad_norm": 0.3346817620662647,
      "learning_rate": 9.218493400092796e-06,
      "loss": 0.0332,
      "step": 2935
    },
    {
      "epoch": 0.7873424510592653,
      "grad_norm": 0.40600667941086627,
      "learning_rate": 9.217655659733807e-06,
      "loss": 0.0326,
      "step": 2936
    },
    {
      "epoch": 0.7876106194690266,
      "grad_norm": 0.2918206742183915,
      "learning_rate": 9.216817508712882e-06,
      "loss": 0.0311,
      "step": 2937
    },
    {
      "epoch": 0.7878787878787878,
      "grad_norm": 0.25306235681448896,
      "learning_rate": 9.21597894711163e-06,
      "loss": 0.0236,
      "step": 2938
    },
    {
      "epoch": 0.7881469562885492,
      "grad_norm": 0.3642812214651202,
      "learning_rate": 9.215139975011705e-06,
      "loss": 0.039,
      "step": 2939
    },
    {
      "epoch": 0.7884151246983105,
      "grad_norm": 0.286237346956489,
      "learning_rate": 9.214300592494789e-06,
      "loss": 0.0275,
      "step": 2940
    },
    {
      "epoch": 0.7886832931080718,
      "grad_norm": 0.4393501334033114,
      "learning_rate": 9.213460799642612e-06,
      "loss": 0.0399,
      "step": 2941
    },
    {
      "epoch": 0.7889514615178332,
      "grad_norm": 0.45173118327172107,
      "learning_rate": 9.212620596536948e-06,
      "loss": 0.0334,
      "step": 2942
    },
    {
      "epoch": 0.7892196299275945,
      "grad_norm": 0.434343633182564,
      "learning_rate": 9.211779983259597e-06,
      "loss": 0.046,
      "step": 2943
    },
    {
      "epoch": 0.7894877983373558,
      "grad_norm": 0.29258152473573695,
      "learning_rate": 9.21093895989241e-06,
      "loss": 0.03,
      "step": 2944
    },
    {
      "epoch": 0.7897559667471172,
      "grad_norm": 0.29418289324012686,
      "learning_rate": 9.210097526517281e-06,
      "loss": 0.0348,
      "step": 2945
    },
    {
      "epoch": 0.7900241351568785,
      "grad_norm": 0.3852292607319929,
      "learning_rate": 9.20925568321613e-06,
      "loss": 0.0349,
      "step": 2946
    },
    {
      "epoch": 0.7902923035666398,
      "grad_norm": 0.4279837255662349,
      "learning_rate": 9.20841343007093e-06,
      "loss": 0.0391,
      "step": 2947
    },
    {
      "epoch": 0.7905604719764012,
      "grad_norm": 0.3371197227831816,
      "learning_rate": 9.207570767163686e-06,
      "loss": 0.0286,
      "step": 2948
    },
    {
      "epoch": 0.7908286403861625,
      "grad_norm": 1.2247903958194084,
      "learning_rate": 9.206727694576448e-06,
      "loss": 0.0413,
      "step": 2949
    },
    {
      "epoch": 0.7910968087959238,
      "grad_norm": 0.32768764007714346,
      "learning_rate": 9.205884212391303e-06,
      "loss": 0.0372,
      "step": 2950
    },
    {
      "epoch": 0.7913649772056852,
      "grad_norm": 0.2957693033485463,
      "learning_rate": 9.205040320690379e-06,
      "loss": 0.0348,
      "step": 2951
    },
    {
      "epoch": 0.7916331456154465,
      "grad_norm": 0.25105726877522855,
      "learning_rate": 9.204196019555842e-06,
      "loss": 0.0286,
      "step": 2952
    },
    {
      "epoch": 0.7919013140252078,
      "grad_norm": 0.846676857505455,
      "learning_rate": 9.203351309069902e-06,
      "loss": 0.0285,
      "step": 2953
    },
    {
      "epoch": 0.7921694824349692,
      "grad_norm": 0.3317145111184569,
      "learning_rate": 9.202506189314803e-06,
      "loss": 0.0405,
      "step": 2954
    },
    {
      "epoch": 0.7924376508447305,
      "grad_norm": 0.3337868633636277,
      "learning_rate": 9.201660660372835e-06,
      "loss": 0.0421,
      "step": 2955
    },
    {
      "epoch": 0.7927058192544918,
      "grad_norm": 0.21465959803285392,
      "learning_rate": 9.200814722326323e-06,
      "loss": 0.0234,
      "step": 2956
    },
    {
      "epoch": 0.7929739876642532,
      "grad_norm": 0.3155521926615943,
      "learning_rate": 9.199968375257635e-06,
      "loss": 0.0374,
      "step": 2957
    },
    {
      "epoch": 0.7932421560740145,
      "grad_norm": 0.4547267880282319,
      "learning_rate": 9.199121619249178e-06,
      "loss": 0.0389,
      "step": 2958
    },
    {
      "epoch": 0.7935103244837758,
      "grad_norm": 0.5314352221096742,
      "learning_rate": 9.198274454383396e-06,
      "loss": 0.047,
      "step": 2959
    },
    {
      "epoch": 0.7937784928935372,
      "grad_norm": 0.29606600019463253,
      "learning_rate": 9.197426880742777e-06,
      "loss": 0.03,
      "step": 2960
    },
    {
      "epoch": 0.7940466613032985,
      "grad_norm": 0.3829191972952014,
      "learning_rate": 9.196578898409847e-06,
      "loss": 0.0358,
      "step": 2961
    },
    {
      "epoch": 0.7943148297130598,
      "grad_norm": 0.6983473976847496,
      "learning_rate": 9.19573050746717e-06,
      "loss": 0.035,
      "step": 2962
    },
    {
      "epoch": 0.7945829981228212,
      "grad_norm": 0.48353845768653064,
      "learning_rate": 9.194881707997355e-06,
      "loss": 0.0355,
      "step": 2963
    },
    {
      "epoch": 0.7948511665325825,
      "grad_norm": 0.34744082322204334,
      "learning_rate": 9.194032500083044e-06,
      "loss": 0.0286,
      "step": 2964
    },
    {
      "epoch": 0.7951193349423438,
      "grad_norm": 0.31098731922195244,
      "learning_rate": 9.193182883806921e-06,
      "loss": 0.0299,
      "step": 2965
    },
    {
      "epoch": 0.7953875033521052,
      "grad_norm": 0.5097885362172042,
      "learning_rate": 9.192332859251719e-06,
      "loss": 0.0296,
      "step": 2966
    },
    {
      "epoch": 0.7956556717618665,
      "grad_norm": 0.325767160332601,
      "learning_rate": 9.191482426500192e-06,
      "loss": 0.0356,
      "step": 2967
    },
    {
      "epoch": 0.7959238401716278,
      "grad_norm": 0.4658919716997365,
      "learning_rate": 9.19063158563515e-06,
      "loss": 0.03,
      "step": 2968
    },
    {
      "epoch": 0.7961920085813892,
      "grad_norm": 0.31387209183212905,
      "learning_rate": 9.189780336739437e-06,
      "loss": 0.0277,
      "step": 2969
    },
    {
      "epoch": 0.7964601769911505,
      "grad_norm": 0.34838515313085294,
      "learning_rate": 9.188928679895938e-06,
      "loss": 0.0377,
      "step": 2970
    },
    {
      "epoch": 0.7967283454009118,
      "grad_norm": 0.40982727028896004,
      "learning_rate": 9.188076615187573e-06,
      "loss": 0.0379,
      "step": 2971
    },
    {
      "epoch": 0.7969965138106732,
      "grad_norm": 0.4140543624781629,
      "learning_rate": 9.18722414269731e-06,
      "loss": 0.0426,
      "step": 2972
    },
    {
      "epoch": 0.7972646822204345,
      "grad_norm": 0.3852544884138152,
      "learning_rate": 9.186371262508144e-06,
      "loss": 0.0358,
      "step": 2973
    },
    {
      "epoch": 0.7975328506301957,
      "grad_norm": 0.3271135808761734,
      "learning_rate": 9.185517974703127e-06,
      "loss": 0.0253,
      "step": 2974
    },
    {
      "epoch": 0.797801019039957,
      "grad_norm": 0.39152692049923954,
      "learning_rate": 9.184664279365334e-06,
      "loss": 0.0315,
      "step": 2975
    },
    {
      "epoch": 0.7980691874497184,
      "grad_norm": 0.6384409622919172,
      "learning_rate": 9.183810176577891e-06,
      "loss": 0.0483,
      "step": 2976
    },
    {
      "epoch": 0.7983373558594797,
      "grad_norm": 0.49404448480020785,
      "learning_rate": 9.182955666423961e-06,
      "loss": 0.0473,
      "step": 2977
    },
    {
      "epoch": 0.798605524269241,
      "grad_norm": 0.5881588925218225,
      "learning_rate": 9.182100748986742e-06,
      "loss": 0.0569,
      "step": 2978
    },
    {
      "epoch": 0.7988736926790024,
      "grad_norm": 0.4696415108203217,
      "learning_rate": 9.181245424349477e-06,
      "loss": 0.0458,
      "step": 2979
    },
    {
      "epoch": 0.7991418610887637,
      "grad_norm": 0.3106350506183395,
      "learning_rate": 9.180389692595444e-06,
      "loss": 0.0293,
      "step": 2980
    },
    {
      "epoch": 0.799410029498525,
      "grad_norm": 0.40883137381720336,
      "learning_rate": 9.179533553807967e-06,
      "loss": 0.0283,
      "step": 2981
    },
    {
      "epoch": 0.7996781979082864,
      "grad_norm": 0.29495381881447597,
      "learning_rate": 9.178677008070404e-06,
      "loss": 0.0265,
      "step": 2982
    },
    {
      "epoch": 0.7999463663180477,
      "grad_norm": 0.3299318535064028,
      "learning_rate": 9.177820055466155e-06,
      "loss": 0.0267,
      "step": 2983
    },
    {
      "epoch": 0.800214534727809,
      "grad_norm": 0.5106851715958182,
      "learning_rate": 9.17696269607866e-06,
      "loss": 0.0396,
      "step": 2984
    },
    {
      "epoch": 0.8004827031375704,
      "grad_norm": 0.37280014915346493,
      "learning_rate": 9.176104929991396e-06,
      "loss": 0.0408,
      "step": 2985
    },
    {
      "epoch": 0.8007508715473317,
      "grad_norm": 0.38444075467190897,
      "learning_rate": 9.175246757287881e-06,
      "loss": 0.0425,
      "step": 2986
    },
    {
      "epoch": 0.801019039957093,
      "grad_norm": 0.3502836852687659,
      "learning_rate": 9.174388178051676e-06,
      "loss": 0.0269,
      "step": 2987
    },
    {
      "epoch": 0.8012872083668544,
      "grad_norm": 0.40021572399901684,
      "learning_rate": 9.173529192366377e-06,
      "loss": 0.0362,
      "step": 2988
    },
    {
      "epoch": 0.8015553767766157,
      "grad_norm": 0.30812843746569213,
      "learning_rate": 9.172669800315619e-06,
      "loss": 0.0295,
      "step": 2989
    },
    {
      "epoch": 0.801823545186377,
      "grad_norm": 0.5901777184890611,
      "learning_rate": 9.171810001983082e-06,
      "loss": 0.0645,
      "step": 2990
    },
    {
      "epoch": 0.8020917135961384,
      "grad_norm": 0.3689056433248043,
      "learning_rate": 9.170949797452481e-06,
      "loss": 0.0278,
      "step": 2991
    },
    {
      "epoch": 0.8023598820058997,
      "grad_norm": 0.35330117841691916,
      "learning_rate": 9.170089186807574e-06,
      "loss": 0.0413,
      "step": 2992
    },
    {
      "epoch": 0.802628050415661,
      "grad_norm": 0.30306525495539743,
      "learning_rate": 9.169228170132151e-06,
      "loss": 0.0254,
      "step": 2993
    },
    {
      "epoch": 0.8028962188254224,
      "grad_norm": 0.47532318464789686,
      "learning_rate": 9.168366747510051e-06,
      "loss": 0.0406,
      "step": 2994
    },
    {
      "epoch": 0.8031643872351837,
      "grad_norm": 0.30685209833731675,
      "learning_rate": 9.16750491902515e-06,
      "loss": 0.0236,
      "step": 2995
    },
    {
      "epoch": 0.803432555644945,
      "grad_norm": 0.9273462803496283,
      "learning_rate": 9.166642684761358e-06,
      "loss": 0.044,
      "step": 2996
    },
    {
      "epoch": 0.8037007240547064,
      "grad_norm": 0.591294086170457,
      "learning_rate": 9.165780044802629e-06,
      "loss": 0.0362,
      "step": 2997
    },
    {
      "epoch": 0.8039688924644677,
      "grad_norm": 0.4094740124956683,
      "learning_rate": 9.164916999232958e-06,
      "loss": 0.0295,
      "step": 2998
    },
    {
      "epoch": 0.804237060874229,
      "grad_norm": 0.3992424506482337,
      "learning_rate": 9.164053548136376e-06,
      "loss": 0.0321,
      "step": 2999
    },
    {
      "epoch": 0.8045052292839904,
      "grad_norm": 0.6171502261179619,
      "learning_rate": 9.163189691596954e-06,
      "loss": 0.0371,
      "step": 3000
    },
    {
      "epoch": 0.8047733976937517,
      "grad_norm": 0.272662381707896,
      "learning_rate": 9.162325429698807e-06,
      "loss": 0.0228,
      "step": 3001
    },
    {
      "epoch": 0.805041566103513,
      "grad_norm": 0.45704383560185574,
      "learning_rate": 9.161460762526082e-06,
      "loss": 0.0349,
      "step": 3002
    },
    {
      "epoch": 0.8053097345132744,
      "grad_norm": 0.39471126398923945,
      "learning_rate": 9.160595690162974e-06,
      "loss": 0.0351,
      "step": 3003
    },
    {
      "epoch": 0.8055779029230357,
      "grad_norm": 0.36693805310353855,
      "learning_rate": 9.159730212693707e-06,
      "loss": 0.0331,
      "step": 3004
    },
    {
      "epoch": 0.805846071332797,
      "grad_norm": 0.3495793064558823,
      "learning_rate": 9.158864330202553e-06,
      "loss": 0.0367,
      "step": 3005
    },
    {
      "epoch": 0.8061142397425584,
      "grad_norm": 0.6080548003635315,
      "learning_rate": 9.157998042773823e-06,
      "loss": 0.0383,
      "step": 3006
    },
    {
      "epoch": 0.8063824081523197,
      "grad_norm": 0.42982678015602827,
      "learning_rate": 9.15713135049186e-06,
      "loss": 0.0327,
      "step": 3007
    },
    {
      "epoch": 0.806650576562081,
      "grad_norm": 0.3839392069262633,
      "learning_rate": 9.156264253441056e-06,
      "loss": 0.0411,
      "step": 3008
    },
    {
      "epoch": 0.8069187449718423,
      "grad_norm": 0.5147548557781123,
      "learning_rate": 9.155396751705836e-06,
      "loss": 0.0415,
      "step": 3009
    },
    {
      "epoch": 0.8071869133816036,
      "grad_norm": 0.27929030668574467,
      "learning_rate": 9.154528845370667e-06,
      "loss": 0.0304,
      "step": 3010
    },
    {
      "epoch": 0.8074550817913649,
      "grad_norm": 0.5720679358707791,
      "learning_rate": 9.153660534520054e-06,
      "loss": 0.0423,
      "step": 3011
    },
    {
      "epoch": 0.8077232502011263,
      "grad_norm": 0.39670541702155265,
      "learning_rate": 9.152791819238542e-06,
      "loss": 0.0406,
      "step": 3012
    },
    {
      "epoch": 0.8079914186108876,
      "grad_norm": 0.33911535696143075,
      "learning_rate": 9.151922699610718e-06,
      "loss": 0.0391,
      "step": 3013
    },
    {
      "epoch": 0.8082595870206489,
      "grad_norm": 0.25756459819393707,
      "learning_rate": 9.151053175721202e-06,
      "loss": 0.0288,
      "step": 3014
    },
    {
      "epoch": 0.8085277554304103,
      "grad_norm": 0.25873870855606657,
      "learning_rate": 9.15018324765466e-06,
      "loss": 0.0307,
      "step": 3015
    },
    {
      "epoch": 0.8087959238401716,
      "grad_norm": 0.2824723327497554,
      "learning_rate": 9.149312915495793e-06,
      "loss": 0.0312,
      "step": 3016
    },
    {
      "epoch": 0.8090640922499329,
      "grad_norm": 0.3470452490431448,
      "learning_rate": 9.148442179329344e-06,
      "loss": 0.0303,
      "step": 3017
    },
    {
      "epoch": 0.8093322606596943,
      "grad_norm": 0.3040693577018474,
      "learning_rate": 9.147571039240095e-06,
      "loss": 0.0325,
      "step": 3018
    },
    {
      "epoch": 0.8096004290694556,
      "grad_norm": 0.2810814305780686,
      "learning_rate": 9.146699495312862e-06,
      "loss": 0.0329,
      "step": 3019
    },
    {
      "epoch": 0.8098685974792169,
      "grad_norm": 0.31857804514296867,
      "learning_rate": 9.145827547632512e-06,
      "loss": 0.0264,
      "step": 3020
    },
    {
      "epoch": 0.8101367658889783,
      "grad_norm": 0.32349264980699577,
      "learning_rate": 9.14495519628394e-06,
      "loss": 0.028,
      "step": 3021
    },
    {
      "epoch": 0.8104049342987396,
      "grad_norm": 0.2674122395201621,
      "learning_rate": 9.144082441352084e-06,
      "loss": 0.0273,
      "step": 3022
    },
    {
      "epoch": 0.8106731027085009,
      "grad_norm": 0.2604662455400156,
      "learning_rate": 9.143209282921924e-06,
      "loss": 0.0251,
      "step": 3023
    },
    {
      "epoch": 0.8109412711182623,
      "grad_norm": 0.3456059845859436,
      "learning_rate": 9.142335721078475e-06,
      "loss": 0.04,
      "step": 3024
    },
    {
      "epoch": 0.8112094395280236,
      "grad_norm": 0.31504184947130953,
      "learning_rate": 9.141461755906795e-06,
      "loss": 0.0271,
      "step": 3025
    },
    {
      "epoch": 0.8114776079377849,
      "grad_norm": 0.2742502967078285,
      "learning_rate": 9.140587387491979e-06,
      "loss": 0.0261,
      "step": 3026
    },
    {
      "epoch": 0.8117457763475463,
      "grad_norm": 0.33228275138759356,
      "learning_rate": 9.139712615919163e-06,
      "loss": 0.0289,
      "step": 3027
    },
    {
      "epoch": 0.8120139447573076,
      "grad_norm": 0.28690398005725837,
      "learning_rate": 9.13883744127352e-06,
      "loss": 0.0339,
      "step": 3028
    },
    {
      "epoch": 0.8122821131670689,
      "grad_norm": 0.36197665868251,
      "learning_rate": 9.137961863640263e-06,
      "loss": 0.0318,
      "step": 3029
    },
    {
      "epoch": 0.8125502815768303,
      "grad_norm": 0.5392236751847658,
      "learning_rate": 9.137085883104648e-06,
      "loss": 0.0551,
      "step": 3030
    },
    {
      "epoch": 0.8128184499865916,
      "grad_norm": 0.4478948689781035,
      "learning_rate": 9.136209499751962e-06,
      "loss": 0.042,
      "step": 3031
    },
    {
      "epoch": 0.8130866183963529,
      "grad_norm": 0.3211957503723893,
      "learning_rate": 9.135332713667538e-06,
      "loss": 0.0257,
      "step": 3032
    },
    {
      "epoch": 0.8133547868061143,
      "grad_norm": 0.26994167133789987,
      "learning_rate": 9.134455524936749e-06,
      "loss": 0.0303,
      "step": 3033
    },
    {
      "epoch": 0.8136229552158756,
      "grad_norm": 0.5568876154616585,
      "learning_rate": 9.133577933644999e-06,
      "loss": 0.0278,
      "step": 3034
    },
    {
      "epoch": 0.8138911236256369,
      "grad_norm": 0.3306401385621047,
      "learning_rate": 9.132699939877744e-06,
      "loss": 0.0228,
      "step": 3035
    },
    {
      "epoch": 0.8141592920353983,
      "grad_norm": 0.39528198039001566,
      "learning_rate": 9.131821543720464e-06,
      "loss": 0.0436,
      "step": 3036
    },
    {
      "epoch": 0.8144274604451596,
      "grad_norm": 0.31628972788774434,
      "learning_rate": 9.130942745258694e-06,
      "loss": 0.0293,
      "step": 3037
    },
    {
      "epoch": 0.8146956288549209,
      "grad_norm": 0.30613185330806464,
      "learning_rate": 9.130063544577993e-06,
      "loss": 0.0279,
      "step": 3038
    },
    {
      "epoch": 0.8149637972646823,
      "grad_norm": 0.374751125623404,
      "learning_rate": 9.129183941763971e-06,
      "loss": 0.037,
      "step": 3039
    },
    {
      "epoch": 0.8152319656744436,
      "grad_norm": 0.4063969657160524,
      "learning_rate": 9.128303936902272e-06,
      "loss": 0.0338,
      "step": 3040
    },
    {
      "epoch": 0.8155001340842049,
      "grad_norm": 0.28772467941568475,
      "learning_rate": 9.127423530078578e-06,
      "loss": 0.0268,
      "step": 3041
    },
    {
      "epoch": 0.8157683024939663,
      "grad_norm": 0.3127625889624516,
      "learning_rate": 9.126542721378613e-06,
      "loss": 0.0304,
      "step": 3042
    },
    {
      "epoch": 0.8160364709037276,
      "grad_norm": 0.3429209130916632,
      "learning_rate": 9.12566151088814e-06,
      "loss": 0.0375,
      "step": 3043
    },
    {
      "epoch": 0.8163046393134888,
      "grad_norm": 0.2551717373100411,
      "learning_rate": 9.124779898692959e-06,
      "loss": 0.0253,
      "step": 3044
    },
    {
      "epoch": 0.8165728077232502,
      "grad_norm": 0.30956001862151933,
      "learning_rate": 9.123897884878909e-06,
      "loss": 0.0356,
      "step": 3045
    },
    {
      "epoch": 0.8168409761330115,
      "grad_norm": 0.6829366931363557,
      "learning_rate": 9.123015469531872e-06,
      "loss": 0.0454,
      "step": 3046
    },
    {
      "epoch": 0.8171091445427728,
      "grad_norm": 0.4532034604499228,
      "learning_rate": 9.122132652737765e-06,
      "loss": 0.0397,
      "step": 3047
    },
    {
      "epoch": 0.8173773129525342,
      "grad_norm": 0.29280328713506043,
      "learning_rate": 9.121249434582545e-06,
      "loss": 0.0296,
      "step": 3048
    },
    {
      "epoch": 0.8176454813622955,
      "grad_norm": 0.30626780408156906,
      "learning_rate": 9.12036581515221e-06,
      "loss": 0.0434,
      "step": 3049
    },
    {
      "epoch": 0.8179136497720568,
      "grad_norm": 0.30361495489996665,
      "learning_rate": 9.119481794532794e-06,
      "loss": 0.0328,
      "step": 3050
    },
    {
      "epoch": 0.8181818181818182,
      "grad_norm": 0.3683483553884046,
      "learning_rate": 9.118597372810374e-06,
      "loss": 0.0328,
      "step": 3051
    },
    {
      "epoch": 0.8184499865915795,
      "grad_norm": 0.3869111290161808,
      "learning_rate": 9.11771255007106e-06,
      "loss": 0.0364,
      "step": 3052
    },
    {
      "epoch": 0.8187181550013408,
      "grad_norm": 0.30593473785642084,
      "learning_rate": 9.11682732640101e-06,
      "loss": 0.0359,
      "step": 3053
    },
    {
      "epoch": 0.8189863234111022,
      "grad_norm": 1.1506624742367686,
      "learning_rate": 9.115941701886412e-06,
      "loss": 0.0372,
      "step": 3054
    },
    {
      "epoch": 0.8192544918208635,
      "grad_norm": 0.38895242082064607,
      "learning_rate": 9.115055676613498e-06,
      "loss": 0.036,
      "step": 3055
    },
    {
      "epoch": 0.8195226602306248,
      "grad_norm": 0.40728702812150686,
      "learning_rate": 9.114169250668539e-06,
      "loss": 0.0309,
      "step": 3056
    },
    {
      "epoch": 0.8197908286403862,
      "grad_norm": 0.30913350372350334,
      "learning_rate": 9.113282424137842e-06,
      "loss": 0.0454,
      "step": 3057
    },
    {
      "epoch": 0.8200589970501475,
      "grad_norm": 0.608336766134122,
      "learning_rate": 9.112395197107757e-06,
      "loss": 0.0313,
      "step": 3058
    },
    {
      "epoch": 0.8203271654599088,
      "grad_norm": 0.3376312625315893,
      "learning_rate": 9.111507569664668e-06,
      "loss": 0.0322,
      "step": 3059
    },
    {
      "epoch": 0.8205953338696702,
      "grad_norm": 0.45436284478093475,
      "learning_rate": 9.110619541895006e-06,
      "loss": 0.0385,
      "step": 3060
    },
    {
      "epoch": 0.8208635022794315,
      "grad_norm": 0.36910999613327167,
      "learning_rate": 9.10973111388523e-06,
      "loss": 0.025,
      "step": 3061
    },
    {
      "epoch": 0.8211316706891928,
      "grad_norm": 0.5460036612796674,
      "learning_rate": 9.108842285721846e-06,
      "loss": 0.0383,
      "step": 3062
    },
    {
      "epoch": 0.8213998390989542,
      "grad_norm": 0.3670920836595705,
      "learning_rate": 9.107953057491399e-06,
      "loss": 0.0375,
      "step": 3063
    },
    {
      "epoch": 0.8216680075087155,
      "grad_norm": 0.37793424664600395,
      "learning_rate": 9.10706342928047e-06,
      "loss": 0.0308,
      "step": 3064
    },
    {
      "epoch": 0.8219361759184768,
      "grad_norm": 0.5431083712142541,
      "learning_rate": 9.106173401175677e-06,
      "loss": 0.052,
      "step": 3065
    },
    {
      "epoch": 0.8222043443282381,
      "grad_norm": 0.42426652187567626,
      "learning_rate": 9.105282973263682e-06,
      "loss": 0.0393,
      "step": 3066
    },
    {
      "epoch": 0.8224725127379995,
      "grad_norm": 0.5028644869569973,
      "learning_rate": 9.104392145631183e-06,
      "loss": 0.0322,
      "step": 3067
    },
    {
      "epoch": 0.8227406811477608,
      "grad_norm": 0.6119837149316172,
      "learning_rate": 9.103500918364917e-06,
      "loss": 0.0469,
      "step": 3068
    },
    {
      "epoch": 0.8230088495575221,
      "grad_norm": 0.38931155191318556,
      "learning_rate": 9.102609291551664e-06,
      "loss": 0.0384,
      "step": 3069
    },
    {
      "epoch": 0.8232770179672835,
      "grad_norm": 0.36962275632493946,
      "learning_rate": 9.101717265278235e-06,
      "loss": 0.0344,
      "step": 3070
    },
    {
      "epoch": 0.8235451863770448,
      "grad_norm": 0.30944197089227476,
      "learning_rate": 9.100824839631485e-06,
      "loss": 0.0314,
      "step": 3071
    },
    {
      "epoch": 0.8238133547868061,
      "grad_norm": 0.3783513842794875,
      "learning_rate": 9.09993201469831e-06,
      "loss": 0.0364,
      "step": 3072
    },
    {
      "epoch": 0.8240815231965675,
      "grad_norm": 0.4389371095230446,
      "learning_rate": 9.099038790565638e-06,
      "loss": 0.0418,
      "step": 3073
    },
    {
      "epoch": 0.8243496916063288,
      "grad_norm": 0.32554529847844593,
      "learning_rate": 9.098145167320442e-06,
      "loss": 0.0262,
      "step": 3074
    },
    {
      "epoch": 0.82461786001609,
      "grad_norm": 0.33860790537922875,
      "learning_rate": 9.09725114504973e-06,
      "loss": 0.0364,
      "step": 3075
    },
    {
      "epoch": 0.8248860284258515,
      "grad_norm": 0.32978994764763964,
      "learning_rate": 9.096356723840555e-06,
      "loss": 0.0332,
      "step": 3076
    },
    {
      "epoch": 0.8251541968356128,
      "grad_norm": 0.5680882997051623,
      "learning_rate": 9.095461903779999e-06,
      "loss": 0.0276,
      "step": 3077
    },
    {
      "epoch": 0.825422365245374,
      "grad_norm": 0.3331176452849128,
      "learning_rate": 9.094566684955191e-06,
      "loss": 0.0311,
      "step": 3078
    },
    {
      "epoch": 0.8256905336551355,
      "grad_norm": 0.3278046495070281,
      "learning_rate": 9.093671067453296e-06,
      "loss": 0.0351,
      "step": 3079
    },
    {
      "epoch": 0.8259587020648967,
      "grad_norm": 0.25079030749389003,
      "learning_rate": 9.092775051361519e-06,
      "loss": 0.0329,
      "step": 3080
    },
    {
      "epoch": 0.826226870474658,
      "grad_norm": 0.4086511740591196,
      "learning_rate": 9.0918786367671e-06,
      "loss": 0.0333,
      "step": 3081
    },
    {
      "epoch": 0.8264950388844194,
      "grad_norm": 0.3127012071838921,
      "learning_rate": 9.090981823757322e-06,
      "loss": 0.0241,
      "step": 3082
    },
    {
      "epoch": 0.8267632072941807,
      "grad_norm": 0.4489518033135561,
      "learning_rate": 9.090084612419506e-06,
      "loss": 0.044,
      "step": 3083
    },
    {
      "epoch": 0.827031375703942,
      "grad_norm": 0.5340948978724632,
      "learning_rate": 9.089187002841008e-06,
      "loss": 0.038,
      "step": 3084
    },
    {
      "epoch": 0.8272995441137034,
      "grad_norm": 0.2978179656184194,
      "learning_rate": 9.08828899510923e-06,
      "loss": 0.018,
      "step": 3085
    },
    {
      "epoch": 0.8275677125234647,
      "grad_norm": 0.40182384092960544,
      "learning_rate": 9.087390589311606e-06,
      "loss": 0.0357,
      "step": 3086
    },
    {
      "epoch": 0.827835880933226,
      "grad_norm": 0.3523924782387512,
      "learning_rate": 9.086491785535613e-06,
      "loss": 0.0371,
      "step": 3087
    },
    {
      "epoch": 0.8281040493429874,
      "grad_norm": 0.33564387106327137,
      "learning_rate": 9.085592583868765e-06,
      "loss": 0.0244,
      "step": 3088
    },
    {
      "epoch": 0.8283722177527487,
      "grad_norm": 0.5159426099321865,
      "learning_rate": 9.084692984398612e-06,
      "loss": 0.0344,
      "step": 3089
    },
    {
      "epoch": 0.82864038616251,
      "grad_norm": 0.37687601396302756,
      "learning_rate": 9.083792987212749e-06,
      "loss": 0.0325,
      "step": 3090
    },
    {
      "epoch": 0.8289085545722714,
      "grad_norm": 0.5465975755904534,
      "learning_rate": 9.082892592398804e-06,
      "loss": 0.033,
      "step": 3091
    },
    {
      "epoch": 0.8291767229820327,
      "grad_norm": 0.2717473455166893,
      "learning_rate": 9.08199180004445e-06,
      "loss": 0.03,
      "step": 3092
    },
    {
      "epoch": 0.829444891391794,
      "grad_norm": 0.5581666388459535,
      "learning_rate": 9.081090610237388e-06,
      "loss": 0.0315,
      "step": 3093
    },
    {
      "epoch": 0.8297130598015554,
      "grad_norm": 0.2894183770250827,
      "learning_rate": 9.080189023065371e-06,
      "loss": 0.0224,
      "step": 3094
    },
    {
      "epoch": 0.8299812282113167,
      "grad_norm": 0.3082148665308753,
      "learning_rate": 9.07928703861618e-06,
      "loss": 0.0389,
      "step": 3095
    },
    {
      "epoch": 0.830249396621078,
      "grad_norm": 0.22991995262418755,
      "learning_rate": 9.078384656977642e-06,
      "loss": 0.0257,
      "step": 3096
    },
    {
      "epoch": 0.8305175650308394,
      "grad_norm": 0.7300450657758292,
      "learning_rate": 9.077481878237618e-06,
      "loss": 0.0577,
      "step": 3097
    },
    {
      "epoch": 0.8307857334406007,
      "grad_norm": 0.30654389943120847,
      "learning_rate": 9.076578702484007e-06,
      "loss": 0.0274,
      "step": 3098
    },
    {
      "epoch": 0.831053901850362,
      "grad_norm": 0.24124307359018687,
      "learning_rate": 9.07567512980475e-06,
      "loss": 0.0218,
      "step": 3099
    },
    {
      "epoch": 0.8313220702601234,
      "grad_norm": 0.3875661642554389,
      "learning_rate": 9.07477116028783e-06,
      "loss": 0.0381,
      "step": 3100
    },
    {
      "epoch": 0.8315902386698847,
      "grad_norm": 0.4334348219983835,
      "learning_rate": 9.073866794021258e-06,
      "loss": 0.0414,
      "step": 3101
    },
    {
      "epoch": 0.831858407079646,
      "grad_norm": 0.8260876271711652,
      "learning_rate": 9.072962031093093e-06,
      "loss": 0.0374,
      "step": 3102
    },
    {
      "epoch": 0.8321265754894074,
      "grad_norm": 0.34637611007844676,
      "learning_rate": 9.072056871591428e-06,
      "loss": 0.0285,
      "step": 3103
    },
    {
      "epoch": 0.8323947438991687,
      "grad_norm": 0.2868896701001528,
      "learning_rate": 9.071151315604396e-06,
      "loss": 0.029,
      "step": 3104
    },
    {
      "epoch": 0.83266291230893,
      "grad_norm": 0.4918340253657461,
      "learning_rate": 9.070245363220172e-06,
      "loss": 0.0319,
      "step": 3105
    },
    {
      "epoch": 0.8329310807186914,
      "grad_norm": 0.3190149869348753,
      "learning_rate": 9.06933901452696e-06,
      "loss": 0.034,
      "step": 3106
    },
    {
      "epoch": 0.8331992491284527,
      "grad_norm": 0.44155784116958563,
      "learning_rate": 9.068432269613012e-06,
      "loss": 0.0392,
      "step": 3107
    },
    {
      "epoch": 0.833467417538214,
      "grad_norm": 0.4072642396779156,
      "learning_rate": 9.067525128566617e-06,
      "loss": 0.0341,
      "step": 3108
    },
    {
      "epoch": 0.8337355859479754,
      "grad_norm": 0.30112994993753417,
      "learning_rate": 9.066617591476099e-06,
      "loss": 0.0289,
      "step": 3109
    },
    {
      "epoch": 0.8340037543577367,
      "grad_norm": 0.44499814112431435,
      "learning_rate": 9.065709658429823e-06,
      "loss": 0.0412,
      "step": 3110
    },
    {
      "epoch": 0.834271922767498,
      "grad_norm": 0.28479295369719443,
      "learning_rate": 9.064801329516192e-06,
      "loss": 0.03,
      "step": 3111
    },
    {
      "epoch": 0.8345400911772594,
      "grad_norm": 0.39129826086620084,
      "learning_rate": 9.063892604823647e-06,
      "loss": 0.0249,
      "step": 3112
    },
    {
      "epoch": 0.8348082595870207,
      "grad_norm": 0.38430350964101617,
      "learning_rate": 9.062983484440668e-06,
      "loss": 0.0386,
      "step": 3113
    },
    {
      "epoch": 0.835076427996782,
      "grad_norm": 0.37062951028795926,
      "learning_rate": 9.062073968455773e-06,
      "loss": 0.0378,
      "step": 3114
    },
    {
      "epoch": 0.8353445964065433,
      "grad_norm": 0.4369090690362323,
      "learning_rate": 9.061164056957523e-06,
      "loss": 0.0406,
      "step": 3115
    },
    {
      "epoch": 0.8356127648163046,
      "grad_norm": 0.26945041050326907,
      "learning_rate": 9.06025375003451e-06,
      "loss": 0.0236,
      "step": 3116
    },
    {
      "epoch": 0.8358809332260659,
      "grad_norm": 0.43478029030002374,
      "learning_rate": 9.059343047775368e-06,
      "loss": 0.032,
      "step": 3117
    },
    {
      "epoch": 0.8361491016358273,
      "grad_norm": 0.2865719118338404,
      "learning_rate": 9.058431950268773e-06,
      "loss": 0.0304,
      "step": 3118
    },
    {
      "epoch": 0.8364172700455886,
      "grad_norm": 0.34997029572149246,
      "learning_rate": 9.057520457603433e-06,
      "loss": 0.0301,
      "step": 3119
    },
    {
      "epoch": 0.8366854384553499,
      "grad_norm": 0.4465879469737071,
      "learning_rate": 9.056608569868098e-06,
      "loss": 0.0383,
      "step": 3120
    },
    {
      "epoch": 0.8369536068651113,
      "grad_norm": 0.6074705756134123,
      "learning_rate": 9.055696287151556e-06,
      "loss": 0.0455,
      "step": 3121
    },
    {
      "epoch": 0.8372217752748726,
      "grad_norm": 0.3705844393107477,
      "learning_rate": 9.054783609542634e-06,
      "loss": 0.0365,
      "step": 3122
    },
    {
      "epoch": 0.8374899436846339,
      "grad_norm": 0.3909780785040752,
      "learning_rate": 9.053870537130198e-06,
      "loss": 0.0384,
      "step": 3123
    },
    {
      "epoch": 0.8377581120943953,
      "grad_norm": 0.3485549622796398,
      "learning_rate": 9.052957070003152e-06,
      "loss": 0.0251,
      "step": 3124
    },
    {
      "epoch": 0.8380262805041566,
      "grad_norm": 0.43383363917469075,
      "learning_rate": 9.052043208250435e-06,
      "loss": 0.0468,
      "step": 3125
    },
    {
      "epoch": 0.8382944489139179,
      "grad_norm": 0.3426200470566251,
      "learning_rate": 9.05112895196103e-06,
      "loss": 0.0377,
      "step": 3126
    },
    {
      "epoch": 0.8385626173236793,
      "grad_norm": 0.31915652098840464,
      "learning_rate": 9.050214301223952e-06,
      "loss": 0.0361,
      "step": 3127
    },
    {
      "epoch": 0.8388307857334406,
      "grad_norm": 0.2967358238652847,
      "learning_rate": 9.049299256128263e-06,
      "loss": 0.0199,
      "step": 3128
    },
    {
      "epoch": 0.8390989541432019,
      "grad_norm": 0.37109728605334674,
      "learning_rate": 9.048383816763056e-06,
      "loss": 0.0332,
      "step": 3129
    },
    {
      "epoch": 0.8393671225529633,
      "grad_norm": 0.4029886474786045,
      "learning_rate": 9.047467983217464e-06,
      "loss": 0.0305,
      "step": 3130
    },
    {
      "epoch": 0.8396352909627246,
      "grad_norm": 0.4449371363808408,
      "learning_rate": 9.04655175558066e-06,
      "loss": 0.0318,
      "step": 3131
    },
    {
      "epoch": 0.8399034593724859,
      "grad_norm": 0.32005270812870085,
      "learning_rate": 9.045635133941855e-06,
      "loss": 0.0364,
      "step": 3132
    },
    {
      "epoch": 0.8401716277822473,
      "grad_norm": 0.49584681897023297,
      "learning_rate": 9.044718118390299e-06,
      "loss": 0.0269,
      "step": 3133
    },
    {
      "epoch": 0.8404397961920086,
      "grad_norm": 0.31728975496481265,
      "learning_rate": 9.043800709015278e-06,
      "loss": 0.0357,
      "step": 3134
    },
    {
      "epoch": 0.8407079646017699,
      "grad_norm": 0.3453864283996984,
      "learning_rate": 9.042882905906118e-06,
      "loss": 0.0332,
      "step": 3135
    },
    {
      "epoch": 0.8409761330115313,
      "grad_norm": 0.4074679565986499,
      "learning_rate": 9.041964709152183e-06,
      "loss": 0.0369,
      "step": 3136
    },
    {
      "epoch": 0.8412443014212926,
      "grad_norm": 0.2910588243536259,
      "learning_rate": 9.041046118842874e-06,
      "loss": 0.0283,
      "step": 3137
    },
    {
      "epoch": 0.8415124698310539,
      "grad_norm": 0.44745177748835996,
      "learning_rate": 9.040127135067636e-06,
      "loss": 0.0365,
      "step": 3138
    },
    {
      "epoch": 0.8417806382408153,
      "grad_norm": 0.29617703208644147,
      "learning_rate": 9.039207757915942e-06,
      "loss": 0.0342,
      "step": 3139
    },
    {
      "epoch": 0.8420488066505766,
      "grad_norm": 0.362173470727667,
      "learning_rate": 9.038287987477314e-06,
      "loss": 0.0358,
      "step": 3140
    },
    {
      "epoch": 0.8423169750603379,
      "grad_norm": 0.341053417337464,
      "learning_rate": 9.037367823841308e-06,
      "loss": 0.0275,
      "step": 3141
    },
    {
      "epoch": 0.8425851434700993,
      "grad_norm": 0.39031860077077973,
      "learning_rate": 9.036447267097514e-06,
      "loss": 0.0362,
      "step": 3142
    },
    {
      "epoch": 0.8428533118798606,
      "grad_norm": 0.3492842547118297,
      "learning_rate": 9.035526317335567e-06,
      "loss": 0.0364,
      "step": 3143
    },
    {
      "epoch": 0.8431214802896219,
      "grad_norm": 0.38848250292252945,
      "learning_rate": 9.034604974645136e-06,
      "loss": 0.0314,
      "step": 3144
    },
    {
      "epoch": 0.8433896486993833,
      "grad_norm": 0.3153718121194499,
      "learning_rate": 9.03368323911593e-06,
      "loss": 0.0336,
      "step": 3145
    },
    {
      "epoch": 0.8436578171091446,
      "grad_norm": 0.3795468649491991,
      "learning_rate": 9.032761110837696e-06,
      "loss": 0.04,
      "step": 3146
    },
    {
      "epoch": 0.8439259855189059,
      "grad_norm": 0.25117989891886827,
      "learning_rate": 9.03183858990022e-06,
      "loss": 0.0316,
      "step": 3147
    },
    {
      "epoch": 0.8441941539286673,
      "grad_norm": 0.4281617229993968,
      "learning_rate": 9.030915676393326e-06,
      "loss": 0.0426,
      "step": 3148
    },
    {
      "epoch": 0.8444623223384286,
      "grad_norm": 0.30421827082506114,
      "learning_rate": 9.029992370406872e-06,
      "loss": 0.0294,
      "step": 3149
    },
    {
      "epoch": 0.8447304907481898,
      "grad_norm": 0.3227374630994128,
      "learning_rate": 9.029068672030764e-06,
      "loss": 0.0268,
      "step": 3150
    },
    {
      "epoch": 0.8449986591579512,
      "grad_norm": 0.7705866503023938,
      "learning_rate": 9.028144581354932e-06,
      "loss": 0.0591,
      "step": 3151
    },
    {
      "epoch": 0.8452668275677125,
      "grad_norm": 0.3443625661788152,
      "learning_rate": 9.02722009846936e-06,
      "loss": 0.0317,
      "step": 3152
    },
    {
      "epoch": 0.8455349959774738,
      "grad_norm": 0.5854392657362937,
      "learning_rate": 9.026295223464056e-06,
      "loss": 0.044,
      "step": 3153
    },
    {
      "epoch": 0.8458031643872351,
      "grad_norm": 0.24733972724772027,
      "learning_rate": 9.025369956429077e-06,
      "loss": 0.0252,
      "step": 3154
    },
    {
      "epoch": 0.8460713327969965,
      "grad_norm": 0.3728444055790479,
      "learning_rate": 9.024444297454515e-06,
      "loss": 0.0325,
      "step": 3155
    },
    {
      "epoch": 0.8463395012067578,
      "grad_norm": 0.22894489727878467,
      "learning_rate": 9.023518246630494e-06,
      "loss": 0.0264,
      "step": 3156
    },
    {
      "epoch": 0.8466076696165191,
      "grad_norm": 0.6619200610442779,
      "learning_rate": 9.022591804047183e-06,
      "loss": 0.0435,
      "step": 3157
    },
    {
      "epoch": 0.8468758380262805,
      "grad_norm": 0.3090323040461443,
      "learning_rate": 9.02166496979479e-06,
      "loss": 0.0282,
      "step": 3158
    },
    {
      "epoch": 0.8471440064360418,
      "grad_norm": 0.4461665208429402,
      "learning_rate": 9.020737743963555e-06,
      "loss": 0.0437,
      "step": 3159
    },
    {
      "epoch": 0.8474121748458031,
      "grad_norm": 0.4745944533188282,
      "learning_rate": 9.01981012664376e-06,
      "loss": 0.0554,
      "step": 3160
    },
    {
      "epoch": 0.8476803432555645,
      "grad_norm": 0.34101599772660124,
      "learning_rate": 9.018882117925726e-06,
      "loss": 0.0243,
      "step": 3161
    },
    {
      "epoch": 0.8479485116653258,
      "grad_norm": 0.25307496336010993,
      "learning_rate": 9.01795371789981e-06,
      "loss": 0.0214,
      "step": 3162
    },
    {
      "epoch": 0.8482166800750871,
      "grad_norm": 0.4041079798718042,
      "learning_rate": 9.017024926656408e-06,
      "loss": 0.0346,
      "step": 3163
    },
    {
      "epoch": 0.8484848484848485,
      "grad_norm": 0.3346664683517816,
      "learning_rate": 9.016095744285955e-06,
      "loss": 0.0352,
      "step": 3164
    },
    {
      "epoch": 0.8487530168946098,
      "grad_norm": 0.3183958910192475,
      "learning_rate": 9.015166170878919e-06,
      "loss": 0.0281,
      "step": 3165
    },
    {
      "epoch": 0.8490211853043711,
      "grad_norm": 0.31849158753590223,
      "learning_rate": 9.014236206525815e-06,
      "loss": 0.0319,
      "step": 3166
    },
    {
      "epoch": 0.8492893537141325,
      "grad_norm": 0.3657656980748535,
      "learning_rate": 9.01330585131719e-06,
      "loss": 0.0369,
      "step": 3167
    },
    {
      "epoch": 0.8495575221238938,
      "grad_norm": 0.39188994611326783,
      "learning_rate": 9.012375105343628e-06,
      "loss": 0.0379,
      "step": 3168
    },
    {
      "epoch": 0.8498256905336551,
      "grad_norm": 0.3985568203003311,
      "learning_rate": 9.011443968695756e-06,
      "loss": 0.0492,
      "step": 3169
    },
    {
      "epoch": 0.8500938589434165,
      "grad_norm": 0.31146790713106864,
      "learning_rate": 9.010512441464233e-06,
      "loss": 0.031,
      "step": 3170
    },
    {
      "epoch": 0.8503620273531778,
      "grad_norm": 0.3813254373964755,
      "learning_rate": 9.009580523739763e-06,
      "loss": 0.0362,
      "step": 3171
    },
    {
      "epoch": 0.8506301957629391,
      "grad_norm": 0.43182819013375173,
      "learning_rate": 9.008648215613082e-06,
      "loss": 0.0401,
      "step": 3172
    },
    {
      "epoch": 0.8508983641727005,
      "grad_norm": 0.3577038071168452,
      "learning_rate": 9.007715517174966e-06,
      "loss": 0.0325,
      "step": 3173
    },
    {
      "epoch": 0.8511665325824618,
      "grad_norm": 0.29550644536017023,
      "learning_rate": 9.006782428516232e-06,
      "loss": 0.0364,
      "step": 3174
    },
    {
      "epoch": 0.8514347009922231,
      "grad_norm": 0.28672215170160686,
      "learning_rate": 9.005848949727729e-06,
      "loss": 0.0343,
      "step": 3175
    },
    {
      "epoch": 0.8517028694019845,
      "grad_norm": 0.5555534676425643,
      "learning_rate": 9.004915080900352e-06,
      "loss": 0.0394,
      "step": 3176
    },
    {
      "epoch": 0.8519710378117458,
      "grad_norm": 0.33193332563266886,
      "learning_rate": 9.003980822125024e-06,
      "loss": 0.0287,
      "step": 3177
    },
    {
      "epoch": 0.8522392062215071,
      "grad_norm": 0.5095974674079591,
      "learning_rate": 9.003046173492713e-06,
      "loss": 0.0311,
      "step": 3178
    },
    {
      "epoch": 0.8525073746312685,
      "grad_norm": 0.47380994656224745,
      "learning_rate": 9.002111135094426e-06,
      "loss": 0.0596,
      "step": 3179
    },
    {
      "epoch": 0.8527755430410298,
      "grad_norm": 0.4185313007132331,
      "learning_rate": 9.001175707021202e-06,
      "loss": 0.0344,
      "step": 3180
    },
    {
      "epoch": 0.853043711450791,
      "grad_norm": 0.5477696917652317,
      "learning_rate": 9.000239889364123e-06,
      "loss": 0.0342,
      "step": 3181
    },
    {
      "epoch": 0.8533118798605525,
      "grad_norm": 0.34776106384848265,
      "learning_rate": 8.999303682214307e-06,
      "loss": 0.0315,
      "step": 3182
    },
    {
      "epoch": 0.8535800482703138,
      "grad_norm": 0.32848918843377956,
      "learning_rate": 8.998367085662908e-06,
      "loss": 0.0292,
      "step": 3183
    },
    {
      "epoch": 0.853848216680075,
      "grad_norm": 0.37073682237609096,
      "learning_rate": 8.997430099801122e-06,
      "loss": 0.0406,
      "step": 3184
    },
    {
      "epoch": 0.8541163850898364,
      "grad_norm": 0.3436111221285223,
      "learning_rate": 8.996492724720181e-06,
      "loss": 0.0393,
      "step": 3185
    },
    {
      "epoch": 0.8543845534995977,
      "grad_norm": 0.3792804269836349,
      "learning_rate": 8.995554960511352e-06,
      "loss": 0.0365,
      "step": 3186
    },
    {
      "epoch": 0.854652721909359,
      "grad_norm": 0.3466226889155668,
      "learning_rate": 8.994616807265945e-06,
      "loss": 0.0322,
      "step": 3187
    },
    {
      "epoch": 0.8549208903191204,
      "grad_norm": 0.43391486593982026,
      "learning_rate": 8.993678265075305e-06,
      "loss": 0.0263,
      "step": 3188
    },
    {
      "epoch": 0.8551890587288817,
      "grad_norm": 0.3492127444934099,
      "learning_rate": 8.992739334030815e-06,
      "loss": 0.0392,
      "step": 3189
    },
    {
      "epoch": 0.855457227138643,
      "grad_norm": 0.24623676316287096,
      "learning_rate": 8.9918000142239e-06,
      "loss": 0.0274,
      "step": 3190
    },
    {
      "epoch": 0.8557253955484044,
      "grad_norm": 0.3669496130756555,
      "learning_rate": 8.990860305746012e-06,
      "loss": 0.0375,
      "step": 3191
    },
    {
      "epoch": 0.8559935639581657,
      "grad_norm": 0.3500109449136609,
      "learning_rate": 8.989920208688652e-06,
      "loss": 0.0302,
      "step": 3192
    },
    {
      "epoch": 0.856261732367927,
      "grad_norm": 0.32621100491343347,
      "learning_rate": 8.988979723143354e-06,
      "loss": 0.0288,
      "step": 3193
    },
    {
      "epoch": 0.8565299007776884,
      "grad_norm": 0.6814033336907981,
      "learning_rate": 8.988038849201692e-06,
      "loss": 0.0335,
      "step": 3194
    },
    {
      "epoch": 0.8567980691874497,
      "grad_norm": 0.358976994800179,
      "learning_rate": 8.987097586955276e-06,
      "loss": 0.0271,
      "step": 3195
    },
    {
      "epoch": 0.857066237597211,
      "grad_norm": 0.2608819323564406,
      "learning_rate": 8.986155936495751e-06,
      "loss": 0.0255,
      "step": 3196
    },
    {
      "epoch": 0.8573344060069724,
      "grad_norm": 0.41455994412143765,
      "learning_rate": 8.985213897914808e-06,
      "loss": 0.0467,
      "step": 3197
    },
    {
      "epoch": 0.8576025744167337,
      "grad_norm": 0.3229304118858955,
      "learning_rate": 8.984271471304167e-06,
      "loss": 0.0239,
      "step": 3198
    },
    {
      "epoch": 0.857870742826495,
      "grad_norm": 0.7222250436027328,
      "learning_rate": 8.98332865675559e-06,
      "loss": 0.052,
      "step": 3199
    },
    {
      "epoch": 0.8581389112362564,
      "grad_norm": 0.35502807835787314,
      "learning_rate": 8.98238545436088e-06,
      "loss": 0.0317,
      "step": 3200
    },
    {
      "epoch": 0.8584070796460177,
      "grad_norm": 0.39044392043991466,
      "learning_rate": 8.981441864211869e-06,
      "loss": 0.0438,
      "step": 3201
    },
    {
      "epoch": 0.858675248055779,
      "grad_norm": 0.3282572212354986,
      "learning_rate": 8.980497886400435e-06,
      "loss": 0.0311,
      "step": 3202
    },
    {
      "epoch": 0.8589434164655404,
      "grad_norm": 0.3019406058203457,
      "learning_rate": 8.97955352101849e-06,
      "loss": 0.0319,
      "step": 3203
    },
    {
      "epoch": 0.8592115848753017,
      "grad_norm": 0.27288360170442766,
      "learning_rate": 8.978608768157984e-06,
      "loss": 0.0296,
      "step": 3204
    },
    {
      "epoch": 0.859479753285063,
      "grad_norm": 0.38901088771041836,
      "learning_rate": 8.977663627910903e-06,
      "loss": 0.0358,
      "step": 3205
    },
    {
      "epoch": 0.8597479216948244,
      "grad_norm": 0.28495683773255065,
      "learning_rate": 8.97671810036928e-06,
      "loss": 0.0287,
      "step": 3206
    },
    {
      "epoch": 0.8600160901045857,
      "grad_norm": 0.7244943475156721,
      "learning_rate": 8.97577218562517e-06,
      "loss": 0.0557,
      "step": 3207
    },
    {
      "epoch": 0.860284258514347,
      "grad_norm": 0.26255045452274306,
      "learning_rate": 8.974825883770678e-06,
      "loss": 0.0285,
      "step": 3208
    },
    {
      "epoch": 0.8605524269241084,
      "grad_norm": 0.4006075356604569,
      "learning_rate": 8.973879194897942e-06,
      "loss": 0.0418,
      "step": 3209
    },
    {
      "epoch": 0.8608205953338697,
      "grad_norm": 0.32337104089515767,
      "learning_rate": 8.972932119099142e-06,
      "loss": 0.0354,
      "step": 3210
    },
    {
      "epoch": 0.861088763743631,
      "grad_norm": 0.372123180523371,
      "learning_rate": 8.971984656466488e-06,
      "loss": 0.041,
      "step": 3211
    },
    {
      "epoch": 0.8613569321533924,
      "grad_norm": 0.36017765223260895,
      "learning_rate": 8.971036807092233e-06,
      "loss": 0.0393,
      "step": 3212
    },
    {
      "epoch": 0.8616251005631537,
      "grad_norm": 0.2943371140137503,
      "learning_rate": 8.970088571068667e-06,
      "loss": 0.026,
      "step": 3213
    },
    {
      "epoch": 0.861893268972915,
      "grad_norm": 0.4696854845032073,
      "learning_rate": 8.969139948488116e-06,
      "loss": 0.0324,
      "step": 3214
    },
    {
      "epoch": 0.8621614373826764,
      "grad_norm": 0.2966347142680746,
      "learning_rate": 8.968190939442948e-06,
      "loss": 0.0258,
      "step": 3215
    },
    {
      "epoch": 0.8624296057924377,
      "grad_norm": 0.33896000040923613,
      "learning_rate": 8.967241544025564e-06,
      "loss": 0.0261,
      "step": 3216
    },
    {
      "epoch": 0.862697774202199,
      "grad_norm": 0.321391771592232,
      "learning_rate": 8.966291762328404e-06,
      "loss": 0.0314,
      "step": 3217
    },
    {
      "epoch": 0.8629659426119604,
      "grad_norm": 0.382194289813668,
      "learning_rate": 8.965341594443944e-06,
      "loss": 0.0443,
      "step": 3218
    },
    {
      "epoch": 0.8632341110217217,
      "grad_norm": 0.3550964040323414,
      "learning_rate": 8.964391040464699e-06,
      "loss": 0.0358,
      "step": 3219
    },
    {
      "epoch": 0.8635022794314829,
      "grad_norm": 0.4411124751548579,
      "learning_rate": 8.963440100483228e-06,
      "loss": 0.04,
      "step": 3220
    },
    {
      "epoch": 0.8637704478412443,
      "grad_norm": 0.631411787320404,
      "learning_rate": 8.962488774592113e-06,
      "loss": 0.0326,
      "step": 3221
    },
    {
      "epoch": 0.8640386162510056,
      "grad_norm": 0.28770758140320707,
      "learning_rate": 8.961537062883989e-06,
      "loss": 0.0368,
      "step": 3222
    },
    {
      "epoch": 0.8643067846607669,
      "grad_norm": 0.5407600269692727,
      "learning_rate": 8.960584965451517e-06,
      "loss": 0.0417,
      "step": 3223
    },
    {
      "epoch": 0.8645749530705283,
      "grad_norm": 0.28969312791378926,
      "learning_rate": 8.959632482387404e-06,
      "loss": 0.0286,
      "step": 3224
    },
    {
      "epoch": 0.8648431214802896,
      "grad_norm": 0.34774294214349366,
      "learning_rate": 8.958679613784387e-06,
      "loss": 0.04,
      "step": 3225
    },
    {
      "epoch": 0.8651112898900509,
      "grad_norm": 0.2809542274150366,
      "learning_rate": 8.957726359735245e-06,
      "loss": 0.0304,
      "step": 3226
    },
    {
      "epoch": 0.8653794582998123,
      "grad_norm": 0.3701205745189522,
      "learning_rate": 8.956772720332799e-06,
      "loss": 0.0481,
      "step": 3227
    },
    {
      "epoch": 0.8656476267095736,
      "grad_norm": 0.38449005232171146,
      "learning_rate": 8.955818695669895e-06,
      "loss": 0.0375,
      "step": 3228
    },
    {
      "epoch": 0.8659157951193349,
      "grad_norm": 0.3150166236925231,
      "learning_rate": 8.954864285839428e-06,
      "loss": 0.0289,
      "step": 3229
    },
    {
      "epoch": 0.8661839635290963,
      "grad_norm": 0.5767512780363198,
      "learning_rate": 8.953909490934327e-06,
      "loss": 0.0586,
      "step": 3230
    },
    {
      "epoch": 0.8664521319388576,
      "grad_norm": 0.2885328817364179,
      "learning_rate": 8.952954311047554e-06,
      "loss": 0.0366,
      "step": 3231
    },
    {
      "epoch": 0.8667203003486189,
      "grad_norm": 0.4288043198371314,
      "learning_rate": 8.951998746272115e-06,
      "loss": 0.0413,
      "step": 3232
    },
    {
      "epoch": 0.8669884687583803,
      "grad_norm": 0.4649732287926261,
      "learning_rate": 8.951042796701051e-06,
      "loss": 0.0377,
      "step": 3233
    },
    {
      "epoch": 0.8672566371681416,
      "grad_norm": 0.378922569249171,
      "learning_rate": 8.95008646242744e-06,
      "loss": 0.0399,
      "step": 3234
    },
    {
      "epoch": 0.8675248055779029,
      "grad_norm": 0.33603473191199146,
      "learning_rate": 8.949129743544396e-06,
      "loss": 0.0322,
      "step": 3235
    },
    {
      "epoch": 0.8677929739876643,
      "grad_norm": 0.3940563696723601,
      "learning_rate": 8.948172640145076e-06,
      "loss": 0.0378,
      "step": 3236
    },
    {
      "epoch": 0.8680611423974256,
      "grad_norm": 0.3213987291497469,
      "learning_rate": 8.947215152322666e-06,
      "loss": 0.0326,
      "step": 3237
    },
    {
      "epoch": 0.8683293108071869,
      "grad_norm": 0.29584201504898233,
      "learning_rate": 8.946257280170398e-06,
      "loss": 0.0234,
      "step": 3238
    },
    {
      "epoch": 0.8685974792169483,
      "grad_norm": 0.3463067154635824,
      "learning_rate": 8.945299023781534e-06,
      "loss": 0.0347,
      "step": 3239
    },
    {
      "epoch": 0.8688656476267096,
      "grad_norm": 0.377574910832234,
      "learning_rate": 8.944340383249379e-06,
      "loss": 0.0382,
      "step": 3240
    },
    {
      "epoch": 0.8691338160364709,
      "grad_norm": 0.4535462750940223,
      "learning_rate": 8.943381358667273e-06,
      "loss": 0.0504,
      "step": 3241
    },
    {
      "epoch": 0.8694019844462322,
      "grad_norm": 0.4041825913856687,
      "learning_rate": 8.942421950128595e-06,
      "loss": 0.0326,
      "step": 3242
    },
    {
      "epoch": 0.8696701528559936,
      "grad_norm": 0.26440936377405416,
      "learning_rate": 8.941462157726757e-06,
      "loss": 0.0264,
      "step": 3243
    },
    {
      "epoch": 0.8699383212657549,
      "grad_norm": 0.3270436586550396,
      "learning_rate": 8.940501981555215e-06,
      "loss": 0.0356,
      "step": 3244
    },
    {
      "epoch": 0.8702064896755162,
      "grad_norm": 0.30588659881944097,
      "learning_rate": 8.939541421707454e-06,
      "loss": 0.0294,
      "step": 3245
    },
    {
      "epoch": 0.8704746580852776,
      "grad_norm": 0.8022143408534514,
      "learning_rate": 8.938580478277006e-06,
      "loss": 0.0319,
      "step": 3246
    },
    {
      "epoch": 0.8707428264950389,
      "grad_norm": 0.4480963503906354,
      "learning_rate": 8.937619151357433e-06,
      "loss": 0.0317,
      "step": 3247
    },
    {
      "epoch": 0.8710109949048002,
      "grad_norm": 0.48739118396863423,
      "learning_rate": 8.936657441042339e-06,
      "loss": 0.0374,
      "step": 3248
    },
    {
      "epoch": 0.8712791633145616,
      "grad_norm": 0.2952567352809718,
      "learning_rate": 8.935695347425359e-06,
      "loss": 0.029,
      "step": 3249
    },
    {
      "epoch": 0.8715473317243229,
      "grad_norm": 0.5099873586801738,
      "learning_rate": 8.934732870600174e-06,
      "loss": 0.0313,
      "step": 3250
    },
    {
      "epoch": 0.8718155001340842,
      "grad_norm": 0.34996629164151183,
      "learning_rate": 8.933770010660494e-06,
      "loss": 0.0322,
      "step": 3251
    },
    {
      "epoch": 0.8720836685438456,
      "grad_norm": 0.46288528816219054,
      "learning_rate": 8.932806767700074e-06,
      "loss": 0.0406,
      "step": 3252
    },
    {
      "epoch": 0.8723518369536069,
      "grad_norm": 0.2612171339194805,
      "learning_rate": 8.9318431418127e-06,
      "loss": 0.023,
      "step": 3253
    },
    {
      "epoch": 0.8726200053633681,
      "grad_norm": 0.5499180765106316,
      "learning_rate": 8.930879133092197e-06,
      "loss": 0.0286,
      "step": 3254
    },
    {
      "epoch": 0.8728881737731295,
      "grad_norm": 0.4750467431180505,
      "learning_rate": 8.92991474163243e-06,
      "loss": 0.0395,
      "step": 3255
    },
    {
      "epoch": 0.8731563421828908,
      "grad_norm": 0.36273049877686386,
      "learning_rate": 8.928949967527298e-06,
      "loss": 0.0324,
      "step": 3256
    },
    {
      "epoch": 0.8734245105926521,
      "grad_norm": 0.5923851200452512,
      "learning_rate": 8.927984810870739e-06,
      "loss": 0.0428,
      "step": 3257
    },
    {
      "epoch": 0.8736926790024135,
      "grad_norm": 0.9218356995878388,
      "learning_rate": 8.927019271756728e-06,
      "loss": 0.0424,
      "step": 3258
    },
    {
      "epoch": 0.8739608474121748,
      "grad_norm": 0.38520121678613917,
      "learning_rate": 8.926053350279276e-06,
      "loss": 0.0358,
      "step": 3259
    },
    {
      "epoch": 0.8742290158219361,
      "grad_norm": 0.5619199815534117,
      "learning_rate": 8.925087046532432e-06,
      "loss": 0.0424,
      "step": 3260
    },
    {
      "epoch": 0.8744971842316975,
      "grad_norm": 0.3592836946330606,
      "learning_rate": 8.924120360610284e-06,
      "loss": 0.0257,
      "step": 3261
    },
    {
      "epoch": 0.8747653526414588,
      "grad_norm": 0.5075692318255762,
      "learning_rate": 8.923153292606955e-06,
      "loss": 0.0459,
      "step": 3262
    },
    {
      "epoch": 0.8750335210512201,
      "grad_norm": 0.3932397047440339,
      "learning_rate": 8.922185842616607e-06,
      "loss": 0.0425,
      "step": 3263
    },
    {
      "epoch": 0.8753016894609815,
      "grad_norm": 0.36168841369329974,
      "learning_rate": 8.921218010733435e-06,
      "loss": 0.0339,
      "step": 3264
    },
    {
      "epoch": 0.8755698578707428,
      "grad_norm": 0.3686775604349555,
      "learning_rate": 8.920249797051675e-06,
      "loss": 0.023,
      "step": 3265
    },
    {
      "epoch": 0.8758380262805041,
      "grad_norm": 0.3090428418666465,
      "learning_rate": 8.919281201665601e-06,
      "loss": 0.0269,
      "step": 3266
    },
    {
      "epoch": 0.8761061946902655,
      "grad_norm": 0.34827805528932293,
      "learning_rate": 8.918312224669523e-06,
      "loss": 0.0249,
      "step": 3267
    },
    {
      "epoch": 0.8763743631000268,
      "grad_norm": 0.3664979037303756,
      "learning_rate": 8.917342866157786e-06,
      "loss": 0.0321,
      "step": 3268
    },
    {
      "epoch": 0.8766425315097881,
      "grad_norm": 0.36225470137596133,
      "learning_rate": 8.916373126224775e-06,
      "loss": 0.0363,
      "step": 3269
    },
    {
      "epoch": 0.8769106999195495,
      "grad_norm": 0.28954922673491484,
      "learning_rate": 8.915403004964908e-06,
      "loss": 0.0412,
      "step": 3270
    },
    {
      "epoch": 0.8771788683293108,
      "grad_norm": 0.41905707171959417,
      "learning_rate": 8.914432502472648e-06,
      "loss": 0.0307,
      "step": 3271
    },
    {
      "epoch": 0.8774470367390721,
      "grad_norm": 0.31273456641517294,
      "learning_rate": 8.913461618842488e-06,
      "loss": 0.0423,
      "step": 3272
    },
    {
      "epoch": 0.8777152051488335,
      "grad_norm": 0.5431537286675475,
      "learning_rate": 8.912490354168959e-06,
      "loss": 0.0251,
      "step": 3273
    },
    {
      "epoch": 0.8779833735585948,
      "grad_norm": 0.2959314442421235,
      "learning_rate": 8.911518708546633e-06,
      "loss": 0.0369,
      "step": 3274
    },
    {
      "epoch": 0.8782515419683561,
      "grad_norm": 0.35466975607671825,
      "learning_rate": 8.910546682070114e-06,
      "loss": 0.0349,
      "step": 3275
    },
    {
      "epoch": 0.8785197103781175,
      "grad_norm": 0.3232715242869063,
      "learning_rate": 8.909574274834047e-06,
      "loss": 0.0274,
      "step": 3276
    },
    {
      "epoch": 0.8787878787878788,
      "grad_norm": 0.31891309892168923,
      "learning_rate": 8.908601486933113e-06,
      "loss": 0.035,
      "step": 3277
    },
    {
      "epoch": 0.8790560471976401,
      "grad_norm": 0.3313864187110509,
      "learning_rate": 8.907628318462029e-06,
      "loss": 0.0303,
      "step": 3278
    },
    {
      "epoch": 0.8793242156074015,
      "grad_norm": 0.31161800667180134,
      "learning_rate": 8.906654769515551e-06,
      "loss": 0.0278,
      "step": 3279
    },
    {
      "epoch": 0.8795923840171628,
      "grad_norm": 0.45887522355644644,
      "learning_rate": 8.905680840188469e-06,
      "loss": 0.0334,
      "step": 3280
    },
    {
      "epoch": 0.8798605524269241,
      "grad_norm": 0.2942099822667207,
      "learning_rate": 8.904706530575613e-06,
      "loss": 0.0236,
      "step": 3281
    },
    {
      "epoch": 0.8801287208366855,
      "grad_norm": 0.2882601686904067,
      "learning_rate": 8.90373184077185e-06,
      "loss": 0.0273,
      "step": 3282
    },
    {
      "epoch": 0.8803968892464468,
      "grad_norm": 0.31027336066663547,
      "learning_rate": 8.902756770872082e-06,
      "loss": 0.0249,
      "step": 3283
    },
    {
      "epoch": 0.8806650576562081,
      "grad_norm": 0.45038515932523937,
      "learning_rate": 8.901781320971248e-06,
      "loss": 0.0324,
      "step": 3284
    },
    {
      "epoch": 0.8809332260659695,
      "grad_norm": 0.8601207000566916,
      "learning_rate": 8.900805491164326e-06,
      "loss": 0.037,
      "step": 3285
    },
    {
      "epoch": 0.8812013944757308,
      "grad_norm": 0.2614037160607683,
      "learning_rate": 8.89982928154633e-06,
      "loss": 0.0307,
      "step": 3286
    },
    {
      "epoch": 0.881469562885492,
      "grad_norm": 0.3785975348403741,
      "learning_rate": 8.89885269221231e-06,
      "loss": 0.0307,
      "step": 3287
    },
    {
      "epoch": 0.8817377312952535,
      "grad_norm": 0.4236452411071127,
      "learning_rate": 8.897875723257354e-06,
      "loss": 0.0429,
      "step": 3288
    },
    {
      "epoch": 0.8820058997050148,
      "grad_norm": 0.3723546714413032,
      "learning_rate": 8.89689837477659e-06,
      "loss": 0.0376,
      "step": 3289
    },
    {
      "epoch": 0.882274068114776,
      "grad_norm": 0.47873485227272405,
      "learning_rate": 8.895920646865174e-06,
      "loss": 0.0368,
      "step": 3290
    },
    {
      "epoch": 0.8825422365245374,
      "grad_norm": 0.4302309795989242,
      "learning_rate": 8.89494253961831e-06,
      "loss": 0.0343,
      "step": 3291
    },
    {
      "epoch": 0.8828104049342987,
      "grad_norm": 0.6503964904763161,
      "learning_rate": 8.893964053131232e-06,
      "loss": 0.0449,
      "step": 3292
    },
    {
      "epoch": 0.88307857334406,
      "grad_norm": 0.5351174767158696,
      "learning_rate": 8.892985187499212e-06,
      "loss": 0.0315,
      "step": 3293
    },
    {
      "epoch": 0.8833467417538214,
      "grad_norm": 0.30873965425267474,
      "learning_rate": 8.892005942817562e-06,
      "loss": 0.0294,
      "step": 3294
    },
    {
      "epoch": 0.8836149101635827,
      "grad_norm": 0.5722857051970847,
      "learning_rate": 8.891026319181625e-06,
      "loss": 0.0371,
      "step": 3295
    },
    {
      "epoch": 0.883883078573344,
      "grad_norm": 0.2791279025873966,
      "learning_rate": 8.890046316686783e-06,
      "loss": 0.031,
      "step": 3296
    },
    {
      "epoch": 0.8841512469831054,
      "grad_norm": 0.340300266612849,
      "learning_rate": 8.889065935428462e-06,
      "loss": 0.0377,
      "step": 3297
    },
    {
      "epoch": 0.8844194153928667,
      "grad_norm": 0.37073189110358273,
      "learning_rate": 8.888085175502114e-06,
      "loss": 0.0301,
      "step": 3298
    },
    {
      "epoch": 0.884687583802628,
      "grad_norm": 0.39141817682558183,
      "learning_rate": 8.887104037003238e-06,
      "loss": 0.0476,
      "step": 3299
    },
    {
      "epoch": 0.8849557522123894,
      "grad_norm": 0.4228666881111964,
      "learning_rate": 8.88612252002736e-06,
      "loss": 0.0349,
      "step": 3300
    },
    {
      "epoch": 0.8852239206221507,
      "grad_norm": 0.4339392215255841,
      "learning_rate": 8.88514062467005e-06,
      "loss": 0.0311,
      "step": 3301
    },
    {
      "epoch": 0.885492089031912,
      "grad_norm": 0.34311183664319495,
      "learning_rate": 8.884158351026912e-06,
      "loss": 0.0394,
      "step": 3302
    },
    {
      "epoch": 0.8857602574416734,
      "grad_norm": 0.3506717474607233,
      "learning_rate": 8.883175699193589e-06,
      "loss": 0.028,
      "step": 3303
    },
    {
      "epoch": 0.8860284258514347,
      "grad_norm": 0.4117176940344392,
      "learning_rate": 8.882192669265757e-06,
      "loss": 0.0336,
      "step": 3304
    },
    {
      "epoch": 0.886296594261196,
      "grad_norm": 0.3948742260453566,
      "learning_rate": 8.881209261339133e-06,
      "loss": 0.0395,
      "step": 3305
    },
    {
      "epoch": 0.8865647626709574,
      "grad_norm": 0.34122840519738834,
      "learning_rate": 8.880225475509465e-06,
      "loss": 0.032,
      "step": 3306
    },
    {
      "epoch": 0.8868329310807187,
      "grad_norm": 0.4820425898133421,
      "learning_rate": 8.879241311872546e-06,
      "loss": 0.0347,
      "step": 3307
    },
    {
      "epoch": 0.88710109949048,
      "grad_norm": 0.900549644164003,
      "learning_rate": 8.878256770524199e-06,
      "loss": 0.0288,
      "step": 3308
    },
    {
      "epoch": 0.8873692679002414,
      "grad_norm": 0.49876627362592213,
      "learning_rate": 8.877271851560287e-06,
      "loss": 0.0378,
      "step": 3309
    },
    {
      "epoch": 0.8876374363100027,
      "grad_norm": 0.38837102569002036,
      "learning_rate": 8.876286555076708e-06,
      "loss": 0.0375,
      "step": 3310
    },
    {
      "epoch": 0.887905604719764,
      "grad_norm": 0.6987808592398832,
      "learning_rate": 8.8753008811694e-06,
      "loss": 0.0476,
      "step": 3311
    },
    {
      "epoch": 0.8881737731295254,
      "grad_norm": 0.3703855496726898,
      "learning_rate": 8.874314829934333e-06,
      "loss": 0.0367,
      "step": 3312
    },
    {
      "epoch": 0.8884419415392867,
      "grad_norm": 0.30861846324516695,
      "learning_rate": 8.873328401467517e-06,
      "loss": 0.0429,
      "step": 3313
    },
    {
      "epoch": 0.888710109949048,
      "grad_norm": 0.324905712302882,
      "learning_rate": 8.872341595864998e-06,
      "loss": 0.0344,
      "step": 3314
    },
    {
      "epoch": 0.8889782783588094,
      "grad_norm": 0.3144290213595244,
      "learning_rate": 8.871354413222859e-06,
      "loss": 0.0245,
      "step": 3315
    },
    {
      "epoch": 0.8892464467685707,
      "grad_norm": 0.3865801837851577,
      "learning_rate": 8.870366853637218e-06,
      "loss": 0.0381,
      "step": 3316
    },
    {
      "epoch": 0.889514615178332,
      "grad_norm": 0.2726863307668217,
      "learning_rate": 8.869378917204235e-06,
      "loss": 0.0264,
      "step": 3317
    },
    {
      "epoch": 0.8897827835880934,
      "grad_norm": 0.31691792052155127,
      "learning_rate": 8.868390604020097e-06,
      "loss": 0.0279,
      "step": 3318
    },
    {
      "epoch": 0.8900509519978547,
      "grad_norm": 0.4878425369129178,
      "learning_rate": 8.867401914181037e-06,
      "loss": 0.0336,
      "step": 3319
    },
    {
      "epoch": 0.890319120407616,
      "grad_norm": 0.36781826793960626,
      "learning_rate": 8.866412847783322e-06,
      "loss": 0.0272,
      "step": 3320
    },
    {
      "epoch": 0.8905872888173774,
      "grad_norm": 0.7612783888577035,
      "learning_rate": 8.86542340492325e-06,
      "loss": 0.0418,
      "step": 3321
    },
    {
      "epoch": 0.8908554572271387,
      "grad_norm": 0.2927581106778206,
      "learning_rate": 8.864433585697165e-06,
      "loss": 0.0298,
      "step": 3322
    },
    {
      "epoch": 0.8911236256369,
      "grad_norm": 0.3719632434211825,
      "learning_rate": 8.863443390201443e-06,
      "loss": 0.035,
      "step": 3323
    },
    {
      "epoch": 0.8913917940466614,
      "grad_norm": 0.5549574892921412,
      "learning_rate": 8.862452818532495e-06,
      "loss": 0.0299,
      "step": 3324
    },
    {
      "epoch": 0.8916599624564227,
      "grad_norm": 0.5077451261812873,
      "learning_rate": 8.86146187078677e-06,
      "loss": 0.0385,
      "step": 3325
    },
    {
      "epoch": 0.8919281308661839,
      "grad_norm": 0.42890304998150874,
      "learning_rate": 8.860470547060753e-06,
      "loss": 0.0405,
      "step": 3326
    },
    {
      "epoch": 0.8921962992759453,
      "grad_norm": 0.3861114806017171,
      "learning_rate": 8.85947884745097e-06,
      "loss": 0.0281,
      "step": 3327
    },
    {
      "epoch": 0.8924644676857066,
      "grad_norm": 0.3594894988476645,
      "learning_rate": 8.858486772053979e-06,
      "loss": 0.0292,
      "step": 3328
    },
    {
      "epoch": 0.8927326360954679,
      "grad_norm": 0.3106272669394502,
      "learning_rate": 8.857494320966374e-06,
      "loss": 0.0255,
      "step": 3329
    },
    {
      "epoch": 0.8930008045052292,
      "grad_norm": 0.6409873548751294,
      "learning_rate": 8.856501494284789e-06,
      "loss": 0.0492,
      "step": 3330
    },
    {
      "epoch": 0.8932689729149906,
      "grad_norm": 0.52757966250911,
      "learning_rate": 8.85550829210589e-06,
      "loss": 0.0456,
      "step": 3331
    },
    {
      "epoch": 0.8935371413247519,
      "grad_norm": 0.4850196871129454,
      "learning_rate": 8.854514714526387e-06,
      "loss": 0.0535,
      "step": 3332
    },
    {
      "epoch": 0.8938053097345132,
      "grad_norm": 0.3396470589916414,
      "learning_rate": 8.85352076164302e-06,
      "loss": 0.03,
      "step": 3333
    },
    {
      "epoch": 0.8940734781442746,
      "grad_norm": 0.31826293536107214,
      "learning_rate": 8.852526433552567e-06,
      "loss": 0.037,
      "step": 3334
    },
    {
      "epoch": 0.8943416465540359,
      "grad_norm": 0.33806972384301787,
      "learning_rate": 8.851531730351843e-06,
      "loss": 0.0385,
      "step": 3335
    },
    {
      "epoch": 0.8946098149637972,
      "grad_norm": 0.33337727025388797,
      "learning_rate": 8.8505366521377e-06,
      "loss": 0.0243,
      "step": 3336
    },
    {
      "epoch": 0.8948779833735586,
      "grad_norm": 0.40307956983732934,
      "learning_rate": 8.849541199007028e-06,
      "loss": 0.0374,
      "step": 3337
    },
    {
      "epoch": 0.8951461517833199,
      "grad_norm": 0.31194524328963236,
      "learning_rate": 8.848545371056747e-06,
      "loss": 0.0327,
      "step": 3338
    },
    {
      "epoch": 0.8954143201930812,
      "grad_norm": 0.29791350784274095,
      "learning_rate": 8.847549168383823e-06,
      "loss": 0.0294,
      "step": 3339
    },
    {
      "epoch": 0.8956824886028426,
      "grad_norm": 0.35764332892848805,
      "learning_rate": 8.84655259108525e-06,
      "loss": 0.0266,
      "step": 3340
    },
    {
      "epoch": 0.8959506570126039,
      "grad_norm": 0.5887583962089283,
      "learning_rate": 8.845555639258065e-06,
      "loss": 0.0441,
      "step": 3341
    },
    {
      "epoch": 0.8962188254223652,
      "grad_norm": 0.3677823172074353,
      "learning_rate": 8.844558312999337e-06,
      "loss": 0.0382,
      "step": 3342
    },
    {
      "epoch": 0.8964869938321266,
      "grad_norm": 0.30989777476938146,
      "learning_rate": 8.843560612406171e-06,
      "loss": 0.0344,
      "step": 3343
    },
    {
      "epoch": 0.8967551622418879,
      "grad_norm": 0.31029728861607003,
      "learning_rate": 8.842562537575715e-06,
      "loss": 0.0266,
      "step": 3344
    },
    {
      "epoch": 0.8970233306516492,
      "grad_norm": 0.28901366386189237,
      "learning_rate": 8.841564088605145e-06,
      "loss": 0.0208,
      "step": 3345
    },
    {
      "epoch": 0.8972914990614106,
      "grad_norm": 0.44138477391091824,
      "learning_rate": 8.840565265591679e-06,
      "loss": 0.0378,
      "step": 3346
    },
    {
      "epoch": 0.8975596674711719,
      "grad_norm": 0.41458639723773355,
      "learning_rate": 8.83956606863257e-06,
      "loss": 0.0369,
      "step": 3347
    },
    {
      "epoch": 0.8978278358809332,
      "grad_norm": 0.3479715390778149,
      "learning_rate": 8.838566497825108e-06,
      "loss": 0.0237,
      "step": 3348
    },
    {
      "epoch": 0.8980960042906946,
      "grad_norm": 0.30128962440086493,
      "learning_rate": 8.837566553266617e-06,
      "loss": 0.0294,
      "step": 3349
    },
    {
      "epoch": 0.8983641727004559,
      "grad_norm": 0.258723872046188,
      "learning_rate": 8.836566235054458e-06,
      "loss": 0.0312,
      "step": 3350
    },
    {
      "epoch": 0.8986323411102172,
      "grad_norm": 0.31627960368159475,
      "learning_rate": 8.835565543286031e-06,
      "loss": 0.0323,
      "step": 3351
    },
    {
      "epoch": 0.8989005095199786,
      "grad_norm": 0.28015933918117214,
      "learning_rate": 8.83456447805877e-06,
      "loss": 0.0255,
      "step": 3352
    },
    {
      "epoch": 0.8991686779297399,
      "grad_norm": 0.47879418858873796,
      "learning_rate": 8.833563039470146e-06,
      "loss": 0.0408,
      "step": 3353
    },
    {
      "epoch": 0.8994368463395012,
      "grad_norm": 0.32973095511955236,
      "learning_rate": 8.832561227617667e-06,
      "loss": 0.0266,
      "step": 3354
    },
    {
      "epoch": 0.8997050147492626,
      "grad_norm": 0.7922216973115608,
      "learning_rate": 8.831559042598879e-06,
      "loss": 0.0447,
      "step": 3355
    },
    {
      "epoch": 0.8999731831590239,
      "grad_norm": 0.30996261247351614,
      "learning_rate": 8.830556484511356e-06,
      "loss": 0.0337,
      "step": 3356
    },
    {
      "epoch": 0.9002413515687852,
      "grad_norm": 0.324663332546099,
      "learning_rate": 8.82955355345272e-06,
      "loss": 0.0298,
      "step": 3357
    },
    {
      "epoch": 0.9005095199785466,
      "grad_norm": 0.3067072650808897,
      "learning_rate": 8.828550249520623e-06,
      "loss": 0.0398,
      "step": 3358
    },
    {
      "epoch": 0.9007776883883079,
      "grad_norm": 0.3793997395327298,
      "learning_rate": 8.82754657281275e-06,
      "loss": 0.0332,
      "step": 3359
    },
    {
      "epoch": 0.9010458567980691,
      "grad_norm": 0.3051300451510978,
      "learning_rate": 8.82654252342683e-06,
      "loss": 0.0266,
      "step": 3360
    },
    {
      "epoch": 0.9013140252078305,
      "grad_norm": 0.4698608248360729,
      "learning_rate": 8.825538101460625e-06,
      "loss": 0.0343,
      "step": 3361
    },
    {
      "epoch": 0.9015821936175918,
      "grad_norm": 0.4392734641558583,
      "learning_rate": 8.82453330701193e-06,
      "loss": 0.0344,
      "step": 3362
    },
    {
      "epoch": 0.9018503620273531,
      "grad_norm": 0.5054275725233702,
      "learning_rate": 8.82352814017858e-06,
      "loss": 0.0418,
      "step": 3363
    },
    {
      "epoch": 0.9021185304371145,
      "grad_norm": 0.42184616039625833,
      "learning_rate": 8.82252260105845e-06,
      "loss": 0.0325,
      "step": 3364
    },
    {
      "epoch": 0.9023866988468758,
      "grad_norm": 0.3165308049364458,
      "learning_rate": 8.82151668974944e-06,
      "loss": 0.0258,
      "step": 3365
    },
    {
      "epoch": 0.9026548672566371,
      "grad_norm": 0.29336659180468483,
      "learning_rate": 8.820510406349496e-06,
      "loss": 0.0329,
      "step": 3366
    },
    {
      "epoch": 0.9029230356663985,
      "grad_norm": 0.3014416750605149,
      "learning_rate": 8.819503750956598e-06,
      "loss": 0.0321,
      "step": 3367
    },
    {
      "epoch": 0.9031912040761598,
      "grad_norm": 0.4039786332984564,
      "learning_rate": 8.81849672366876e-06,
      "loss": 0.0338,
      "step": 3368
    },
    {
      "epoch": 0.9034593724859211,
      "grad_norm": 0.3363985523291303,
      "learning_rate": 8.817489324584035e-06,
      "loss": 0.0209,
      "step": 3369
    },
    {
      "epoch": 0.9037275408956825,
      "grad_norm": 0.5406397513027865,
      "learning_rate": 8.816481553800508e-06,
      "loss": 0.0385,
      "step": 3370
    },
    {
      "epoch": 0.9039957093054438,
      "grad_norm": 0.4226334837778787,
      "learning_rate": 8.815473411416305e-06,
      "loss": 0.0256,
      "step": 3371
    },
    {
      "epoch": 0.9042638777152051,
      "grad_norm": 0.3296580986693705,
      "learning_rate": 8.814464897529587e-06,
      "loss": 0.0317,
      "step": 3372
    },
    {
      "epoch": 0.9045320461249665,
      "grad_norm": 0.3171868429539967,
      "learning_rate": 8.813456012238548e-06,
      "loss": 0.0285,
      "step": 3373
    },
    {
      "epoch": 0.9048002145347278,
      "grad_norm": 0.35259781151707476,
      "learning_rate": 8.812446755641422e-06,
      "loss": 0.0328,
      "step": 3374
    },
    {
      "epoch": 0.9050683829444891,
      "grad_norm": 0.4107802128087662,
      "learning_rate": 8.811437127836477e-06,
      "loss": 0.0422,
      "step": 3375
    },
    {
      "epoch": 0.9053365513542505,
      "grad_norm": 0.9102784371618802,
      "learning_rate": 8.810427128922016e-06,
      "loss": 0.0549,
      "step": 3376
    },
    {
      "epoch": 0.9056047197640118,
      "grad_norm": 0.2976432925411997,
      "learning_rate": 8.809416758996386e-06,
      "loss": 0.0286,
      "step": 3377
    },
    {
      "epoch": 0.9058728881737731,
      "grad_norm": 0.4101456242582264,
      "learning_rate": 8.808406018157957e-06,
      "loss": 0.0375,
      "step": 3378
    },
    {
      "epoch": 0.9061410565835345,
      "grad_norm": 0.7880858506767984,
      "learning_rate": 8.807394906505146e-06,
      "loss": 0.0479,
      "step": 3379
    },
    {
      "epoch": 0.9064092249932958,
      "grad_norm": 0.4130148510419459,
      "learning_rate": 8.806383424136403e-06,
      "loss": 0.0319,
      "step": 3380
    },
    {
      "epoch": 0.9066773934030571,
      "grad_norm": 0.34816711663518507,
      "learning_rate": 8.80537157115021e-06,
      "loss": 0.0342,
      "step": 3381
    },
    {
      "epoch": 0.9069455618128185,
      "grad_norm": 0.704650126882243,
      "learning_rate": 8.804359347645091e-06,
      "loss": 0.0655,
      "step": 3382
    },
    {
      "epoch": 0.9072137302225798,
      "grad_norm": 0.498878898192836,
      "learning_rate": 8.803346753719602e-06,
      "loss": 0.0355,
      "step": 3383
    },
    {
      "epoch": 0.9074818986323411,
      "grad_norm": 0.358983957861865,
      "learning_rate": 8.802333789472338e-06,
      "loss": 0.044,
      "step": 3384
    },
    {
      "epoch": 0.9077500670421025,
      "grad_norm": 0.5446881841330066,
      "learning_rate": 8.801320455001928e-06,
      "loss": 0.0308,
      "step": 3385
    },
    {
      "epoch": 0.9080182354518638,
      "grad_norm": 0.4286690498701049,
      "learning_rate": 8.800306750407034e-06,
      "loss": 0.0349,
      "step": 3386
    },
    {
      "epoch": 0.9082864038616251,
      "grad_norm": 0.3838394017896056,
      "learning_rate": 8.799292675786365e-06,
      "loss": 0.0329,
      "step": 3387
    },
    {
      "epoch": 0.9085545722713865,
      "grad_norm": 0.4076643387926635,
      "learning_rate": 8.798278231238655e-06,
      "loss": 0.0301,
      "step": 3388
    },
    {
      "epoch": 0.9088227406811478,
      "grad_norm": 0.42922836932543895,
      "learning_rate": 8.797263416862678e-06,
      "loss": 0.036,
      "step": 3389
    },
    {
      "epoch": 0.9090909090909091,
      "grad_norm": 0.3852969741160308,
      "learning_rate": 8.796248232757243e-06,
      "loss": 0.0385,
      "step": 3390
    },
    {
      "epoch": 0.9093590775006705,
      "grad_norm": 0.826275552472805,
      "learning_rate": 8.795232679021197e-06,
      "loss": 0.0468,
      "step": 3391
    },
    {
      "epoch": 0.9096272459104318,
      "grad_norm": 0.3395180810836074,
      "learning_rate": 8.79421675575342e-06,
      "loss": 0.0341,
      "step": 3392
    },
    {
      "epoch": 0.909895414320193,
      "grad_norm": 0.5598328947810235,
      "learning_rate": 8.793200463052832e-06,
      "loss": 0.0542,
      "step": 3393
    },
    {
      "epoch": 0.9101635827299545,
      "grad_norm": 0.3010822094662956,
      "learning_rate": 8.792183801018384e-06,
      "loss": 0.0324,
      "step": 3394
    },
    {
      "epoch": 0.9104317511397158,
      "grad_norm": 0.299682936763229,
      "learning_rate": 8.791166769749068e-06,
      "loss": 0.027,
      "step": 3395
    },
    {
      "epoch": 0.910699919549477,
      "grad_norm": 0.3549950679173996,
      "learning_rate": 8.79014936934391e-06,
      "loss": 0.0306,
      "step": 3396
    },
    {
      "epoch": 0.9109680879592384,
      "grad_norm": 0.3216899509017038,
      "learning_rate": 8.78913159990197e-06,
      "loss": 0.0262,
      "step": 3397
    },
    {
      "epoch": 0.9112362563689997,
      "grad_norm": 0.3896680482040204,
      "learning_rate": 8.788113461522346e-06,
      "loss": 0.0346,
      "step": 3398
    },
    {
      "epoch": 0.911504424778761,
      "grad_norm": 0.3242145948279843,
      "learning_rate": 8.787094954304172e-06,
      "loss": 0.0294,
      "step": 3399
    },
    {
      "epoch": 0.9117725931885224,
      "grad_norm": 0.44211969895353237,
      "learning_rate": 8.786076078346615e-06,
      "loss": 0.0458,
      "step": 3400
    },
    {
      "epoch": 0.9120407615982837,
      "grad_norm": 0.4171091732377363,
      "learning_rate": 8.785056833748883e-06,
      "loss": 0.0461,
      "step": 3401
    },
    {
      "epoch": 0.912308930008045,
      "grad_norm": 0.3025890450589386,
      "learning_rate": 8.78403722061022e-06,
      "loss": 0.0303,
      "step": 3402
    },
    {
      "epoch": 0.9125770984178064,
      "grad_norm": 0.39468709179311234,
      "learning_rate": 8.783017239029894e-06,
      "loss": 0.0406,
      "step": 3403
    },
    {
      "epoch": 0.9128452668275677,
      "grad_norm": 0.23899144352990173,
      "learning_rate": 8.781996889107226e-06,
      "loss": 0.0226,
      "step": 3404
    },
    {
      "epoch": 0.913113435237329,
      "grad_norm": 0.4167370601959788,
      "learning_rate": 8.780976170941561e-06,
      "loss": 0.0424,
      "step": 3405
    },
    {
      "epoch": 0.9133816036470904,
      "grad_norm": 0.6560744760719583,
      "learning_rate": 8.779955084632284e-06,
      "loss": 0.039,
      "step": 3406
    },
    {
      "epoch": 0.9136497720568517,
      "grad_norm": 0.4032588849662207,
      "learning_rate": 8.778933630278817e-06,
      "loss": 0.032,
      "step": 3407
    },
    {
      "epoch": 0.913917940466613,
      "grad_norm": 0.3298897366921061,
      "learning_rate": 8.777911807980615e-06,
      "loss": 0.027,
      "step": 3408
    },
    {
      "epoch": 0.9141861088763744,
      "grad_norm": 0.3313719192088538,
      "learning_rate": 8.77688961783717e-06,
      "loss": 0.0358,
      "step": 3409
    },
    {
      "epoch": 0.9144542772861357,
      "grad_norm": 0.3052567224309724,
      "learning_rate": 8.775867059948011e-06,
      "loss": 0.0372,
      "step": 3410
    },
    {
      "epoch": 0.914722445695897,
      "grad_norm": 0.49090418032028404,
      "learning_rate": 8.7748441344127e-06,
      "loss": 0.0396,
      "step": 3411
    },
    {
      "epoch": 0.9149906141056584,
      "grad_norm": 0.535214036038146,
      "learning_rate": 8.77382084133084e-06,
      "loss": 0.0279,
      "step": 3412
    },
    {
      "epoch": 0.9152587825154197,
      "grad_norm": 0.7593444634931664,
      "learning_rate": 8.772797180802061e-06,
      "loss": 0.0404,
      "step": 3413
    },
    {
      "epoch": 0.915526950925181,
      "grad_norm": 0.694009308072015,
      "learning_rate": 8.77177315292604e-06,
      "loss": 0.0447,
      "step": 3414
    },
    {
      "epoch": 0.9157951193349424,
      "grad_norm": 0.46289340729487094,
      "learning_rate": 8.770748757802478e-06,
      "loss": 0.0488,
      "step": 3415
    },
    {
      "epoch": 0.9160632877447037,
      "grad_norm": 0.3729024557999195,
      "learning_rate": 8.769723995531123e-06,
      "loss": 0.043,
      "step": 3416
    },
    {
      "epoch": 0.916331456154465,
      "grad_norm": 0.282382262049196,
      "learning_rate": 8.76869886621175e-06,
      "loss": 0.0288,
      "step": 3417
    },
    {
      "epoch": 0.9165996245642264,
      "grad_norm": 0.3647329979242271,
      "learning_rate": 8.767673369944174e-06,
      "loss": 0.036,
      "step": 3418
    },
    {
      "epoch": 0.9168677929739877,
      "grad_norm": 0.5492621290030115,
      "learning_rate": 8.766647506828246e-06,
      "loss": 0.0381,
      "step": 3419
    },
    {
      "epoch": 0.917135961383749,
      "grad_norm": 0.42478480924233625,
      "learning_rate": 8.76562127696385e-06,
      "loss": 0.0276,
      "step": 3420
    },
    {
      "epoch": 0.9174041297935103,
      "grad_norm": 0.3516947363251873,
      "learning_rate": 8.764594680450907e-06,
      "loss": 0.0262,
      "step": 3421
    },
    {
      "epoch": 0.9176722982032717,
      "grad_norm": 0.6767930162581993,
      "learning_rate": 8.763567717389374e-06,
      "loss": 0.0362,
      "step": 3422
    },
    {
      "epoch": 0.917940466613033,
      "grad_norm": 0.5268727666848313,
      "learning_rate": 8.762540387879245e-06,
      "loss": 0.0263,
      "step": 3423
    },
    {
      "epoch": 0.9182086350227943,
      "grad_norm": 0.3294366556338562,
      "learning_rate": 8.76151269202055e-06,
      "loss": 0.0395,
      "step": 3424
    },
    {
      "epoch": 0.9184768034325557,
      "grad_norm": 0.412935295743048,
      "learning_rate": 8.760484629913349e-06,
      "loss": 0.0353,
      "step": 3425
    },
    {
      "epoch": 0.918744971842317,
      "grad_norm": 0.41242589394257234,
      "learning_rate": 8.759456201657743e-06,
      "loss": 0.0283,
      "step": 3426
    },
    {
      "epoch": 0.9190131402520783,
      "grad_norm": 0.6378129400959122,
      "learning_rate": 8.758427407353868e-06,
      "loss": 0.0399,
      "step": 3427
    },
    {
      "epoch": 0.9192813086618397,
      "grad_norm": 0.36620448421197715,
      "learning_rate": 8.757398247101896e-06,
      "loss": 0.0289,
      "step": 3428
    },
    {
      "epoch": 0.919549477071601,
      "grad_norm": 0.44196634307598237,
      "learning_rate": 8.756368721002031e-06,
      "loss": 0.0422,
      "step": 3429
    },
    {
      "epoch": 0.9198176454813622,
      "grad_norm": 0.820781953925549,
      "learning_rate": 8.755338829154517e-06,
      "loss": 0.0532,
      "step": 3430
    },
    {
      "epoch": 0.9200858138911236,
      "grad_norm": 0.4193592014543343,
      "learning_rate": 8.754308571659633e-06,
      "loss": 0.0439,
      "step": 3431
    },
    {
      "epoch": 0.9203539823008849,
      "grad_norm": 0.44182622778399266,
      "learning_rate": 8.753277948617688e-06,
      "loss": 0.0381,
      "step": 3432
    },
    {
      "epoch": 0.9206221507106462,
      "grad_norm": 0.48822212540505355,
      "learning_rate": 8.752246960129037e-06,
      "loss": 0.0339,
      "step": 3433
    },
    {
      "epoch": 0.9208903191204076,
      "grad_norm": 0.28524736457012784,
      "learning_rate": 8.75121560629406e-06,
      "loss": 0.0322,
      "step": 3434
    },
    {
      "epoch": 0.9211584875301689,
      "grad_norm": 0.36135301850127477,
      "learning_rate": 8.75018388721318e-06,
      "loss": 0.0374,
      "step": 3435
    },
    {
      "epoch": 0.9214266559399302,
      "grad_norm": 0.5109162939722136,
      "learning_rate": 8.749151802986848e-06,
      "loss": 0.0289,
      "step": 3436
    },
    {
      "epoch": 0.9216948243496916,
      "grad_norm": 0.24387996746349797,
      "learning_rate": 8.748119353715562e-06,
      "loss": 0.0292,
      "step": 3437
    },
    {
      "epoch": 0.9219629927594529,
      "grad_norm": 0.2721041202750463,
      "learning_rate": 8.747086539499844e-06,
      "loss": 0.0297,
      "step": 3438
    },
    {
      "epoch": 0.9222311611692142,
      "grad_norm": 0.5447837286688044,
      "learning_rate": 8.746053360440258e-06,
      "loss": 0.037,
      "step": 3439
    },
    {
      "epoch": 0.9224993295789756,
      "grad_norm": 0.61975082715662,
      "learning_rate": 8.745019816637404e-06,
      "loss": 0.0493,
      "step": 3440
    },
    {
      "epoch": 0.9227674979887369,
      "grad_norm": 0.44113927796891955,
      "learning_rate": 8.74398590819191e-06,
      "loss": 0.0362,
      "step": 3441
    },
    {
      "epoch": 0.9230356663984982,
      "grad_norm": 0.45648420759892344,
      "learning_rate": 8.742951635204451e-06,
      "loss": 0.0467,
      "step": 3442
    },
    {
      "epoch": 0.9233038348082596,
      "grad_norm": 0.40932278296883784,
      "learning_rate": 8.741916997775727e-06,
      "loss": 0.0339,
      "step": 3443
    },
    {
      "epoch": 0.9235720032180209,
      "grad_norm": 0.3230665570246833,
      "learning_rate": 8.74088199600648e-06,
      "loss": 0.0367,
      "step": 3444
    },
    {
      "epoch": 0.9238401716277822,
      "grad_norm": 0.33853428956400516,
      "learning_rate": 8.739846629997483e-06,
      "loss": 0.0399,
      "step": 3445
    },
    {
      "epoch": 0.9241083400375436,
      "grad_norm": 0.34244815286203717,
      "learning_rate": 8.738810899849552e-06,
      "loss": 0.0379,
      "step": 3446
    },
    {
      "epoch": 0.9243765084473049,
      "grad_norm": 0.38652611343802634,
      "learning_rate": 8.73777480566353e-06,
      "loss": 0.0371,
      "step": 3447
    },
    {
      "epoch": 0.9246446768570662,
      "grad_norm": 0.3622205138979121,
      "learning_rate": 8.736738347540296e-06,
      "loss": 0.0356,
      "step": 3448
    },
    {
      "epoch": 0.9249128452668276,
      "grad_norm": 0.2430869320434499,
      "learning_rate": 8.73570152558077e-06,
      "loss": 0.023,
      "step": 3449
    },
    {
      "epoch": 0.9251810136765889,
      "grad_norm": 0.32667291740777665,
      "learning_rate": 8.734664339885908e-06,
      "loss": 0.0274,
      "step": 3450
    },
    {
      "epoch": 0.9254491820863502,
      "grad_norm": 0.5214866564969718,
      "learning_rate": 8.733626790556692e-06,
      "loss": 0.0341,
      "step": 3451
    },
    {
      "epoch": 0.9257173504961116,
      "grad_norm": 0.4031660735112484,
      "learning_rate": 8.732588877694148e-06,
      "loss": 0.044,
      "step": 3452
    },
    {
      "epoch": 0.9259855189058729,
      "grad_norm": 0.5090607295337447,
      "learning_rate": 8.731550601399336e-06,
      "loss": 0.04,
      "step": 3453
    },
    {
      "epoch": 0.9262536873156342,
      "grad_norm": 0.2754062649743409,
      "learning_rate": 8.730511961773347e-06,
      "loss": 0.0331,
      "step": 3454
    },
    {
      "epoch": 0.9265218557253956,
      "grad_norm": 0.34410703272911203,
      "learning_rate": 8.729472958917316e-06,
      "loss": 0.0283,
      "step": 3455
    },
    {
      "epoch": 0.9267900241351569,
      "grad_norm": 0.28724429531507517,
      "learning_rate": 8.728433592932402e-06,
      "loss": 0.0293,
      "step": 3456
    },
    {
      "epoch": 0.9270581925449182,
      "grad_norm": 0.4618735599154895,
      "learning_rate": 8.727393863919809e-06,
      "loss": 0.0383,
      "step": 3457
    },
    {
      "epoch": 0.9273263609546796,
      "grad_norm": 0.32350893327829805,
      "learning_rate": 8.72635377198077e-06,
      "loss": 0.0288,
      "step": 3458
    },
    {
      "epoch": 0.9275945293644409,
      "grad_norm": 0.41559622051719486,
      "learning_rate": 8.725313317216558e-06,
      "loss": 0.0341,
      "step": 3459
    },
    {
      "epoch": 0.9278626977742022,
      "grad_norm": 0.4024445185653892,
      "learning_rate": 8.724272499728479e-06,
      "loss": 0.0321,
      "step": 3460
    },
    {
      "epoch": 0.9281308661839636,
      "grad_norm": 0.6736470771994812,
      "learning_rate": 8.723231319617875e-06,
      "loss": 0.0308,
      "step": 3461
    },
    {
      "epoch": 0.9283990345937249,
      "grad_norm": 0.4698535114534645,
      "learning_rate": 8.72218977698612e-06,
      "loss": 0.0445,
      "step": 3462
    },
    {
      "epoch": 0.9286672030034862,
      "grad_norm": 0.4191570139774224,
      "learning_rate": 8.721147871934632e-06,
      "loss": 0.0299,
      "step": 3463
    },
    {
      "epoch": 0.9289353714132476,
      "grad_norm": 0.2745301507101826,
      "learning_rate": 8.720105604564853e-06,
      "loss": 0.0312,
      "step": 3464
    },
    {
      "epoch": 0.9292035398230089,
      "grad_norm": 0.5027932031584817,
      "learning_rate": 8.71906297497827e-06,
      "loss": 0.0348,
      "step": 3465
    },
    {
      "epoch": 0.9294717082327701,
      "grad_norm": 0.3461598921463335,
      "learning_rate": 8.718019983276398e-06,
      "loss": 0.0283,
      "step": 3466
    },
    {
      "epoch": 0.9297398766425315,
      "grad_norm": 0.4111005201466307,
      "learning_rate": 8.716976629560791e-06,
      "loss": 0.0446,
      "step": 3467
    },
    {
      "epoch": 0.9300080450522928,
      "grad_norm": 0.22090084559272372,
      "learning_rate": 8.715932913933039e-06,
      "loss": 0.0235,
      "step": 3468
    },
    {
      "epoch": 0.9302762134620541,
      "grad_norm": 0.29757557387345096,
      "learning_rate": 8.714888836494762e-06,
      "loss": 0.0316,
      "step": 3469
    },
    {
      "epoch": 0.9305443818718155,
      "grad_norm": 0.3756043675996844,
      "learning_rate": 8.713844397347624e-06,
      "loss": 0.0285,
      "step": 3470
    },
    {
      "epoch": 0.9308125502815768,
      "grad_norm": 0.3420941353744393,
      "learning_rate": 8.712799596593317e-06,
      "loss": 0.0322,
      "step": 3471
    },
    {
      "epoch": 0.9310807186913381,
      "grad_norm": 0.38742701931654105,
      "learning_rate": 8.71175443433357e-06,
      "loss": 0.0269,
      "step": 3472
    },
    {
      "epoch": 0.9313488871010995,
      "grad_norm": 0.3039553662019852,
      "learning_rate": 8.71070891067015e-06,
      "loss": 0.0321,
      "step": 3473
    },
    {
      "epoch": 0.9316170555108608,
      "grad_norm": 0.2603762774187866,
      "learning_rate": 8.709663025704852e-06,
      "loss": 0.0257,
      "step": 3474
    },
    {
      "epoch": 0.9318852239206221,
      "grad_norm": 0.4011365345342904,
      "learning_rate": 8.708616779539518e-06,
      "loss": 0.0433,
      "step": 3475
    },
    {
      "epoch": 0.9321533923303835,
      "grad_norm": 0.357826966875023,
      "learning_rate": 8.707570172276013e-06,
      "loss": 0.0374,
      "step": 3476
    },
    {
      "epoch": 0.9324215607401448,
      "grad_norm": 0.5352465010550165,
      "learning_rate": 8.706523204016242e-06,
      "loss": 0.0348,
      "step": 3477
    },
    {
      "epoch": 0.9326897291499061,
      "grad_norm": 0.4656758186023748,
      "learning_rate": 8.705475874862149e-06,
      "loss": 0.0418,
      "step": 3478
    },
    {
      "epoch": 0.9329578975596675,
      "grad_norm": 0.2819884277921083,
      "learning_rate": 8.704428184915706e-06,
      "loss": 0.0291,
      "step": 3479
    },
    {
      "epoch": 0.9332260659694288,
      "grad_norm": 0.3788414876227959,
      "learning_rate": 8.703380134278928e-06,
      "loss": 0.0397,
      "step": 3480
    },
    {
      "epoch": 0.9334942343791901,
      "grad_norm": 0.35115623110238153,
      "learning_rate": 8.702331723053856e-06,
      "loss": 0.0287,
      "step": 3481
    },
    {
      "epoch": 0.9337624027889515,
      "grad_norm": 0.3830012409872812,
      "learning_rate": 8.701282951342576e-06,
      "loss": 0.0275,
      "step": 3482
    },
    {
      "epoch": 0.9340305711987128,
      "grad_norm": 0.38353914155370866,
      "learning_rate": 8.7002338192472e-06,
      "loss": 0.0297,
      "step": 3483
    },
    {
      "epoch": 0.9342987396084741,
      "grad_norm": 0.3548797181718097,
      "learning_rate": 8.699184326869883e-06,
      "loss": 0.0306,
      "step": 3484
    },
    {
      "epoch": 0.9345669080182355,
      "grad_norm": 0.46752604753855237,
      "learning_rate": 8.698134474312808e-06,
      "loss": 0.0293,
      "step": 3485
    },
    {
      "epoch": 0.9348350764279968,
      "grad_norm": 0.5277310151457333,
      "learning_rate": 8.697084261678198e-06,
      "loss": 0.028,
      "step": 3486
    },
    {
      "epoch": 0.9351032448377581,
      "grad_norm": 0.5228027595623886,
      "learning_rate": 8.696033689068309e-06,
      "loss": 0.0269,
      "step": 3487
    },
    {
      "epoch": 0.9353714132475195,
      "grad_norm": 0.2855481317424195,
      "learning_rate": 8.69498275658543e-06,
      "loss": 0.0282,
      "step": 3488
    },
    {
      "epoch": 0.9356395816572808,
      "grad_norm": 0.46545715257434983,
      "learning_rate": 8.693931464331893e-06,
      "loss": 0.0429,
      "step": 3489
    },
    {
      "epoch": 0.9359077500670421,
      "grad_norm": 0.33826932897132,
      "learning_rate": 8.692879812410056e-06,
      "loss": 0.0365,
      "step": 3490
    },
    {
      "epoch": 0.9361759184768035,
      "grad_norm": 0.27034163115876086,
      "learning_rate": 8.691827800922314e-06,
      "loss": 0.018,
      "step": 3491
    },
    {
      "epoch": 0.9364440868865648,
      "grad_norm": 0.2966105529902935,
      "learning_rate": 8.690775429971103e-06,
      "loss": 0.0299,
      "step": 3492
    },
    {
      "epoch": 0.9367122552963261,
      "grad_norm": 0.5051992027831839,
      "learning_rate": 8.689722699658887e-06,
      "loss": 0.0501,
      "step": 3493
    },
    {
      "epoch": 0.9369804237060875,
      "grad_norm": 0.5659243650659339,
      "learning_rate": 8.688669610088168e-06,
      "loss": 0.0337,
      "step": 3494
    },
    {
      "epoch": 0.9372485921158488,
      "grad_norm": 0.5983067254823169,
      "learning_rate": 8.68761616136148e-06,
      "loss": 0.0476,
      "step": 3495
    },
    {
      "epoch": 0.9375167605256101,
      "grad_norm": 0.3767861591315911,
      "learning_rate": 8.686562353581398e-06,
      "loss": 0.0356,
      "step": 3496
    },
    {
      "epoch": 0.9377849289353715,
      "grad_norm": 0.5100621616522427,
      "learning_rate": 8.685508186850529e-06,
      "loss": 0.0537,
      "step": 3497
    },
    {
      "epoch": 0.9380530973451328,
      "grad_norm": 0.2526780218401074,
      "learning_rate": 8.684453661271511e-06,
      "loss": 0.0244,
      "step": 3498
    },
    {
      "epoch": 0.938321265754894,
      "grad_norm": 0.414386692575756,
      "learning_rate": 8.683398776947022e-06,
      "loss": 0.0294,
      "step": 3499
    },
    {
      "epoch": 0.9385894341646555,
      "grad_norm": 0.36250574876565156,
      "learning_rate": 8.682343533979774e-06,
      "loss": 0.0343,
      "step": 3500
    },
    {
      "epoch": 0.9388576025744167,
      "grad_norm": 0.3349926664960653,
      "learning_rate": 8.681287932472512e-06,
      "loss": 0.0254,
      "step": 3501
    },
    {
      "epoch": 0.939125770984178,
      "grad_norm": 0.4367327587110873,
      "learning_rate": 8.680231972528019e-06,
      "loss": 0.0312,
      "step": 3502
    },
    {
      "epoch": 0.9393939393939394,
      "grad_norm": 0.3143020796655468,
      "learning_rate": 8.679175654249107e-06,
      "loss": 0.0347,
      "step": 3503
    },
    {
      "epoch": 0.9396621078037007,
      "grad_norm": 0.2827420421407842,
      "learning_rate": 8.678118977738632e-06,
      "loss": 0.0323,
      "step": 3504
    },
    {
      "epoch": 0.939930276213462,
      "grad_norm": 0.7881596764792689,
      "learning_rate": 8.677061943099476e-06,
      "loss": 0.0229,
      "step": 3505
    },
    {
      "epoch": 0.9401984446232234,
      "grad_norm": 0.22302746717536126,
      "learning_rate": 8.676004550434563e-06,
      "loss": 0.026,
      "step": 3506
    },
    {
      "epoch": 0.9404666130329847,
      "grad_norm": 0.3731899474737974,
      "learning_rate": 8.674946799846844e-06,
      "loss": 0.031,
      "step": 3507
    },
    {
      "epoch": 0.940734781442746,
      "grad_norm": 0.33740896175859675,
      "learning_rate": 8.673888691439314e-06,
      "loss": 0.0293,
      "step": 3508
    },
    {
      "epoch": 0.9410029498525073,
      "grad_norm": 0.26470976805274643,
      "learning_rate": 8.672830225314994e-06,
      "loss": 0.0232,
      "step": 3509
    },
    {
      "epoch": 0.9412711182622687,
      "grad_norm": 0.3627384719239412,
      "learning_rate": 8.671771401576948e-06,
      "loss": 0.0356,
      "step": 3510
    },
    {
      "epoch": 0.94153928667203,
      "grad_norm": 0.2843430303576068,
      "learning_rate": 8.670712220328268e-06,
      "loss": 0.0332,
      "step": 3511
    },
    {
      "epoch": 0.9418074550817913,
      "grad_norm": 0.3531038305328944,
      "learning_rate": 8.669652681672085e-06,
      "loss": 0.038,
      "step": 3512
    },
    {
      "epoch": 0.9420756234915527,
      "grad_norm": 0.5380350044943221,
      "learning_rate": 8.668592785711562e-06,
      "loss": 0.0383,
      "step": 3513
    },
    {
      "epoch": 0.942343791901314,
      "grad_norm": 0.3625632128159095,
      "learning_rate": 8.667532532549901e-06,
      "loss": 0.0389,
      "step": 3514
    },
    {
      "epoch": 0.9426119603110753,
      "grad_norm": 0.5321597287764244,
      "learning_rate": 8.666471922290334e-06,
      "loss": 0.0321,
      "step": 3515
    },
    {
      "epoch": 0.9428801287208367,
      "grad_norm": 0.2629506409769947,
      "learning_rate": 8.665410955036129e-06,
      "loss": 0.0262,
      "step": 3516
    },
    {
      "epoch": 0.943148297130598,
      "grad_norm": 0.6597690825186551,
      "learning_rate": 8.664349630890592e-06,
      "loss": 0.0359,
      "step": 3517
    },
    {
      "epoch": 0.9434164655403593,
      "grad_norm": 0.3440416485674077,
      "learning_rate": 8.663287949957059e-06,
      "loss": 0.0284,
      "step": 3518
    },
    {
      "epoch": 0.9436846339501207,
      "grad_norm": 0.5020157659771631,
      "learning_rate": 8.662225912338906e-06,
      "loss": 0.0456,
      "step": 3519
    },
    {
      "epoch": 0.943952802359882,
      "grad_norm": 0.3237051835938411,
      "learning_rate": 8.661163518139537e-06,
      "loss": 0.0376,
      "step": 3520
    },
    {
      "epoch": 0.9442209707696433,
      "grad_norm": 0.6234433385623344,
      "learning_rate": 8.660100767462398e-06,
      "loss": 0.047,
      "step": 3521
    },
    {
      "epoch": 0.9444891391794047,
      "grad_norm": 0.3018094850498469,
      "learning_rate": 8.659037660410965e-06,
      "loss": 0.0221,
      "step": 3522
    },
    {
      "epoch": 0.944757307589166,
      "grad_norm": 0.4534977464659579,
      "learning_rate": 8.657974197088747e-06,
      "loss": 0.0286,
      "step": 3523
    },
    {
      "epoch": 0.9450254759989273,
      "grad_norm": 0.4638779444898991,
      "learning_rate": 8.656910377599294e-06,
      "loss": 0.0519,
      "step": 3524
    },
    {
      "epoch": 0.9452936444086887,
      "grad_norm": 0.3722790087068143,
      "learning_rate": 8.65584620204619e-06,
      "loss": 0.038,
      "step": 3525
    },
    {
      "epoch": 0.94556181281845,
      "grad_norm": 0.4024395799425294,
      "learning_rate": 8.654781670533043e-06,
      "loss": 0.0262,
      "step": 3526
    },
    {
      "epoch": 0.9458299812282113,
      "grad_norm": 0.6244448385619118,
      "learning_rate": 8.65371678316351e-06,
      "loss": 0.041,
      "step": 3527
    },
    {
      "epoch": 0.9460981496379727,
      "grad_norm": 0.5481710451507913,
      "learning_rate": 8.652651540041272e-06,
      "loss": 0.0333,
      "step": 3528
    },
    {
      "epoch": 0.946366318047734,
      "grad_norm": 0.40051380102013634,
      "learning_rate": 8.651585941270056e-06,
      "loss": 0.0366,
      "step": 3529
    },
    {
      "epoch": 0.9466344864574953,
      "grad_norm": 0.33217248672407196,
      "learning_rate": 8.650519986953607e-06,
      "loss": 0.0245,
      "step": 3530
    },
    {
      "epoch": 0.9469026548672567,
      "grad_norm": 0.27813446074695647,
      "learning_rate": 8.64945367719572e-06,
      "loss": 0.0239,
      "step": 3531
    },
    {
      "epoch": 0.947170823277018,
      "grad_norm": 0.3921792402048546,
      "learning_rate": 8.64838701210022e-06,
      "loss": 0.0327,
      "step": 3532
    },
    {
      "epoch": 0.9474389916867793,
      "grad_norm": 0.32849280702804706,
      "learning_rate": 8.647319991770959e-06,
      "loss": 0.0371,
      "step": 3533
    },
    {
      "epoch": 0.9477071600965407,
      "grad_norm": 0.34770606469846554,
      "learning_rate": 8.646252616311836e-06,
      "loss": 0.0304,
      "step": 3534
    },
    {
      "epoch": 0.947975328506302,
      "grad_norm": 0.28908639654540036,
      "learning_rate": 8.645184885826776e-06,
      "loss": 0.0386,
      "step": 3535
    },
    {
      "epoch": 0.9482434969160632,
      "grad_norm": 0.23792373829134256,
      "learning_rate": 8.644116800419741e-06,
      "loss": 0.0316,
      "step": 3536
    },
    {
      "epoch": 0.9485116653258246,
      "grad_norm": 0.3191023292441791,
      "learning_rate": 8.64304836019473e-06,
      "loss": 0.0365,
      "step": 3537
    },
    {
      "epoch": 0.9487798337355859,
      "grad_norm": 0.6874460414117828,
      "learning_rate": 8.64197956525577e-06,
      "loss": 0.029,
      "step": 3538
    },
    {
      "epoch": 0.9490480021453472,
      "grad_norm": 0.43920993916557255,
      "learning_rate": 8.640910415706931e-06,
      "loss": 0.0431,
      "step": 3539
    },
    {
      "epoch": 0.9493161705551086,
      "grad_norm": 0.45930704010464496,
      "learning_rate": 8.639840911652309e-06,
      "loss": 0.0302,
      "step": 3540
    },
    {
      "epoch": 0.9495843389648699,
      "grad_norm": 0.36442804312876265,
      "learning_rate": 8.638771053196043e-06,
      "loss": 0.0341,
      "step": 3541
    },
    {
      "epoch": 0.9498525073746312,
      "grad_norm": 0.4167081809713381,
      "learning_rate": 8.6377008404423e-06,
      "loss": 0.0343,
      "step": 3542
    },
    {
      "epoch": 0.9501206757843926,
      "grad_norm": 0.5302044618722979,
      "learning_rate": 8.636630273495284e-06,
      "loss": 0.0259,
      "step": 3543
    },
    {
      "epoch": 0.9503888441941539,
      "grad_norm": 0.3762756929809467,
      "learning_rate": 8.635559352459232e-06,
      "loss": 0.0306,
      "step": 3544
    },
    {
      "epoch": 0.9506570126039152,
      "grad_norm": 0.3277359014067011,
      "learning_rate": 8.634488077438419e-06,
      "loss": 0.0304,
      "step": 3545
    },
    {
      "epoch": 0.9509251810136766,
      "grad_norm": 0.3348328328896239,
      "learning_rate": 8.63341644853715e-06,
      "loss": 0.0269,
      "step": 3546
    },
    {
      "epoch": 0.9511933494234379,
      "grad_norm": 0.40890586199432016,
      "learning_rate": 8.63234446585977e-06,
      "loss": 0.0315,
      "step": 3547
    },
    {
      "epoch": 0.9514615178331992,
      "grad_norm": 0.5167310419009069,
      "learning_rate": 8.631272129510653e-06,
      "loss": 0.0425,
      "step": 3548
    },
    {
      "epoch": 0.9517296862429606,
      "grad_norm": 0.24842406418806748,
      "learning_rate": 8.630199439594209e-06,
      "loss": 0.0278,
      "step": 3549
    },
    {
      "epoch": 0.9519978546527219,
      "grad_norm": 0.3142024869226592,
      "learning_rate": 8.629126396214884e-06,
      "loss": 0.0303,
      "step": 3550
    },
    {
      "epoch": 0.9522660230624832,
      "grad_norm": 0.6790042026125757,
      "learning_rate": 8.628052999477156e-06,
      "loss": 0.0459,
      "step": 3551
    },
    {
      "epoch": 0.9525341914722446,
      "grad_norm": 0.6341699008332742,
      "learning_rate": 8.626979249485538e-06,
      "loss": 0.0265,
      "step": 3552
    },
    {
      "epoch": 0.9528023598820059,
      "grad_norm": 0.40537804865859745,
      "learning_rate": 8.625905146344585e-06,
      "loss": 0.0323,
      "step": 3553
    },
    {
      "epoch": 0.9530705282917672,
      "grad_norm": 0.7606901496524828,
      "learning_rate": 8.624830690158869e-06,
      "loss": 0.0447,
      "step": 3554
    },
    {
      "epoch": 0.9533386967015286,
      "grad_norm": 0.24681365532420824,
      "learning_rate": 8.623755881033016e-06,
      "loss": 0.0249,
      "step": 3555
    },
    {
      "epoch": 0.9536068651112899,
      "grad_norm": 0.3451250793849473,
      "learning_rate": 8.622680719071673e-06,
      "loss": 0.0301,
      "step": 3556
    },
    {
      "epoch": 0.9538750335210512,
      "grad_norm": 0.45135034503313176,
      "learning_rate": 8.621605204379525e-06,
      "loss": 0.0298,
      "step": 3557
    },
    {
      "epoch": 0.9541432019308126,
      "grad_norm": 0.3253546747082488,
      "learning_rate": 8.620529337061295e-06,
      "loss": 0.0338,
      "step": 3558
    },
    {
      "epoch": 0.9544113703405739,
      "grad_norm": 0.4531375412289693,
      "learning_rate": 8.619453117221737e-06,
      "loss": 0.0386,
      "step": 3559
    },
    {
      "epoch": 0.9546795387503352,
      "grad_norm": 0.31633192625082573,
      "learning_rate": 8.618376544965636e-06,
      "loss": 0.0359,
      "step": 3560
    },
    {
      "epoch": 0.9549477071600966,
      "grad_norm": 0.3033622808161688,
      "learning_rate": 8.617299620397817e-06,
      "loss": 0.0286,
      "step": 3561
    },
    {
      "epoch": 0.9552158755698579,
      "grad_norm": 0.8627661596775185,
      "learning_rate": 8.61622234362314e-06,
      "loss": 0.0397,
      "step": 3562
    },
    {
      "epoch": 0.9554840439796192,
      "grad_norm": 0.4169885695082447,
      "learning_rate": 8.615144714746493e-06,
      "loss": 0.0367,
      "step": 3563
    },
    {
      "epoch": 0.9557522123893806,
      "grad_norm": 0.26010141517055063,
      "learning_rate": 8.614066733872803e-06,
      "loss": 0.0236,
      "step": 3564
    },
    {
      "epoch": 0.9560203807991419,
      "grad_norm": 0.3150500922750423,
      "learning_rate": 8.61298840110703e-06,
      "loss": 0.0315,
      "step": 3565
    },
    {
      "epoch": 0.9562885492089032,
      "grad_norm": 0.44559932471656305,
      "learning_rate": 8.61190971655417e-06,
      "loss": 0.0316,
      "step": 3566
    },
    {
      "epoch": 0.9565567176186646,
      "grad_norm": 0.43338971782395724,
      "learning_rate": 8.61083068031925e-06,
      "loss": 0.0434,
      "step": 3567
    },
    {
      "epoch": 0.9568248860284259,
      "grad_norm": 0.4800334563938576,
      "learning_rate": 8.609751292507332e-06,
      "loss": 0.0334,
      "step": 3568
    },
    {
      "epoch": 0.9570930544381872,
      "grad_norm": 0.6113471916596498,
      "learning_rate": 8.608671553223515e-06,
      "loss": 0.0446,
      "step": 3569
    },
    {
      "epoch": 0.9573612228479486,
      "grad_norm": 0.4067172924078759,
      "learning_rate": 8.60759146257293e-06,
      "loss": 0.0283,
      "step": 3570
    },
    {
      "epoch": 0.9576293912577098,
      "grad_norm": 0.24974496459760726,
      "learning_rate": 8.606511020660743e-06,
      "loss": 0.0351,
      "step": 3571
    },
    {
      "epoch": 0.9578975596674711,
      "grad_norm": 0.23015473531430836,
      "learning_rate": 8.605430227592152e-06,
      "loss": 0.0197,
      "step": 3572
    },
    {
      "epoch": 0.9581657280772325,
      "grad_norm": 0.5053490005188851,
      "learning_rate": 8.604349083472392e-06,
      "loss": 0.0304,
      "step": 3573
    },
    {
      "epoch": 0.9584338964869938,
      "grad_norm": 0.4218803396825629,
      "learning_rate": 8.603267588406732e-06,
      "loss": 0.033,
      "step": 3574
    },
    {
      "epoch": 0.9587020648967551,
      "grad_norm": 0.3185953959767205,
      "learning_rate": 8.602185742500473e-06,
      "loss": 0.0322,
      "step": 3575
    },
    {
      "epoch": 0.9589702333065165,
      "grad_norm": 0.2783640410490793,
      "learning_rate": 8.601103545858951e-06,
      "loss": 0.0307,
      "step": 3576
    },
    {
      "epoch": 0.9592384017162778,
      "grad_norm": 0.3495298034297191,
      "learning_rate": 8.600020998587537e-06,
      "loss": 0.0264,
      "step": 3577
    },
    {
      "epoch": 0.9595065701260391,
      "grad_norm": 0.28682301255627823,
      "learning_rate": 8.59893810079164e-06,
      "loss": 0.0283,
      "step": 3578
    },
    {
      "epoch": 0.9597747385358005,
      "grad_norm": 0.28204027645970686,
      "learning_rate": 8.59785485257669e-06,
      "loss": 0.0245,
      "step": 3579
    },
    {
      "epoch": 0.9600429069455618,
      "grad_norm": 0.3151392967216016,
      "learning_rate": 8.596771254048167e-06,
      "loss": 0.031,
      "step": 3580
    },
    {
      "epoch": 0.9603110753553231,
      "grad_norm": 0.49012244798563714,
      "learning_rate": 8.595687305311578e-06,
      "loss": 0.0422,
      "step": 3581
    },
    {
      "epoch": 0.9605792437650845,
      "grad_norm": 0.6291615130829359,
      "learning_rate": 8.59460300647246e-06,
      "loss": 0.0417,
      "step": 3582
    },
    {
      "epoch": 0.9608474121748458,
      "grad_norm": 0.7142187479303892,
      "learning_rate": 8.593518357636391e-06,
      "loss": 0.0328,
      "step": 3583
    },
    {
      "epoch": 0.9611155805846071,
      "grad_norm": 0.3275506431842715,
      "learning_rate": 8.592433358908983e-06,
      "loss": 0.031,
      "step": 3584
    },
    {
      "epoch": 0.9613837489943685,
      "grad_norm": 0.2988858656577797,
      "learning_rate": 8.591348010395874e-06,
      "loss": 0.0275,
      "step": 3585
    },
    {
      "epoch": 0.9616519174041298,
      "grad_norm": 0.3757247528570066,
      "learning_rate": 8.590262312202745e-06,
      "loss": 0.0306,
      "step": 3586
    },
    {
      "epoch": 0.9619200858138911,
      "grad_norm": 0.273062698872959,
      "learning_rate": 8.589176264435306e-06,
      "loss": 0.0317,
      "step": 3587
    },
    {
      "epoch": 0.9621882542236525,
      "grad_norm": 0.4989624237172853,
      "learning_rate": 8.588089867199302e-06,
      "loss": 0.0285,
      "step": 3588
    },
    {
      "epoch": 0.9624564226334138,
      "grad_norm": 0.3460364337520605,
      "learning_rate": 8.587003120600516e-06,
      "loss": 0.0303,
      "step": 3589
    },
    {
      "epoch": 0.9627245910431751,
      "grad_norm": 0.6028209878297713,
      "learning_rate": 8.58591602474476e-06,
      "loss": 0.035,
      "step": 3590
    },
    {
      "epoch": 0.9629927594529365,
      "grad_norm": 0.3627544850398372,
      "learning_rate": 8.58482857973788e-06,
      "loss": 0.0306,
      "step": 3591
    },
    {
      "epoch": 0.9632609278626978,
      "grad_norm": 0.5036391619583225,
      "learning_rate": 8.58374078568576e-06,
      "loss": 0.0286,
      "step": 3592
    },
    {
      "epoch": 0.9635290962724591,
      "grad_norm": 0.4005179781099706,
      "learning_rate": 8.582652642694311e-06,
      "loss": 0.037,
      "step": 3593
    },
    {
      "epoch": 0.9637972646822205,
      "grad_norm": 0.3949093922850974,
      "learning_rate": 8.581564150869488e-06,
      "loss": 0.0326,
      "step": 3594
    },
    {
      "epoch": 0.9640654330919818,
      "grad_norm": 0.28835274838047764,
      "learning_rate": 8.580475310317273e-06,
      "loss": 0.0326,
      "step": 3595
    },
    {
      "epoch": 0.9643336015017431,
      "grad_norm": 0.3104035876542182,
      "learning_rate": 8.579386121143682e-06,
      "loss": 0.0288,
      "step": 3596
    },
    {
      "epoch": 0.9646017699115044,
      "grad_norm": 0.34041386153468695,
      "learning_rate": 8.578296583454767e-06,
      "loss": 0.0333,
      "step": 3597
    },
    {
      "epoch": 0.9648699383212658,
      "grad_norm": 0.500109460322815,
      "learning_rate": 8.577206697356614e-06,
      "loss": 0.0441,
      "step": 3598
    },
    {
      "epoch": 0.9651381067310271,
      "grad_norm": 0.3013686309288792,
      "learning_rate": 8.576116462955345e-06,
      "loss": 0.0184,
      "step": 3599
    },
    {
      "epoch": 0.9654062751407884,
      "grad_norm": 0.5428248015501635,
      "learning_rate": 8.575025880357109e-06,
      "loss": 0.0324,
      "step": 3600
    },
    {
      "epoch": 0.9656744435505498,
      "grad_norm": 0.30118618146595383,
      "learning_rate": 8.573934949668093e-06,
      "loss": 0.0341,
      "step": 3601
    },
    {
      "epoch": 0.9659426119603111,
      "grad_norm": 0.32486728670072945,
      "learning_rate": 8.572843670994521e-06,
      "loss": 0.0462,
      "step": 3602
    },
    {
      "epoch": 0.9662107803700724,
      "grad_norm": 0.39848629559963616,
      "learning_rate": 8.571752044442645e-06,
      "loss": 0.041,
      "step": 3603
    },
    {
      "epoch": 0.9664789487798338,
      "grad_norm": 0.3016615814992247,
      "learning_rate": 8.570660070118758e-06,
      "loss": 0.0232,
      "step": 3604
    },
    {
      "epoch": 0.966747117189595,
      "grad_norm": 0.3903766925165771,
      "learning_rate": 8.569567748129177e-06,
      "loss": 0.0304,
      "step": 3605
    },
    {
      "epoch": 0.9670152855993563,
      "grad_norm": 0.39932121786725394,
      "learning_rate": 8.568475078580262e-06,
      "loss": 0.0293,
      "step": 3606
    },
    {
      "epoch": 0.9672834540091177,
      "grad_norm": 0.39562142160648417,
      "learning_rate": 8.567382061578404e-06,
      "loss": 0.0415,
      "step": 3607
    },
    {
      "epoch": 0.967551622418879,
      "grad_norm": 0.26051531713373094,
      "learning_rate": 8.566288697230024e-06,
      "loss": 0.0256,
      "step": 3608
    },
    {
      "epoch": 0.9678197908286403,
      "grad_norm": 0.7105109594981944,
      "learning_rate": 8.565194985641582e-06,
      "loss": 0.0333,
      "step": 3609
    },
    {
      "epoch": 0.9680879592384017,
      "grad_norm": 0.3720483660672912,
      "learning_rate": 8.564100926919572e-06,
      "loss": 0.0451,
      "step": 3610
    },
    {
      "epoch": 0.968356127648163,
      "grad_norm": 0.784727960567823,
      "learning_rate": 8.563006521170516e-06,
      "loss": 0.034,
      "step": 3611
    },
    {
      "epoch": 0.9686242960579243,
      "grad_norm": 0.6062223015771004,
      "learning_rate": 8.561911768500974e-06,
      "loss": 0.032,
      "step": 3612
    },
    {
      "epoch": 0.9688924644676857,
      "grad_norm": 0.39504268843170937,
      "learning_rate": 8.56081666901754e-06,
      "loss": 0.0364,
      "step": 3613
    },
    {
      "epoch": 0.969160632877447,
      "grad_norm": 0.3030056471490794,
      "learning_rate": 8.559721222826841e-06,
      "loss": 0.0237,
      "step": 3614
    },
    {
      "epoch": 0.9694288012872083,
      "grad_norm": 0.7599004933675892,
      "learning_rate": 8.558625430035537e-06,
      "loss": 0.0305,
      "step": 3615
    },
    {
      "epoch": 0.9696969696969697,
      "grad_norm": 0.343388948770909,
      "learning_rate": 8.557529290750324e-06,
      "loss": 0.0311,
      "step": 3616
    },
    {
      "epoch": 0.969965138106731,
      "grad_norm": 0.3402353120384854,
      "learning_rate": 8.556432805077927e-06,
      "loss": 0.0476,
      "step": 3617
    },
    {
      "epoch": 0.9702333065164923,
      "grad_norm": 0.4781030988722,
      "learning_rate": 8.555335973125111e-06,
      "loss": 0.0382,
      "step": 3618
    },
    {
      "epoch": 0.9705014749262537,
      "grad_norm": 0.4992659874134966,
      "learning_rate": 8.554238794998672e-06,
      "loss": 0.0397,
      "step": 3619
    },
    {
      "epoch": 0.970769643336015,
      "grad_norm": 0.31974148482946496,
      "learning_rate": 8.553141270805435e-06,
      "loss": 0.0257,
      "step": 3620
    },
    {
      "epoch": 0.9710378117457763,
      "grad_norm": 0.2328888531367756,
      "learning_rate": 8.552043400652269e-06,
      "loss": 0.0261,
      "step": 3621
    },
    {
      "epoch": 0.9713059801555377,
      "grad_norm": 0.3657806404679391,
      "learning_rate": 8.550945184646065e-06,
      "loss": 0.0373,
      "step": 3622
    },
    {
      "epoch": 0.971574148565299,
      "grad_norm": 0.32882856077966904,
      "learning_rate": 8.549846622893758e-06,
      "loss": 0.0345,
      "step": 3623
    },
    {
      "epoch": 0.9718423169750603,
      "grad_norm": 0.4022086541233728,
      "learning_rate": 8.548747715502309e-06,
      "loss": 0.0268,
      "step": 3624
    },
    {
      "epoch": 0.9721104853848217,
      "grad_norm": 0.39835162949469044,
      "learning_rate": 8.547648462578717e-06,
      "loss": 0.0363,
      "step": 3625
    },
    {
      "epoch": 0.972378653794583,
      "grad_norm": 0.330564160035514,
      "learning_rate": 8.546548864230011e-06,
      "loss": 0.0328,
      "step": 3626
    },
    {
      "epoch": 0.9726468222043443,
      "grad_norm": 0.3309434377910914,
      "learning_rate": 8.54544892056326e-06,
      "loss": 0.0231,
      "step": 3627
    },
    {
      "epoch": 0.9729149906141057,
      "grad_norm": 0.39382364885742527,
      "learning_rate": 8.544348631685561e-06,
      "loss": 0.0369,
      "step": 3628
    },
    {
      "epoch": 0.973183159023867,
      "grad_norm": 0.34978941019018345,
      "learning_rate": 8.543247997704045e-06,
      "loss": 0.0343,
      "step": 3629
    },
    {
      "epoch": 0.9734513274336283,
      "grad_norm": 0.2924714109476883,
      "learning_rate": 8.542147018725878e-06,
      "loss": 0.0248,
      "step": 3630
    },
    {
      "epoch": 0.9737194958433897,
      "grad_norm": 0.4231194015496994,
      "learning_rate": 8.54104569485826e-06,
      "loss": 0.0429,
      "step": 3631
    },
    {
      "epoch": 0.973987664253151,
      "grad_norm": 0.30205461411936785,
      "learning_rate": 8.539944026208426e-06,
      "loss": 0.0287,
      "step": 3632
    },
    {
      "epoch": 0.9742558326629123,
      "grad_norm": 0.4436855941384288,
      "learning_rate": 8.538842012883639e-06,
      "loss": 0.0253,
      "step": 3633
    },
    {
      "epoch": 0.9745240010726737,
      "grad_norm": 0.3042032110696812,
      "learning_rate": 8.537739654991201e-06,
      "loss": 0.0284,
      "step": 3634
    },
    {
      "epoch": 0.974792169482435,
      "grad_norm": 0.3740067423378039,
      "learning_rate": 8.536636952638444e-06,
      "loss": 0.0305,
      "step": 3635
    },
    {
      "epoch": 0.9750603378921963,
      "grad_norm": 0.4220913053207132,
      "learning_rate": 8.535533905932739e-06,
      "loss": 0.0402,
      "step": 3636
    },
    {
      "epoch": 0.9753285063019577,
      "grad_norm": 0.4331822385387482,
      "learning_rate": 8.534430514981483e-06,
      "loss": 0.0373,
      "step": 3637
    },
    {
      "epoch": 0.975596674711719,
      "grad_norm": 0.39750156887269056,
      "learning_rate": 8.533326779892111e-06,
      "loss": 0.0277,
      "step": 3638
    },
    {
      "epoch": 0.9758648431214803,
      "grad_norm": 1.3201139789972984,
      "learning_rate": 8.53222270077209e-06,
      "loss": 0.0479,
      "step": 3639
    },
    {
      "epoch": 0.9761330115312417,
      "grad_norm": 0.40423632278071825,
      "learning_rate": 8.531118277728925e-06,
      "loss": 0.042,
      "step": 3640
    },
    {
      "epoch": 0.976401179941003,
      "grad_norm": 0.364463374774072,
      "learning_rate": 8.530013510870148e-06,
      "loss": 0.0367,
      "step": 3641
    },
    {
      "epoch": 0.9766693483507642,
      "grad_norm": 0.2982704885536568,
      "learning_rate": 8.528908400303326e-06,
      "loss": 0.0359,
      "step": 3642
    },
    {
      "epoch": 0.9769375167605256,
      "grad_norm": 0.36406833969195906,
      "learning_rate": 8.527802946136063e-06,
      "loss": 0.036,
      "step": 3643
    },
    {
      "epoch": 0.9772056851702869,
      "grad_norm": 0.2926632869322191,
      "learning_rate": 8.526697148475993e-06,
      "loss": 0.0232,
      "step": 3644
    },
    {
      "epoch": 0.9774738535800482,
      "grad_norm": 0.3104761628676533,
      "learning_rate": 8.525591007430785e-06,
      "loss": 0.0244,
      "step": 3645
    },
    {
      "epoch": 0.9777420219898096,
      "grad_norm": 0.7557963682686898,
      "learning_rate": 8.52448452310814e-06,
      "loss": 0.0355,
      "step": 3646
    },
    {
      "epoch": 0.9780101903995709,
      "grad_norm": 0.35280727238867876,
      "learning_rate": 8.523377695615797e-06,
      "loss": 0.0334,
      "step": 3647
    },
    {
      "epoch": 0.9782783588093322,
      "grad_norm": 0.36715952591403095,
      "learning_rate": 8.52227052506152e-06,
      "loss": 0.0309,
      "step": 3648
    },
    {
      "epoch": 0.9785465272190936,
      "grad_norm": 0.3839059885291677,
      "learning_rate": 8.521163011553114e-06,
      "loss": 0.0254,
      "step": 3649
    },
    {
      "epoch": 0.9788146956288549,
      "grad_norm": 0.3918615767690337,
      "learning_rate": 8.520055155198413e-06,
      "loss": 0.0333,
      "step": 3650
    },
    {
      "epoch": 0.9790828640386162,
      "grad_norm": 0.32140123859303615,
      "learning_rate": 8.518946956105288e-06,
      "loss": 0.0289,
      "step": 3651
    },
    {
      "epoch": 0.9793510324483776,
      "grad_norm": 0.31017674200415773,
      "learning_rate": 8.51783841438164e-06,
      "loss": 0.0234,
      "step": 3652
    },
    {
      "epoch": 0.9796192008581389,
      "grad_norm": 0.29975746247711027,
      "learning_rate": 8.516729530135406e-06,
      "loss": 0.0298,
      "step": 3653
    },
    {
      "epoch": 0.9798873692679002,
      "grad_norm": 0.3268654142759821,
      "learning_rate": 8.515620303474554e-06,
      "loss": 0.0272,
      "step": 3654
    },
    {
      "epoch": 0.9801555376776616,
      "grad_norm": 0.3142866388604586,
      "learning_rate": 8.514510734507087e-06,
      "loss": 0.0227,
      "step": 3655
    },
    {
      "epoch": 0.9804237060874229,
      "grad_norm": 0.34433421524887536,
      "learning_rate": 8.513400823341042e-06,
      "loss": 0.0324,
      "step": 3656
    },
    {
      "epoch": 0.9806918744971842,
      "grad_norm": 0.26529099639928977,
      "learning_rate": 8.512290570084486e-06,
      "loss": 0.0224,
      "step": 3657
    },
    {
      "epoch": 0.9809600429069456,
      "grad_norm": 0.36606696508016767,
      "learning_rate": 8.51117997484552e-06,
      "loss": 0.0349,
      "step": 3658
    },
    {
      "epoch": 0.9812282113167069,
      "grad_norm": 0.31790399298441924,
      "learning_rate": 8.510069037732284e-06,
      "loss": 0.0314,
      "step": 3659
    },
    {
      "epoch": 0.9814963797264682,
      "grad_norm": 0.5595156414326738,
      "learning_rate": 8.508957758852943e-06,
      "loss": 0.0407,
      "step": 3660
    },
    {
      "epoch": 0.9817645481362296,
      "grad_norm": 0.28438168158365773,
      "learning_rate": 8.507846138315702e-06,
      "loss": 0.0268,
      "step": 3661
    },
    {
      "epoch": 0.9820327165459909,
      "grad_norm": 0.41442313395237773,
      "learning_rate": 8.506734176228797e-06,
      "loss": 0.0357,
      "step": 3662
    },
    {
      "epoch": 0.9823008849557522,
      "grad_norm": 0.30821055404697556,
      "learning_rate": 8.505621872700493e-06,
      "loss": 0.0373,
      "step": 3663
    },
    {
      "epoch": 0.9825690533655136,
      "grad_norm": 0.4322438902686305,
      "learning_rate": 8.504509227839097e-06,
      "loss": 0.0309,
      "step": 3664
    },
    {
      "epoch": 0.9828372217752749,
      "grad_norm": 0.24182023383172518,
      "learning_rate": 8.503396241752942e-06,
      "loss": 0.0234,
      "step": 3665
    },
    {
      "epoch": 0.9831053901850362,
      "grad_norm": 0.3329531361639742,
      "learning_rate": 8.502282914550395e-06,
      "loss": 0.0294,
      "step": 3666
    },
    {
      "epoch": 0.9833735585947976,
      "grad_norm": 0.30467797241152317,
      "learning_rate": 8.501169246339857e-06,
      "loss": 0.0271,
      "step": 3667
    },
    {
      "epoch": 0.9836417270045589,
      "grad_norm": 2.0568412835692036,
      "learning_rate": 8.500055237229768e-06,
      "loss": 0.0427,
      "step": 3668
    },
    {
      "epoch": 0.9839098954143202,
      "grad_norm": 0.298483773343428,
      "learning_rate": 8.498940887328592e-06,
      "loss": 0.0369,
      "step": 3669
    },
    {
      "epoch": 0.9841780638240816,
      "grad_norm": 0.4765706278252705,
      "learning_rate": 8.497826196744832e-06,
      "loss": 0.0252,
      "step": 3670
    },
    {
      "epoch": 0.9844462322338429,
      "grad_norm": 0.32576955375356076,
      "learning_rate": 8.496711165587021e-06,
      "loss": 0.0315,
      "step": 3671
    },
    {
      "epoch": 0.9847144006436042,
      "grad_norm": 0.2771121438909427,
      "learning_rate": 8.495595793963726e-06,
      "loss": 0.0189,
      "step": 3672
    },
    {
      "epoch": 0.9849825690533656,
      "grad_norm": 0.5073905161745529,
      "learning_rate": 8.494480081983552e-06,
      "loss": 0.0306,
      "step": 3673
    },
    {
      "epoch": 0.9852507374631269,
      "grad_norm": 0.2921129310714333,
      "learning_rate": 8.493364029755126e-06,
      "loss": 0.0303,
      "step": 3674
    },
    {
      "epoch": 0.9855189058728882,
      "grad_norm": 0.302137244795995,
      "learning_rate": 8.492247637387123e-06,
      "loss": 0.0201,
      "step": 3675
    },
    {
      "epoch": 0.9857870742826496,
      "grad_norm": 0.31201247644287644,
      "learning_rate": 8.491130904988239e-06,
      "loss": 0.0238,
      "step": 3676
    },
    {
      "epoch": 0.9860552426924108,
      "grad_norm": 0.33907238930914346,
      "learning_rate": 8.490013832667205e-06,
      "loss": 0.0333,
      "step": 3677
    },
    {
      "epoch": 0.9863234111021721,
      "grad_norm": 0.3367281057477326,
      "learning_rate": 8.488896420532793e-06,
      "loss": 0.0256,
      "step": 3678
    },
    {
      "epoch": 0.9865915795119335,
      "grad_norm": 0.4660794227660885,
      "learning_rate": 8.487778668693797e-06,
      "loss": 0.0303,
      "step": 3679
    },
    {
      "epoch": 0.9868597479216948,
      "grad_norm": 0.4623525994882106,
      "learning_rate": 8.486660577259054e-06,
      "loss": 0.0289,
      "step": 3680
    },
    {
      "epoch": 0.9871279163314561,
      "grad_norm": 0.3131601269450747,
      "learning_rate": 8.485542146337424e-06,
      "loss": 0.0285,
      "step": 3681
    },
    {
      "epoch": 0.9873960847412175,
      "grad_norm": 0.3414207166407291,
      "learning_rate": 8.484423376037813e-06,
      "loss": 0.034,
      "step": 3682
    },
    {
      "epoch": 0.9876642531509788,
      "grad_norm": 0.33735721905316585,
      "learning_rate": 8.483304266469148e-06,
      "loss": 0.0295,
      "step": 3683
    },
    {
      "epoch": 0.9879324215607401,
      "grad_norm": 0.21880821900721562,
      "learning_rate": 8.482184817740393e-06,
      "loss": 0.0258,
      "step": 3684
    },
    {
      "epoch": 0.9882005899705014,
      "grad_norm": 0.4774625732927812,
      "learning_rate": 8.481065029960548e-06,
      "loss": 0.0388,
      "step": 3685
    },
    {
      "epoch": 0.9884687583802628,
      "grad_norm": 0.3485597208300633,
      "learning_rate": 8.479944903238644e-06,
      "loss": 0.0302,
      "step": 3686
    },
    {
      "epoch": 0.9887369267900241,
      "grad_norm": 0.32281229103472747,
      "learning_rate": 8.478824437683742e-06,
      "loss": 0.0291,
      "step": 3687
    },
    {
      "epoch": 0.9890050951997854,
      "grad_norm": 0.2962120415204118,
      "learning_rate": 8.477703633404943e-06,
      "loss": 0.0257,
      "step": 3688
    },
    {
      "epoch": 0.9892732636095468,
      "grad_norm": 0.2295530027038438,
      "learning_rate": 8.476582490511373e-06,
      "loss": 0.0206,
      "step": 3689
    },
    {
      "epoch": 0.9895414320193081,
      "grad_norm": 0.27983002006826563,
      "learning_rate": 8.475461009112196e-06,
      "loss": 0.0357,
      "step": 3690
    },
    {
      "epoch": 0.9898096004290694,
      "grad_norm": 0.485469828273517,
      "learning_rate": 8.474339189316607e-06,
      "loss": 0.0256,
      "step": 3691
    },
    {
      "epoch": 0.9900777688388308,
      "grad_norm": 0.47746549093288887,
      "learning_rate": 8.473217031233835e-06,
      "loss": 0.0394,
      "step": 3692
    },
    {
      "epoch": 0.9903459372485921,
      "grad_norm": 0.35861345590939137,
      "learning_rate": 8.472094534973143e-06,
      "loss": 0.0325,
      "step": 3693
    },
    {
      "epoch": 0.9906141056583534,
      "grad_norm": 0.6419213139539572,
      "learning_rate": 8.470971700643826e-06,
      "loss": 0.0335,
      "step": 3694
    },
    {
      "epoch": 0.9908822740681148,
      "grad_norm": 0.31569654672949454,
      "learning_rate": 8.469848528355208e-06,
      "loss": 0.0342,
      "step": 3695
    },
    {
      "epoch": 0.9911504424778761,
      "grad_norm": 0.39403525569129794,
      "learning_rate": 8.46872501821665e-06,
      "loss": 0.0409,
      "step": 3696
    },
    {
      "epoch": 0.9914186108876374,
      "grad_norm": 0.29754479097228625,
      "learning_rate": 8.467601170337548e-06,
      "loss": 0.0373,
      "step": 3697
    },
    {
      "epoch": 0.9916867792973988,
      "grad_norm": 0.274223177068339,
      "learning_rate": 8.466476984827326e-06,
      "loss": 0.0267,
      "step": 3698
    },
    {
      "epoch": 0.9919549477071601,
      "grad_norm": 0.48924391873126055,
      "learning_rate": 8.465352461795443e-06,
      "loss": 0.0355,
      "step": 3699
    },
    {
      "epoch": 0.9922231161169214,
      "grad_norm": 0.4956224694892889,
      "learning_rate": 8.464227601351391e-06,
      "loss": 0.0529,
      "step": 3700
    },
    {
      "epoch": 0.9924912845266828,
      "grad_norm": 0.40193231427195764,
      "learning_rate": 8.463102403604697e-06,
      "loss": 0.0377,
      "step": 3701
    },
    {
      "epoch": 0.9927594529364441,
      "grad_norm": 0.3688838728014438,
      "learning_rate": 8.461976868664912e-06,
      "loss": 0.0303,
      "step": 3702
    },
    {
      "epoch": 0.9930276213462054,
      "grad_norm": 0.8738824779200808,
      "learning_rate": 8.460850996641635e-06,
      "loss": 0.0664,
      "step": 3703
    },
    {
      "epoch": 0.9932957897559668,
      "grad_norm": 0.3243956347912429,
      "learning_rate": 8.459724787644481e-06,
      "loss": 0.0306,
      "step": 3704
    },
    {
      "epoch": 0.9935639581657281,
      "grad_norm": 0.35510414528393397,
      "learning_rate": 8.458598241783112e-06,
      "loss": 0.0326,
      "step": 3705
    },
    {
      "epoch": 0.9938321265754894,
      "grad_norm": 0.2966461560419278,
      "learning_rate": 8.457471359167215e-06,
      "loss": 0.032,
      "step": 3706
    },
    {
      "epoch": 0.9941002949852508,
      "grad_norm": 0.4555609553132581,
      "learning_rate": 8.45634413990651e-06,
      "loss": 0.0399,
      "step": 3707
    },
    {
      "epoch": 0.9943684633950121,
      "grad_norm": 0.23939977132081122,
      "learning_rate": 8.455216584110753e-06,
      "loss": 0.0269,
      "step": 3708
    },
    {
      "epoch": 0.9946366318047734,
      "grad_norm": 0.36194181468132897,
      "learning_rate": 8.45408869188973e-06,
      "loss": 0.0414,
      "step": 3709
    },
    {
      "epoch": 0.9949048002145348,
      "grad_norm": 0.31030928597820945,
      "learning_rate": 8.452960463353262e-06,
      "loss": 0.027,
      "step": 3710
    },
    {
      "epoch": 0.995172968624296,
      "grad_norm": 0.4404306219599595,
      "learning_rate": 8.451831898611202e-06,
      "loss": 0.0377,
      "step": 3711
    },
    {
      "epoch": 0.9954411370340573,
      "grad_norm": 0.43689770072349654,
      "learning_rate": 8.450702997773435e-06,
      "loss": 0.0464,
      "step": 3712
    },
    {
      "epoch": 0.9957093054438187,
      "grad_norm": 0.35043458295306473,
      "learning_rate": 8.449573760949876e-06,
      "loss": 0.0268,
      "step": 3713
    },
    {
      "epoch": 0.99597747385358,
      "grad_norm": 0.9409836069390604,
      "learning_rate": 8.44844418825048e-06,
      "loss": 0.0427,
      "step": 3714
    },
    {
      "epoch": 0.9962456422633413,
      "grad_norm": 0.2992776504581052,
      "learning_rate": 8.447314279785228e-06,
      "loss": 0.0305,
      "step": 3715
    },
    {
      "epoch": 0.9965138106731027,
      "grad_norm": 0.320474019308063,
      "learning_rate": 8.446184035664136e-06,
      "loss": 0.0282,
      "step": 3716
    },
    {
      "epoch": 0.996781979082864,
      "grad_norm": 0.22390935275976268,
      "learning_rate": 8.445053455997256e-06,
      "loss": 0.0291,
      "step": 3717
    },
    {
      "epoch": 0.9970501474926253,
      "grad_norm": 0.30609713309834785,
      "learning_rate": 8.443922540894666e-06,
      "loss": 0.0262,
      "step": 3718
    },
    {
      "epoch": 0.9973183159023867,
      "grad_norm": 0.2915056754638032,
      "learning_rate": 8.442791290466483e-06,
      "loss": 0.0251,
      "step": 3719
    },
    {
      "epoch": 0.997586484312148,
      "grad_norm": 0.5315435716537974,
      "learning_rate": 8.441659704822853e-06,
      "loss": 0.0501,
      "step": 3720
    },
    {
      "epoch": 0.9978546527219093,
      "grad_norm": 0.2797431325832765,
      "learning_rate": 8.440527784073953e-06,
      "loss": 0.0313,
      "step": 3721
    },
    {
      "epoch": 0.9981228211316707,
      "grad_norm": 0.28667142303323095,
      "learning_rate": 8.439395528329998e-06,
      "loss": 0.028,
      "step": 3722
    },
    {
      "epoch": 0.998390989541432,
      "grad_norm": 0.3869771203519002,
      "learning_rate": 8.438262937701232e-06,
      "loss": 0.0372,
      "step": 3723
    },
    {
      "epoch": 0.9986591579511933,
      "grad_norm": 0.48439110013001574,
      "learning_rate": 8.437130012297932e-06,
      "loss": 0.0397,
      "step": 3724
    },
    {
      "epoch": 0.9989273263609547,
      "grad_norm": 0.3479838437239292,
      "learning_rate": 8.435996752230408e-06,
      "loss": 0.0364,
      "step": 3725
    },
    {
      "epoch": 0.999195494770716,
      "grad_norm": 0.3807811851975286,
      "learning_rate": 8.434863157609003e-06,
      "loss": 0.0371,
      "step": 3726
    },
    {
      "epoch": 0.9994636631804773,
      "grad_norm": 0.25942704416765444,
      "learning_rate": 8.433729228544092e-06,
      "loss": 0.0262,
      "step": 3727
    },
    {
      "epoch": 0.9997318315902387,
      "grad_norm": 0.5239163375832191,
      "learning_rate": 8.43259496514608e-06,
      "loss": 0.0325,
      "step": 3728
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.40559477338466393,
      "learning_rate": 8.431460367525412e-06,
      "loss": 0.0378,
      "step": 3729
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.0330052524805069,
      "eval_runtime": 300.123,
      "eval_samples_per_second": 83.702,
      "eval_steps_per_second": 1.309,
      "step": 3729
    },
    {
      "epoch": 1.0002681684097614,
      "grad_norm": 0.4085592416947716,
      "learning_rate": 8.430325435792557e-06,
      "loss": 0.0299,
      "step": 3730
    },
    {
      "epoch": 1.0005363368195226,
      "grad_norm": 0.46192065977986485,
      "learning_rate": 8.429190170058024e-06,
      "loss": 0.0277,
      "step": 3731
    },
    {
      "epoch": 1.000804505229284,
      "grad_norm": 0.27098473567763354,
      "learning_rate": 8.428054570432345e-06,
      "loss": 0.0214,
      "step": 3732
    },
    {
      "epoch": 1.0010726736390454,
      "grad_norm": 0.2669374902576558,
      "learning_rate": 8.426918637026097e-06,
      "loss": 0.0221,
      "step": 3733
    },
    {
      "epoch": 1.0013408420488066,
      "grad_norm": 0.3660660784561016,
      "learning_rate": 8.425782369949877e-06,
      "loss": 0.0319,
      "step": 3734
    },
    {
      "epoch": 1.001609010458568,
      "grad_norm": 0.3476119667558623,
      "learning_rate": 8.424645769314324e-06,
      "loss": 0.0219,
      "step": 3735
    },
    {
      "epoch": 1.0018771788683294,
      "grad_norm": 0.3795493525156423,
      "learning_rate": 8.423508835230106e-06,
      "loss": 0.0296,
      "step": 3736
    },
    {
      "epoch": 1.0021453472780906,
      "grad_norm": 0.24058794037465714,
      "learning_rate": 8.422371567807921e-06,
      "loss": 0.025,
      "step": 3737
    },
    {
      "epoch": 1.002413515687852,
      "grad_norm": 0.3015092909201774,
      "learning_rate": 8.421233967158503e-06,
      "loss": 0.0283,
      "step": 3738
    },
    {
      "epoch": 1.0026816840976134,
      "grad_norm": 0.40630201578035013,
      "learning_rate": 8.420096033392615e-06,
      "loss": 0.0315,
      "step": 3739
    },
    {
      "epoch": 1.0029498525073746,
      "grad_norm": 0.3218054422630295,
      "learning_rate": 8.418957766621058e-06,
      "loss": 0.0254,
      "step": 3740
    },
    {
      "epoch": 1.003218020917136,
      "grad_norm": 0.32459839233578236,
      "learning_rate": 8.417819166954659e-06,
      "loss": 0.0234,
      "step": 3741
    },
    {
      "epoch": 1.0034861893268974,
      "grad_norm": 0.42755527210206806,
      "learning_rate": 8.416680234504284e-06,
      "loss": 0.0332,
      "step": 3742
    },
    {
      "epoch": 1.0037543577366586,
      "grad_norm": 0.3576850275051218,
      "learning_rate": 8.415540969380824e-06,
      "loss": 0.0294,
      "step": 3743
    },
    {
      "epoch": 1.00402252614642,
      "grad_norm": 0.272713424689016,
      "learning_rate": 8.41440137169521e-06,
      "loss": 0.019,
      "step": 3744
    },
    {
      "epoch": 1.0042906945561814,
      "grad_norm": 0.3670898597024907,
      "learning_rate": 8.413261441558397e-06,
      "loss": 0.0323,
      "step": 3745
    },
    {
      "epoch": 1.0045588629659425,
      "grad_norm": 0.4390344580459956,
      "learning_rate": 8.41212117908138e-06,
      "loss": 0.0335,
      "step": 3746
    },
    {
      "epoch": 1.004827031375704,
      "grad_norm": 0.40433151403062706,
      "learning_rate": 8.410980584375184e-06,
      "loss": 0.0241,
      "step": 3747
    },
    {
      "epoch": 1.0050951997854654,
      "grad_norm": 0.26900384574546155,
      "learning_rate": 8.409839657550864e-06,
      "loss": 0.0235,
      "step": 3748
    },
    {
      "epoch": 1.0053633681952265,
      "grad_norm": 0.3193739196282925,
      "learning_rate": 8.408698398719509e-06,
      "loss": 0.0322,
      "step": 3749
    },
    {
      "epoch": 1.005631536604988,
      "grad_norm": 0.6659207090847489,
      "learning_rate": 8.40755680799224e-06,
      "loss": 0.0394,
      "step": 3750
    },
    {
      "epoch": 1.0058997050147493,
      "grad_norm": 0.4990929017232432,
      "learning_rate": 8.406414885480215e-06,
      "loss": 0.0326,
      "step": 3751
    },
    {
      "epoch": 1.0061678734245105,
      "grad_norm": 0.28010703533005055,
      "learning_rate": 8.405272631294612e-06,
      "loss": 0.0235,
      "step": 3752
    },
    {
      "epoch": 1.006436041834272,
      "grad_norm": 0.2762779329761583,
      "learning_rate": 8.404130045546657e-06,
      "loss": 0.0225,
      "step": 3753
    },
    {
      "epoch": 1.0067042102440333,
      "grad_norm": 0.30014224071003875,
      "learning_rate": 8.402987128347596e-06,
      "loss": 0.0254,
      "step": 3754
    },
    {
      "epoch": 1.0069723786537945,
      "grad_norm": 0.3498981961975498,
      "learning_rate": 8.401843879808713e-06,
      "loss": 0.032,
      "step": 3755
    },
    {
      "epoch": 1.007240547063556,
      "grad_norm": 0.32819334380532583,
      "learning_rate": 8.400700300041322e-06,
      "loss": 0.0348,
      "step": 3756
    },
    {
      "epoch": 1.0075087154733173,
      "grad_norm": 0.23572749407751117,
      "learning_rate": 8.399556389156773e-06,
      "loss": 0.0199,
      "step": 3757
    },
    {
      "epoch": 1.0077768838830785,
      "grad_norm": 0.40191951510931007,
      "learning_rate": 8.398412147266444e-06,
      "loss": 0.031,
      "step": 3758
    },
    {
      "epoch": 1.00804505229284,
      "grad_norm": 0.2771216370970497,
      "learning_rate": 8.397267574481746e-06,
      "loss": 0.0245,
      "step": 3759
    },
    {
      "epoch": 1.0083132207026013,
      "grad_norm": 0.41700642889115425,
      "learning_rate": 8.396122670914124e-06,
      "loss": 0.0279,
      "step": 3760
    },
    {
      "epoch": 1.0085813891123625,
      "grad_norm": 0.32155443627374597,
      "learning_rate": 8.394977436675056e-06,
      "loss": 0.0301,
      "step": 3761
    },
    {
      "epoch": 1.008849557522124,
      "grad_norm": 0.35877587432304964,
      "learning_rate": 8.393831871876046e-06,
      "loss": 0.0246,
      "step": 3762
    },
    {
      "epoch": 1.0091177259318853,
      "grad_norm": 0.47724872482040814,
      "learning_rate": 8.392685976628638e-06,
      "loss": 0.0325,
      "step": 3763
    },
    {
      "epoch": 1.0093858943416465,
      "grad_norm": 0.2064204727749518,
      "learning_rate": 8.391539751044405e-06,
      "loss": 0.0186,
      "step": 3764
    },
    {
      "epoch": 1.009654062751408,
      "grad_norm": 0.35911668532173563,
      "learning_rate": 8.39039319523495e-06,
      "loss": 0.0265,
      "step": 3765
    },
    {
      "epoch": 1.0099222311611693,
      "grad_norm": 0.3409964516919302,
      "learning_rate": 8.389246309311909e-06,
      "loss": 0.0283,
      "step": 3766
    },
    {
      "epoch": 1.0101903995709305,
      "grad_norm": 0.27576876707319053,
      "learning_rate": 8.388099093386957e-06,
      "loss": 0.0221,
      "step": 3767
    },
    {
      "epoch": 1.010458567980692,
      "grad_norm": 0.3094779445128216,
      "learning_rate": 8.38695154757179e-06,
      "loss": 0.0216,
      "step": 3768
    },
    {
      "epoch": 1.0107267363904533,
      "grad_norm": 0.30393419031360175,
      "learning_rate": 8.38580367197814e-06,
      "loss": 0.0307,
      "step": 3769
    },
    {
      "epoch": 1.0109949048002145,
      "grad_norm": 0.2738097178406625,
      "learning_rate": 8.384655466717778e-06,
      "loss": 0.0245,
      "step": 3770
    },
    {
      "epoch": 1.0112630732099759,
      "grad_norm": 0.7482951141666783,
      "learning_rate": 8.3835069319025e-06,
      "loss": 0.0231,
      "step": 3771
    },
    {
      "epoch": 1.0115312416197373,
      "grad_norm": 0.5591449305281037,
      "learning_rate": 8.382358067644132e-06,
      "loss": 0.0443,
      "step": 3772
    },
    {
      "epoch": 1.0117994100294985,
      "grad_norm": 0.22570568957075818,
      "learning_rate": 8.381208874054542e-06,
      "loss": 0.0195,
      "step": 3773
    },
    {
      "epoch": 1.0120675784392599,
      "grad_norm": 0.31093540589036334,
      "learning_rate": 8.380059351245621e-06,
      "loss": 0.0317,
      "step": 3774
    },
    {
      "epoch": 1.0123357468490213,
      "grad_norm": 0.30725587911981944,
      "learning_rate": 8.378909499329294e-06,
      "loss": 0.0228,
      "step": 3775
    },
    {
      "epoch": 1.0126039152587825,
      "grad_norm": 0.42345997836301325,
      "learning_rate": 8.377759318417519e-06,
      "loss": 0.0348,
      "step": 3776
    },
    {
      "epoch": 1.0128720836685439,
      "grad_norm": 0.28414294566052456,
      "learning_rate": 8.376608808622287e-06,
      "loss": 0.0289,
      "step": 3777
    },
    {
      "epoch": 1.0131402520783053,
      "grad_norm": 0.4654614677584702,
      "learning_rate": 8.375457970055622e-06,
      "loss": 0.0391,
      "step": 3778
    },
    {
      "epoch": 1.0134084204880665,
      "grad_norm": 0.2791578766369954,
      "learning_rate": 8.374306802829574e-06,
      "loss": 0.027,
      "step": 3779
    },
    {
      "epoch": 1.0136765888978279,
      "grad_norm": 0.3448044939649245,
      "learning_rate": 8.373155307056233e-06,
      "loss": 0.0352,
      "step": 3780
    },
    {
      "epoch": 1.0139447573075893,
      "grad_norm": 0.47713641216543373,
      "learning_rate": 8.372003482847715e-06,
      "loss": 0.0402,
      "step": 3781
    },
    {
      "epoch": 1.0142129257173504,
      "grad_norm": 0.4443335553917476,
      "learning_rate": 8.370851330316169e-06,
      "loss": 0.0331,
      "step": 3782
    },
    {
      "epoch": 1.0144810941271118,
      "grad_norm": 0.269267544885955,
      "learning_rate": 8.369698849573778e-06,
      "loss": 0.0253,
      "step": 3783
    },
    {
      "epoch": 1.0147492625368733,
      "grad_norm": 0.4586000036646455,
      "learning_rate": 8.36854604073276e-06,
      "loss": 0.028,
      "step": 3784
    },
    {
      "epoch": 1.0150174309466344,
      "grad_norm": 0.42019985201543586,
      "learning_rate": 8.367392903905354e-06,
      "loss": 0.0356,
      "step": 3785
    },
    {
      "epoch": 1.0152855993563958,
      "grad_norm": 0.38361208979041245,
      "learning_rate": 8.366239439203842e-06,
      "loss": 0.0271,
      "step": 3786
    },
    {
      "epoch": 1.0155537677661572,
      "grad_norm": 0.2922204364996332,
      "learning_rate": 8.365085646740533e-06,
      "loss": 0.0342,
      "step": 3787
    },
    {
      "epoch": 1.0158219361759184,
      "grad_norm": 0.35345300203593727,
      "learning_rate": 8.363931526627769e-06,
      "loss": 0.0365,
      "step": 3788
    },
    {
      "epoch": 1.0160901045856798,
      "grad_norm": 0.37501489821123735,
      "learning_rate": 8.362777078977922e-06,
      "loss": 0.0293,
      "step": 3789
    },
    {
      "epoch": 1.0163582729954412,
      "grad_norm": 0.30180584712850295,
      "learning_rate": 8.3616223039034e-06,
      "loss": 0.0254,
      "step": 3790
    },
    {
      "epoch": 1.0166264414052024,
      "grad_norm": 0.28679505730314925,
      "learning_rate": 8.36046720151664e-06,
      "loss": 0.0255,
      "step": 3791
    },
    {
      "epoch": 1.0168946098149638,
      "grad_norm": 0.2970591186403319,
      "learning_rate": 8.35931177193011e-06,
      "loss": 0.028,
      "step": 3792
    },
    {
      "epoch": 1.0171627782247252,
      "grad_norm": 0.2543144193869907,
      "learning_rate": 8.35815601525631e-06,
      "loss": 0.0232,
      "step": 3793
    },
    {
      "epoch": 1.0174309466344864,
      "grad_norm": 0.3018916098642797,
      "learning_rate": 8.356999931607776e-06,
      "loss": 0.0212,
      "step": 3794
    },
    {
      "epoch": 1.0176991150442478,
      "grad_norm": 0.33552176408734113,
      "learning_rate": 8.355843521097071e-06,
      "loss": 0.0389,
      "step": 3795
    },
    {
      "epoch": 1.0179672834540092,
      "grad_norm": 0.283386370700812,
      "learning_rate": 8.354686783836791e-06,
      "loss": 0.0345,
      "step": 3796
    },
    {
      "epoch": 1.0182354518637704,
      "grad_norm": 0.322363003285709,
      "learning_rate": 8.353529719939568e-06,
      "loss": 0.025,
      "step": 3797
    },
    {
      "epoch": 1.0185036202735318,
      "grad_norm": 0.3169390203133281,
      "learning_rate": 8.352372329518058e-06,
      "loss": 0.0315,
      "step": 3798
    },
    {
      "epoch": 1.0187717886832932,
      "grad_norm": 0.2941212797810095,
      "learning_rate": 8.351214612684954e-06,
      "loss": 0.0217,
      "step": 3799
    },
    {
      "epoch": 1.0190399570930544,
      "grad_norm": 0.4832800729816997,
      "learning_rate": 8.35005656955298e-06,
      "loss": 0.0327,
      "step": 3800
    },
    {
      "epoch": 1.0193081255028158,
      "grad_norm": 0.2730801793906435,
      "learning_rate": 8.348898200234894e-06,
      "loss": 0.0185,
      "step": 3801
    },
    {
      "epoch": 1.0195762939125772,
      "grad_norm": 0.3334436308894534,
      "learning_rate": 8.34773950484348e-06,
      "loss": 0.0265,
      "step": 3802
    },
    {
      "epoch": 1.0198444623223384,
      "grad_norm": 0.6072727480769846,
      "learning_rate": 8.34658048349156e-06,
      "loss": 0.0381,
      "step": 3803
    },
    {
      "epoch": 1.0201126307320998,
      "grad_norm": 0.3565312767690091,
      "learning_rate": 8.345421136291983e-06,
      "loss": 0.0309,
      "step": 3804
    },
    {
      "epoch": 1.0203807991418612,
      "grad_norm": 0.5155837498721744,
      "learning_rate": 8.34426146335763e-06,
      "loss": 0.0383,
      "step": 3805
    },
    {
      "epoch": 1.0206489675516224,
      "grad_norm": 0.2794960117448267,
      "learning_rate": 8.343101464801418e-06,
      "loss": 0.0252,
      "step": 3806
    },
    {
      "epoch": 1.0209171359613838,
      "grad_norm": 0.3065205659296354,
      "learning_rate": 8.341941140736292e-06,
      "loss": 0.0214,
      "step": 3807
    },
    {
      "epoch": 1.0211853043711452,
      "grad_norm": 0.29031315599694557,
      "learning_rate": 8.340780491275232e-06,
      "loss": 0.0256,
      "step": 3808
    },
    {
      "epoch": 1.0214534727809064,
      "grad_norm": 0.3335494786164975,
      "learning_rate": 8.339619516531243e-06,
      "loss": 0.0302,
      "step": 3809
    },
    {
      "epoch": 1.0217216411906678,
      "grad_norm": 0.3518610473281327,
      "learning_rate": 8.338458216617368e-06,
      "loss": 0.0315,
      "step": 3810
    },
    {
      "epoch": 1.0219898096004292,
      "grad_norm": 0.2766770290407197,
      "learning_rate": 8.33729659164668e-06,
      "loss": 0.0289,
      "step": 3811
    },
    {
      "epoch": 1.0222579780101904,
      "grad_norm": 0.3273592234950132,
      "learning_rate": 8.336134641732284e-06,
      "loss": 0.0206,
      "step": 3812
    },
    {
      "epoch": 1.0225261464199518,
      "grad_norm": 0.30383067605075476,
      "learning_rate": 8.334972366987315e-06,
      "loss": 0.0254,
      "step": 3813
    },
    {
      "epoch": 1.0227943148297132,
      "grad_norm": 0.26127846155456075,
      "learning_rate": 8.33380976752494e-06,
      "loss": 0.0199,
      "step": 3814
    },
    {
      "epoch": 1.0230624832394744,
      "grad_norm": 0.28937918653128086,
      "learning_rate": 8.332646843458361e-06,
      "loss": 0.0381,
      "step": 3815
    },
    {
      "epoch": 1.0233306516492358,
      "grad_norm": 0.4017441387816911,
      "learning_rate": 8.331483594900807e-06,
      "loss": 0.0292,
      "step": 3816
    },
    {
      "epoch": 1.023598820058997,
      "grad_norm": 0.7819605173276373,
      "learning_rate": 8.330320021965538e-06,
      "loss": 0.0401,
      "step": 3817
    },
    {
      "epoch": 1.0238669884687583,
      "grad_norm": 0.27368879784032657,
      "learning_rate": 8.329156124765852e-06,
      "loss": 0.0299,
      "step": 3818
    },
    {
      "epoch": 1.0241351568785197,
      "grad_norm": 0.3752563169493104,
      "learning_rate": 8.327991903415071e-06,
      "loss": 0.0348,
      "step": 3819
    },
    {
      "epoch": 1.0244033252882812,
      "grad_norm": 0.34666768827658945,
      "learning_rate": 8.326827358026554e-06,
      "loss": 0.0286,
      "step": 3820
    },
    {
      "epoch": 1.0246714936980423,
      "grad_norm": 0.28213166251876426,
      "learning_rate": 8.325662488713692e-06,
      "loss": 0.0259,
      "step": 3821
    },
    {
      "epoch": 1.0249396621078037,
      "grad_norm": 0.4862306698248169,
      "learning_rate": 8.324497295589902e-06,
      "loss": 0.035,
      "step": 3822
    },
    {
      "epoch": 1.025207830517565,
      "grad_norm": 0.35464146836772437,
      "learning_rate": 8.323331778768636e-06,
      "loss": 0.0302,
      "step": 3823
    },
    {
      "epoch": 1.0254759989273263,
      "grad_norm": 0.3762974267368971,
      "learning_rate": 8.322165938363378e-06,
      "loss": 0.026,
      "step": 3824
    },
    {
      "epoch": 1.0257441673370877,
      "grad_norm": 0.3214660876082944,
      "learning_rate": 8.320999774487644e-06,
      "loss": 0.0254,
      "step": 3825
    },
    {
      "epoch": 1.026012335746849,
      "grad_norm": 0.22857557781654275,
      "learning_rate": 8.319833287254978e-06,
      "loss": 0.0207,
      "step": 3826
    },
    {
      "epoch": 1.0262805041566103,
      "grad_norm": 0.337835605928234,
      "learning_rate": 8.31866647677896e-06,
      "loss": 0.0217,
      "step": 3827
    },
    {
      "epoch": 1.0265486725663717,
      "grad_norm": 0.30305957386391386,
      "learning_rate": 8.317499343173196e-06,
      "loss": 0.0294,
      "step": 3828
    },
    {
      "epoch": 1.026816840976133,
      "grad_norm": 0.2996183089214558,
      "learning_rate": 8.31633188655133e-06,
      "loss": 0.0233,
      "step": 3829
    },
    {
      "epoch": 1.0270850093858943,
      "grad_norm": 0.3157018484871345,
      "learning_rate": 8.315164107027032e-06,
      "loss": 0.027,
      "step": 3830
    },
    {
      "epoch": 1.0273531777956557,
      "grad_norm": 0.3795048687387756,
      "learning_rate": 8.313996004714007e-06,
      "loss": 0.0287,
      "step": 3831
    },
    {
      "epoch": 1.027621346205417,
      "grad_norm": 0.3330762033868233,
      "learning_rate": 8.31282757972599e-06,
      "loss": 0.0334,
      "step": 3832
    },
    {
      "epoch": 1.0278895146151783,
      "grad_norm": 0.4211117371737151,
      "learning_rate": 8.311658832176747e-06,
      "loss": 0.0373,
      "step": 3833
    },
    {
      "epoch": 1.0281576830249397,
      "grad_norm": 0.23944118013247687,
      "learning_rate": 8.310489762180076e-06,
      "loss": 0.0225,
      "step": 3834
    },
    {
      "epoch": 1.0284258514347009,
      "grad_norm": 0.47375174348866994,
      "learning_rate": 8.309320369849804e-06,
      "loss": 0.0248,
      "step": 3835
    },
    {
      "epoch": 1.0286940198444623,
      "grad_norm": 0.3568185149623637,
      "learning_rate": 8.308150655299795e-06,
      "loss": 0.0292,
      "step": 3836
    },
    {
      "epoch": 1.0289621882542237,
      "grad_norm": 0.29398404465578026,
      "learning_rate": 8.30698061864394e-06,
      "loss": 0.024,
      "step": 3837
    },
    {
      "epoch": 1.0292303566639849,
      "grad_norm": 0.25828520343396716,
      "learning_rate": 8.305810259996163e-06,
      "loss": 0.0225,
      "step": 3838
    },
    {
      "epoch": 1.0294985250737463,
      "grad_norm": 0.4260845095853011,
      "learning_rate": 8.304639579470416e-06,
      "loss": 0.0223,
      "step": 3839
    },
    {
      "epoch": 1.0297666934835077,
      "grad_norm": 0.32662873648817065,
      "learning_rate": 8.303468577180688e-06,
      "loss": 0.0207,
      "step": 3840
    },
    {
      "epoch": 1.0300348618932689,
      "grad_norm": 0.2474136480743761,
      "learning_rate": 8.302297253240995e-06,
      "loss": 0.0221,
      "step": 3841
    },
    {
      "epoch": 1.0303030303030303,
      "grad_norm": 0.29506853028474905,
      "learning_rate": 8.301125607765385e-06,
      "loss": 0.0342,
      "step": 3842
    },
    {
      "epoch": 1.0305711987127917,
      "grad_norm": 0.22539710793886594,
      "learning_rate": 8.29995364086794e-06,
      "loss": 0.0226,
      "step": 3843
    },
    {
      "epoch": 1.0308393671225529,
      "grad_norm": 0.19963115176027918,
      "learning_rate": 8.298781352662768e-06,
      "loss": 0.0203,
      "step": 3844
    },
    {
      "epoch": 1.0311075355323143,
      "grad_norm": 0.2855574254135013,
      "learning_rate": 8.297608743264017e-06,
      "loss": 0.0286,
      "step": 3845
    },
    {
      "epoch": 1.0313757039420757,
      "grad_norm": 0.28464176338512165,
      "learning_rate": 8.296435812785855e-06,
      "loss": 0.0305,
      "step": 3846
    },
    {
      "epoch": 1.0316438723518369,
      "grad_norm": 0.3489872617769799,
      "learning_rate": 8.295262561342493e-06,
      "loss": 0.0279,
      "step": 3847
    },
    {
      "epoch": 1.0319120407615983,
      "grad_norm": 0.3102699032514512,
      "learning_rate": 8.29408898904816e-06,
      "loss": 0.0212,
      "step": 3848
    },
    {
      "epoch": 1.0321802091713597,
      "grad_norm": 0.3235451019302266,
      "learning_rate": 8.292915096017128e-06,
      "loss": 0.0374,
      "step": 3849
    },
    {
      "epoch": 1.0324483775811208,
      "grad_norm": 0.5010230408849221,
      "learning_rate": 8.2917408823637e-06,
      "loss": 0.0439,
      "step": 3850
    },
    {
      "epoch": 1.0327165459908823,
      "grad_norm": 0.394496270455239,
      "learning_rate": 8.290566348202199e-06,
      "loss": 0.0286,
      "step": 3851
    },
    {
      "epoch": 1.0329847144006437,
      "grad_norm": 0.2985601829037895,
      "learning_rate": 8.289391493646988e-06,
      "loss": 0.0325,
      "step": 3852
    },
    {
      "epoch": 1.0332528828104048,
      "grad_norm": 0.30639940628632806,
      "learning_rate": 8.288216318812461e-06,
      "loss": 0.0282,
      "step": 3853
    },
    {
      "epoch": 1.0335210512201662,
      "grad_norm": 0.38466733970420747,
      "learning_rate": 8.287040823813041e-06,
      "loss": 0.0479,
      "step": 3854
    },
    {
      "epoch": 1.0337892196299276,
      "grad_norm": 0.28080142910781136,
      "learning_rate": 8.285865008763185e-06,
      "loss": 0.0251,
      "step": 3855
    },
    {
      "epoch": 1.0340573880396888,
      "grad_norm": 0.3422093961001927,
      "learning_rate": 8.284688873777375e-06,
      "loss": 0.0305,
      "step": 3856
    },
    {
      "epoch": 1.0343255564494502,
      "grad_norm": 0.39872738116836304,
      "learning_rate": 8.28351241897013e-06,
      "loss": 0.0289,
      "step": 3857
    },
    {
      "epoch": 1.0345937248592116,
      "grad_norm": 0.2538782521239731,
      "learning_rate": 8.282335644455998e-06,
      "loss": 0.0268,
      "step": 3858
    },
    {
      "epoch": 1.0348618932689728,
      "grad_norm": 0.3766756735196799,
      "learning_rate": 8.281158550349559e-06,
      "loss": 0.0267,
      "step": 3859
    },
    {
      "epoch": 1.0351300616787342,
      "grad_norm": 0.2872915280957078,
      "learning_rate": 8.279981136765424e-06,
      "loss": 0.0222,
      "step": 3860
    },
    {
      "epoch": 1.0353982300884956,
      "grad_norm": 0.32415337756830676,
      "learning_rate": 8.278803403818234e-06,
      "loss": 0.0334,
      "step": 3861
    },
    {
      "epoch": 1.0356663984982568,
      "grad_norm": 0.3623314879061271,
      "learning_rate": 8.277625351622662e-06,
      "loss": 0.0323,
      "step": 3862
    },
    {
      "epoch": 1.0359345669080182,
      "grad_norm": 0.39562800800957365,
      "learning_rate": 8.27644698029341e-06,
      "loss": 0.0371,
      "step": 3863
    },
    {
      "epoch": 1.0362027353177796,
      "grad_norm": 0.3041544310230412,
      "learning_rate": 8.275268289945215e-06,
      "loss": 0.0329,
      "step": 3864
    },
    {
      "epoch": 1.0364709037275408,
      "grad_norm": 0.30754787519176163,
      "learning_rate": 8.274089280692842e-06,
      "loss": 0.0229,
      "step": 3865
    },
    {
      "epoch": 1.0367390721373022,
      "grad_norm": 0.36251501627351373,
      "learning_rate": 8.27290995265109e-06,
      "loss": 0.0377,
      "step": 3866
    },
    {
      "epoch": 1.0370072405470636,
      "grad_norm": 0.29145796904724314,
      "learning_rate": 8.271730305934781e-06,
      "loss": 0.0363,
      "step": 3867
    },
    {
      "epoch": 1.0372754089568248,
      "grad_norm": 0.29271289674244216,
      "learning_rate": 8.270550340658782e-06,
      "loss": 0.0315,
      "step": 3868
    },
    {
      "epoch": 1.0375435773665862,
      "grad_norm": 0.4601282488362476,
      "learning_rate": 8.26937005693798e-06,
      "loss": 0.0259,
      "step": 3869
    },
    {
      "epoch": 1.0378117457763476,
      "grad_norm": 0.33055012338726836,
      "learning_rate": 8.268189454887292e-06,
      "loss": 0.0287,
      "step": 3870
    },
    {
      "epoch": 1.0380799141861088,
      "grad_norm": 0.3116507009776977,
      "learning_rate": 8.267008534621676e-06,
      "loss": 0.0287,
      "step": 3871
    },
    {
      "epoch": 1.0383480825958702,
      "grad_norm": 0.539901879925059,
      "learning_rate": 8.265827296256114e-06,
      "loss": 0.0358,
      "step": 3872
    },
    {
      "epoch": 1.0386162510056316,
      "grad_norm": 0.2676443502395585,
      "learning_rate": 8.264645739905618e-06,
      "loss": 0.0285,
      "step": 3873
    },
    {
      "epoch": 1.0388844194153928,
      "grad_norm": 0.2784292362651536,
      "learning_rate": 8.263463865685231e-06,
      "loss": 0.0273,
      "step": 3874
    },
    {
      "epoch": 1.0391525878251542,
      "grad_norm": 0.2592479966043478,
      "learning_rate": 8.262281673710035e-06,
      "loss": 0.0212,
      "step": 3875
    },
    {
      "epoch": 1.0394207562349156,
      "grad_norm": 0.25196195681290257,
      "learning_rate": 8.261099164095132e-06,
      "loss": 0.0243,
      "step": 3876
    },
    {
      "epoch": 1.0396889246446768,
      "grad_norm": 0.449428140724111,
      "learning_rate": 8.259916336955662e-06,
      "loss": 0.0286,
      "step": 3877
    },
    {
      "epoch": 1.0399570930544382,
      "grad_norm": 0.3084548153357562,
      "learning_rate": 8.258733192406793e-06,
      "loss": 0.0202,
      "step": 3878
    },
    {
      "epoch": 1.0402252614641996,
      "grad_norm": 0.3944415065709748,
      "learning_rate": 8.257549730563726e-06,
      "loss": 0.0285,
      "step": 3879
    },
    {
      "epoch": 1.0404934298739608,
      "grad_norm": 0.4391545746233972,
      "learning_rate": 8.256365951541689e-06,
      "loss": 0.0312,
      "step": 3880
    },
    {
      "epoch": 1.0407615982837222,
      "grad_norm": 0.30001147859520644,
      "learning_rate": 8.255181855455944e-06,
      "loss": 0.0205,
      "step": 3881
    },
    {
      "epoch": 1.0410297666934836,
      "grad_norm": 0.39110410990857647,
      "learning_rate": 8.253997442421786e-06,
      "loss": 0.0322,
      "step": 3882
    },
    {
      "epoch": 1.0412979351032448,
      "grad_norm": 0.2723746850971545,
      "learning_rate": 8.252812712554535e-06,
      "loss": 0.0267,
      "step": 3883
    },
    {
      "epoch": 1.0415661035130062,
      "grad_norm": 0.41680566928831214,
      "learning_rate": 8.251627665969548e-06,
      "loss": 0.033,
      "step": 3884
    },
    {
      "epoch": 1.0418342719227676,
      "grad_norm": 0.32525191155937244,
      "learning_rate": 8.250442302782206e-06,
      "loss": 0.0233,
      "step": 3885
    },
    {
      "epoch": 1.0421024403325287,
      "grad_norm": 0.33399947802976115,
      "learning_rate": 8.249256623107928e-06,
      "loss": 0.0411,
      "step": 3886
    },
    {
      "epoch": 1.0423706087422902,
      "grad_norm": 0.4595065460795739,
      "learning_rate": 8.248070627062156e-06,
      "loss": 0.0234,
      "step": 3887
    },
    {
      "epoch": 1.0426387771520516,
      "grad_norm": 0.3864656176156845,
      "learning_rate": 8.246884314760376e-06,
      "loss": 0.0311,
      "step": 3888
    },
    {
      "epoch": 1.0429069455618127,
      "grad_norm": 0.3578768444335209,
      "learning_rate": 8.245697686318086e-06,
      "loss": 0.0247,
      "step": 3889
    },
    {
      "epoch": 1.0431751139715741,
      "grad_norm": 0.2482538524658289,
      "learning_rate": 8.244510741850831e-06,
      "loss": 0.0207,
      "step": 3890
    },
    {
      "epoch": 1.0434432823813355,
      "grad_norm": 0.37534496980449944,
      "learning_rate": 8.24332348147418e-06,
      "loss": 0.0316,
      "step": 3891
    },
    {
      "epoch": 1.0437114507910967,
      "grad_norm": 0.2907304945498778,
      "learning_rate": 8.242135905303731e-06,
      "loss": 0.0201,
      "step": 3892
    },
    {
      "epoch": 1.0439796192008581,
      "grad_norm": 0.3257092226721669,
      "learning_rate": 8.240948013455117e-06,
      "loss": 0.0278,
      "step": 3893
    },
    {
      "epoch": 1.0442477876106195,
      "grad_norm": 0.35073029390515487,
      "learning_rate": 8.239759806044001e-06,
      "loss": 0.0295,
      "step": 3894
    },
    {
      "epoch": 1.0445159560203807,
      "grad_norm": 0.30462722276583604,
      "learning_rate": 8.238571283186074e-06,
      "loss": 0.0271,
      "step": 3895
    },
    {
      "epoch": 1.0447841244301421,
      "grad_norm": 0.4259263989554586,
      "learning_rate": 8.237382444997057e-06,
      "loss": 0.0476,
      "step": 3896
    },
    {
      "epoch": 1.0450522928399035,
      "grad_norm": 0.42279144990955564,
      "learning_rate": 8.236193291592707e-06,
      "loss": 0.0245,
      "step": 3897
    },
    {
      "epoch": 1.0453204612496647,
      "grad_norm": 0.5125161513850827,
      "learning_rate": 8.23500382308881e-06,
      "loss": 0.0343,
      "step": 3898
    },
    {
      "epoch": 1.0455886296594261,
      "grad_norm": 0.3807787884102021,
      "learning_rate": 8.233814039601178e-06,
      "loss": 0.0322,
      "step": 3899
    },
    {
      "epoch": 1.0458567980691875,
      "grad_norm": 0.3259210123790663,
      "learning_rate": 8.232623941245658e-06,
      "loss": 0.0269,
      "step": 3900
    },
    {
      "epoch": 1.0461249664789487,
      "grad_norm": 0.4489841358960552,
      "learning_rate": 8.231433528138128e-06,
      "loss": 0.0322,
      "step": 3901
    },
    {
      "epoch": 1.04639313488871,
      "grad_norm": 0.3891554092911375,
      "learning_rate": 8.230242800394496e-06,
      "loss": 0.0232,
      "step": 3902
    },
    {
      "epoch": 1.0466613032984715,
      "grad_norm": 0.3236736418976952,
      "learning_rate": 8.229051758130697e-06,
      "loss": 0.0413,
      "step": 3903
    },
    {
      "epoch": 1.0469294717082327,
      "grad_norm": 0.4731956143790975,
      "learning_rate": 8.227860401462702e-06,
      "loss": 0.0428,
      "step": 3904
    },
    {
      "epoch": 1.047197640117994,
      "grad_norm": 0.3034438191866115,
      "learning_rate": 8.226668730506509e-06,
      "loss": 0.0243,
      "step": 3905
    },
    {
      "epoch": 1.0474658085277555,
      "grad_norm": 0.23666139585077156,
      "learning_rate": 8.225476745378149e-06,
      "loss": 0.0186,
      "step": 3906
    },
    {
      "epoch": 1.0477339769375167,
      "grad_norm": 0.26601601814055414,
      "learning_rate": 8.224284446193682e-06,
      "loss": 0.0221,
      "step": 3907
    },
    {
      "epoch": 1.048002145347278,
      "grad_norm": 0.2708272749856384,
      "learning_rate": 8.223091833069197e-06,
      "loss": 0.0351,
      "step": 3908
    },
    {
      "epoch": 1.0482703137570395,
      "grad_norm": 0.29135360701756563,
      "learning_rate": 8.22189890612082e-06,
      "loss": 0.0242,
      "step": 3909
    },
    {
      "epoch": 1.0485384821668007,
      "grad_norm": 0.28996393847561214,
      "learning_rate": 8.220705665464699e-06,
      "loss": 0.0243,
      "step": 3910
    },
    {
      "epoch": 1.048806650576562,
      "grad_norm": 0.3175549122744191,
      "learning_rate": 8.219512111217016e-06,
      "loss": 0.0243,
      "step": 3911
    },
    {
      "epoch": 1.0490748189863235,
      "grad_norm": 0.2598688258498686,
      "learning_rate": 8.21831824349399e-06,
      "loss": 0.0268,
      "step": 3912
    },
    {
      "epoch": 1.0493429873960847,
      "grad_norm": 0.3497872500920714,
      "learning_rate": 8.21712406241186e-06,
      "loss": 0.0266,
      "step": 3913
    },
    {
      "epoch": 1.049611155805846,
      "grad_norm": 0.41673824946410626,
      "learning_rate": 8.2159295680869e-06,
      "loss": 0.0337,
      "step": 3914
    },
    {
      "epoch": 1.0498793242156075,
      "grad_norm": 0.36579571937782474,
      "learning_rate": 8.214734760635418e-06,
      "loss": 0.0274,
      "step": 3915
    },
    {
      "epoch": 1.0501474926253687,
      "grad_norm": 0.4705266609663164,
      "learning_rate": 8.213539640173746e-06,
      "loss": 0.0334,
      "step": 3916
    },
    {
      "epoch": 1.05041566103513,
      "grad_norm": 0.24524251519608842,
      "learning_rate": 8.21234420681825e-06,
      "loss": 0.0209,
      "step": 3917
    },
    {
      "epoch": 1.0506838294448915,
      "grad_norm": 0.2800023852075173,
      "learning_rate": 8.21114846068533e-06,
      "loss": 0.0321,
      "step": 3918
    },
    {
      "epoch": 1.0509519978546527,
      "grad_norm": 0.842326507784826,
      "learning_rate": 8.209952401891409e-06,
      "loss": 0.0506,
      "step": 3919
    },
    {
      "epoch": 1.051220166264414,
      "grad_norm": 0.2850190962926409,
      "learning_rate": 8.208756030552944e-06,
      "loss": 0.0218,
      "step": 3920
    },
    {
      "epoch": 1.0514883346741755,
      "grad_norm": 0.3038562543432837,
      "learning_rate": 8.207559346786424e-06,
      "loss": 0.0325,
      "step": 3921
    },
    {
      "epoch": 1.0517565030839366,
      "grad_norm": 0.46708199822661717,
      "learning_rate": 8.206362350708365e-06,
      "loss": 0.0282,
      "step": 3922
    },
    {
      "epoch": 1.052024671493698,
      "grad_norm": 0.7569133649961993,
      "learning_rate": 8.205165042435317e-06,
      "loss": 0.0336,
      "step": 3923
    },
    {
      "epoch": 1.0522928399034595,
      "grad_norm": 0.3111984150983337,
      "learning_rate": 8.20396742208386e-06,
      "loss": 0.0252,
      "step": 3924
    },
    {
      "epoch": 1.0525610083132206,
      "grad_norm": 0.4982732376998536,
      "learning_rate": 8.202769489770602e-06,
      "loss": 0.029,
      "step": 3925
    },
    {
      "epoch": 1.052829176722982,
      "grad_norm": 0.3474166551775235,
      "learning_rate": 8.201571245612179e-06,
      "loss": 0.0231,
      "step": 3926
    },
    {
      "epoch": 1.0530973451327434,
      "grad_norm": 0.3632753671134533,
      "learning_rate": 8.200372689725265e-06,
      "loss": 0.0309,
      "step": 3927
    },
    {
      "epoch": 1.0533655135425046,
      "grad_norm": 0.24294925159143982,
      "learning_rate": 8.19917382222656e-06,
      "loss": 0.0227,
      "step": 3928
    },
    {
      "epoch": 1.053633681952266,
      "grad_norm": 0.3093868874005352,
      "learning_rate": 8.197974643232794e-06,
      "loss": 0.0192,
      "step": 3929
    },
    {
      "epoch": 1.0539018503620274,
      "grad_norm": 0.34198263219409675,
      "learning_rate": 8.196775152860725e-06,
      "loss": 0.0233,
      "step": 3930
    },
    {
      "epoch": 1.0541700187717886,
      "grad_norm": 0.2904204092953673,
      "learning_rate": 8.195575351227148e-06,
      "loss": 0.0226,
      "step": 3931
    },
    {
      "epoch": 1.05443818718155,
      "grad_norm": 0.28397167378299093,
      "learning_rate": 8.194375238448884e-06,
      "loss": 0.027,
      "step": 3932
    },
    {
      "epoch": 1.0547063555913114,
      "grad_norm": 0.35042399503223015,
      "learning_rate": 8.193174814642782e-06,
      "loss": 0.0235,
      "step": 3933
    },
    {
      "epoch": 1.0549745240010726,
      "grad_norm": 0.349350310943682,
      "learning_rate": 8.191974079925727e-06,
      "loss": 0.0233,
      "step": 3934
    },
    {
      "epoch": 1.055242692410834,
      "grad_norm": 0.38937277649520335,
      "learning_rate": 8.190773034414631e-06,
      "loss": 0.0317,
      "step": 3935
    },
    {
      "epoch": 1.0555108608205954,
      "grad_norm": 0.2714228418371403,
      "learning_rate": 8.189571678226436e-06,
      "loss": 0.0242,
      "step": 3936
    },
    {
      "epoch": 1.0557790292303566,
      "grad_norm": 0.3033725002359017,
      "learning_rate": 8.188370011478114e-06,
      "loss": 0.0328,
      "step": 3937
    },
    {
      "epoch": 1.056047197640118,
      "grad_norm": 0.25273246606998884,
      "learning_rate": 8.187168034286668e-06,
      "loss": 0.0221,
      "step": 3938
    },
    {
      "epoch": 1.0563153660498794,
      "grad_norm": 0.367957599015788,
      "learning_rate": 8.185965746769134e-06,
      "loss": 0.0297,
      "step": 3939
    },
    {
      "epoch": 1.0565835344596406,
      "grad_norm": 0.3640407952586068,
      "learning_rate": 8.184763149042574e-06,
      "loss": 0.0314,
      "step": 3940
    },
    {
      "epoch": 1.056851702869402,
      "grad_norm": 0.310981058659115,
      "learning_rate": 8.18356024122408e-06,
      "loss": 0.025,
      "step": 3941
    },
    {
      "epoch": 1.0571198712791634,
      "grad_norm": 0.4646813103790253,
      "learning_rate": 8.182357023430778e-06,
      "loss": 0.031,
      "step": 3942
    },
    {
      "epoch": 1.0573880396889246,
      "grad_norm": 0.28785398293696013,
      "learning_rate": 8.181153495779823e-06,
      "loss": 0.031,
      "step": 3943
    },
    {
      "epoch": 1.057656208098686,
      "grad_norm": 0.27760908368375986,
      "learning_rate": 8.179949658388398e-06,
      "loss": 0.0277,
      "step": 3944
    },
    {
      "epoch": 1.0579243765084474,
      "grad_norm": 0.2571912854244579,
      "learning_rate": 8.178745511373714e-06,
      "loss": 0.0253,
      "step": 3945
    },
    {
      "epoch": 1.0581925449182086,
      "grad_norm": 0.34089477172819593,
      "learning_rate": 8.177541054853022e-06,
      "loss": 0.0294,
      "step": 3946
    },
    {
      "epoch": 1.05846071332797,
      "grad_norm": 0.34671190931635476,
      "learning_rate": 8.176336288943592e-06,
      "loss": 0.0336,
      "step": 3947
    },
    {
      "epoch": 1.0587288817377314,
      "grad_norm": 0.45043297039886504,
      "learning_rate": 8.175131213762732e-06,
      "loss": 0.0351,
      "step": 3948
    },
    {
      "epoch": 1.0589970501474926,
      "grad_norm": 0.25241412512169514,
      "learning_rate": 8.173925829427776e-06,
      "loss": 0.02,
      "step": 3949
    },
    {
      "epoch": 1.059265218557254,
      "grad_norm": 0.36921106447317537,
      "learning_rate": 8.172720136056087e-06,
      "loss": 0.0262,
      "step": 3950
    },
    {
      "epoch": 1.0595333869670154,
      "grad_norm": 0.29990404493953793,
      "learning_rate": 8.171514133765062e-06,
      "loss": 0.0255,
      "step": 3951
    },
    {
      "epoch": 1.0598015553767766,
      "grad_norm": 0.25368062851736867,
      "learning_rate": 8.170307822672127e-06,
      "loss": 0.0196,
      "step": 3952
    },
    {
      "epoch": 1.060069723786538,
      "grad_norm": 0.2653707732623665,
      "learning_rate": 8.169101202894735e-06,
      "loss": 0.0232,
      "step": 3953
    },
    {
      "epoch": 1.0603378921962994,
      "grad_norm": 0.2749175424405886,
      "learning_rate": 8.167894274550374e-06,
      "loss": 0.026,
      "step": 3954
    },
    {
      "epoch": 1.0606060606060606,
      "grad_norm": 0.29656461527362604,
      "learning_rate": 8.166687037756558e-06,
      "loss": 0.0298,
      "step": 3955
    },
    {
      "epoch": 1.060874229015822,
      "grad_norm": 0.2764124505783319,
      "learning_rate": 8.165479492630831e-06,
      "loss": 0.0315,
      "step": 3956
    },
    {
      "epoch": 1.0611423974255834,
      "grad_norm": 0.30121642778279656,
      "learning_rate": 8.164271639290771e-06,
      "loss": 0.0201,
      "step": 3957
    },
    {
      "epoch": 1.0614105658353445,
      "grad_norm": 0.22477680343907117,
      "learning_rate": 8.163063477853983e-06,
      "loss": 0.019,
      "step": 3958
    },
    {
      "epoch": 1.061678734245106,
      "grad_norm": 0.34861880902156456,
      "learning_rate": 8.1618550084381e-06,
      "loss": 0.0258,
      "step": 3959
    },
    {
      "epoch": 1.0619469026548674,
      "grad_norm": 0.39784897431861116,
      "learning_rate": 8.160646231160792e-06,
      "loss": 0.0359,
      "step": 3960
    },
    {
      "epoch": 1.0622150710646285,
      "grad_norm": 0.3638237368655141,
      "learning_rate": 8.159437146139751e-06,
      "loss": 0.0279,
      "step": 3961
    },
    {
      "epoch": 1.06248323947439,
      "grad_norm": 0.2640167172934997,
      "learning_rate": 8.158227753492702e-06,
      "loss": 0.0221,
      "step": 3962
    },
    {
      "epoch": 1.0627514078841513,
      "grad_norm": 0.3637955799726516,
      "learning_rate": 8.157018053337401e-06,
      "loss": 0.0356,
      "step": 3963
    },
    {
      "epoch": 1.0630195762939125,
      "grad_norm": 0.35121023919622213,
      "learning_rate": 8.155808045791635e-06,
      "loss": 0.0418,
      "step": 3964
    },
    {
      "epoch": 1.063287744703674,
      "grad_norm": 0.33740026660793343,
      "learning_rate": 8.15459773097322e-06,
      "loss": 0.0202,
      "step": 3965
    },
    {
      "epoch": 1.0635559131134353,
      "grad_norm": 0.33070590433530217,
      "learning_rate": 8.153387108999998e-06,
      "loss": 0.0201,
      "step": 3966
    },
    {
      "epoch": 1.0638240815231965,
      "grad_norm": 0.2767549144089937,
      "learning_rate": 8.152176179989843e-06,
      "loss": 0.0268,
      "step": 3967
    },
    {
      "epoch": 1.064092249932958,
      "grad_norm": 0.3318908117189271,
      "learning_rate": 8.150964944060665e-06,
      "loss": 0.0263,
      "step": 3968
    },
    {
      "epoch": 1.0643604183427193,
      "grad_norm": 0.296243539509169,
      "learning_rate": 8.149753401330395e-06,
      "loss": 0.0333,
      "step": 3969
    },
    {
      "epoch": 1.0646285867524805,
      "grad_norm": 0.3618070580911903,
      "learning_rate": 8.148541551917e-06,
      "loss": 0.037,
      "step": 3970
    },
    {
      "epoch": 1.064896755162242,
      "grad_norm": 0.2422412686978263,
      "learning_rate": 8.14732939593847e-06,
      "loss": 0.0283,
      "step": 3971
    },
    {
      "epoch": 1.0651649235720033,
      "grad_norm": 0.43128567606317986,
      "learning_rate": 8.146116933512836e-06,
      "loss": 0.0282,
      "step": 3972
    },
    {
      "epoch": 1.0654330919817645,
      "grad_norm": 0.3552815259027964,
      "learning_rate": 8.144904164758147e-06,
      "loss": 0.0266,
      "step": 3973
    },
    {
      "epoch": 1.065701260391526,
      "grad_norm": 0.40991021268771566,
      "learning_rate": 8.14369108979249e-06,
      "loss": 0.0373,
      "step": 3974
    },
    {
      "epoch": 1.0659694288012873,
      "grad_norm": 0.4585034175360701,
      "learning_rate": 8.142477708733977e-06,
      "loss": 0.0325,
      "step": 3975
    },
    {
      "epoch": 1.0662375972110485,
      "grad_norm": 0.2914855856042394,
      "learning_rate": 8.141264021700755e-06,
      "loss": 0.0193,
      "step": 3976
    },
    {
      "epoch": 1.06650576562081,
      "grad_norm": 0.28897058798545794,
      "learning_rate": 8.140050028810992e-06,
      "loss": 0.0287,
      "step": 3977
    },
    {
      "epoch": 1.0667739340305713,
      "grad_norm": 0.3306548149138281,
      "learning_rate": 8.138835730182897e-06,
      "loss": 0.0249,
      "step": 3978
    },
    {
      "epoch": 1.0670421024403325,
      "grad_norm": 0.39512041903544326,
      "learning_rate": 8.1376211259347e-06,
      "loss": 0.0329,
      "step": 3979
    },
    {
      "epoch": 1.067310270850094,
      "grad_norm": 0.2811496613617414,
      "learning_rate": 8.136406216184662e-06,
      "loss": 0.0217,
      "step": 3980
    },
    {
      "epoch": 1.0675784392598553,
      "grad_norm": 0.2565723393861113,
      "learning_rate": 8.13519100105108e-06,
      "loss": 0.024,
      "step": 3981
    },
    {
      "epoch": 1.0678466076696165,
      "grad_norm": 0.2789733120821343,
      "learning_rate": 8.133975480652273e-06,
      "loss": 0.0255,
      "step": 3982
    },
    {
      "epoch": 1.0681147760793779,
      "grad_norm": 0.24664027471438452,
      "learning_rate": 8.132759655106595e-06,
      "loss": 0.0222,
      "step": 3983
    },
    {
      "epoch": 1.068382944489139,
      "grad_norm": 0.282852694286996,
      "learning_rate": 8.131543524532426e-06,
      "loss": 0.0176,
      "step": 3984
    },
    {
      "epoch": 1.0686511128989005,
      "grad_norm": 0.2660293135939434,
      "learning_rate": 8.130327089048179e-06,
      "loss": 0.0253,
      "step": 3985
    },
    {
      "epoch": 1.0689192813086619,
      "grad_norm": 0.22955810939491159,
      "learning_rate": 8.129110348772291e-06,
      "loss": 0.0191,
      "step": 3986
    },
    {
      "epoch": 1.0691874497184233,
      "grad_norm": 0.23306371231669049,
      "learning_rate": 8.127893303823237e-06,
      "loss": 0.019,
      "step": 3987
    },
    {
      "epoch": 1.0694556181281845,
      "grad_norm": 0.6135945724220748,
      "learning_rate": 8.126675954319519e-06,
      "loss": 0.0275,
      "step": 3988
    },
    {
      "epoch": 1.0697237865379459,
      "grad_norm": 0.36718940422748597,
      "learning_rate": 8.125458300379663e-06,
      "loss": 0.0346,
      "step": 3989
    },
    {
      "epoch": 1.069991954947707,
      "grad_norm": 0.29321727987001633,
      "learning_rate": 8.12424034212223e-06,
      "loss": 0.0292,
      "step": 3990
    },
    {
      "epoch": 1.0702601233574685,
      "grad_norm": 0.22561550980456965,
      "learning_rate": 8.12302207966581e-06,
      "loss": 0.0184,
      "step": 3991
    },
    {
      "epoch": 1.0705282917672299,
      "grad_norm": 0.2852455716702975,
      "learning_rate": 8.121803513129021e-06,
      "loss": 0.0274,
      "step": 3992
    },
    {
      "epoch": 1.0707964601769913,
      "grad_norm": 0.2751951925624213,
      "learning_rate": 8.120584642630513e-06,
      "loss": 0.0173,
      "step": 3993
    },
    {
      "epoch": 1.0710646285867524,
      "grad_norm": 0.32032060637510307,
      "learning_rate": 8.119365468288963e-06,
      "loss": 0.0267,
      "step": 3994
    },
    {
      "epoch": 1.0713327969965138,
      "grad_norm": 0.3583189484360853,
      "learning_rate": 8.118145990223078e-06,
      "loss": 0.0425,
      "step": 3995
    },
    {
      "epoch": 1.071600965406275,
      "grad_norm": 0.2937272559409034,
      "learning_rate": 8.116926208551598e-06,
      "loss": 0.0193,
      "step": 3996
    },
    {
      "epoch": 1.0718691338160364,
      "grad_norm": 0.2382120323371079,
      "learning_rate": 8.115706123393287e-06,
      "loss": 0.0261,
      "step": 3997
    },
    {
      "epoch": 1.0721373022257978,
      "grad_norm": 0.378921276441055,
      "learning_rate": 8.114485734866944e-06,
      "loss": 0.0283,
      "step": 3998
    },
    {
      "epoch": 1.0724054706355592,
      "grad_norm": 0.27047055759150507,
      "learning_rate": 8.113265043091393e-06,
      "loss": 0.0229,
      "step": 3999
    },
    {
      "epoch": 1.0726736390453204,
      "grad_norm": 0.25535675650451845,
      "learning_rate": 8.112044048185492e-06,
      "loss": 0.0286,
      "step": 4000
    },
    {
      "epoch": 1.0729418074550818,
      "grad_norm": 0.19169275795802673,
      "learning_rate": 8.110822750268124e-06,
      "loss": 0.0154,
      "step": 4001
    },
    {
      "epoch": 1.073209975864843,
      "grad_norm": 0.538203409576204,
      "learning_rate": 8.109601149458202e-06,
      "loss": 0.0471,
      "step": 4002
    },
    {
      "epoch": 1.0734781442746044,
      "grad_norm": 0.29095760048083846,
      "learning_rate": 8.108379245874673e-06,
      "loss": 0.0356,
      "step": 4003
    },
    {
      "epoch": 1.0737463126843658,
      "grad_norm": 0.35789636388950097,
      "learning_rate": 8.107157039636509e-06,
      "loss": 0.0348,
      "step": 4004
    },
    {
      "epoch": 1.0740144810941272,
      "grad_norm": 0.3621126745542607,
      "learning_rate": 8.105934530862714e-06,
      "loss": 0.0274,
      "step": 4005
    },
    {
      "epoch": 1.0742826495038884,
      "grad_norm": 0.32130110514766413,
      "learning_rate": 8.104711719672318e-06,
      "loss": 0.0279,
      "step": 4006
    },
    {
      "epoch": 1.0745508179136498,
      "grad_norm": 0.38446401161196014,
      "learning_rate": 8.103488606184387e-06,
      "loss": 0.0272,
      "step": 4007
    },
    {
      "epoch": 1.074818986323411,
      "grad_norm": 0.3013428203215738,
      "learning_rate": 8.10226519051801e-06,
      "loss": 0.0253,
      "step": 4008
    },
    {
      "epoch": 1.0750871547331724,
      "grad_norm": 0.28859934631746204,
      "learning_rate": 8.101041472792307e-06,
      "loss": 0.0202,
      "step": 4009
    },
    {
      "epoch": 1.0753553231429338,
      "grad_norm": 0.3038656480376863,
      "learning_rate": 8.099817453126428e-06,
      "loss": 0.0254,
      "step": 4010
    },
    {
      "epoch": 1.0756234915526952,
      "grad_norm": 0.36494270594228884,
      "learning_rate": 8.098593131639555e-06,
      "loss": 0.0245,
      "step": 4011
    },
    {
      "epoch": 1.0758916599624564,
      "grad_norm": 0.30542772176140154,
      "learning_rate": 8.097368508450894e-06,
      "loss": 0.028,
      "step": 4012
    },
    {
      "epoch": 1.0761598283722178,
      "grad_norm": 0.326473523512798,
      "learning_rate": 8.096143583679685e-06,
      "loss": 0.0248,
      "step": 4013
    },
    {
      "epoch": 1.076427996781979,
      "grad_norm": 0.27202087299970124,
      "learning_rate": 8.094918357445197e-06,
      "loss": 0.0211,
      "step": 4014
    },
    {
      "epoch": 1.0766961651917404,
      "grad_norm": 0.3484324401446948,
      "learning_rate": 8.093692829866723e-06,
      "loss": 0.0247,
      "step": 4015
    },
    {
      "epoch": 1.0769643336015018,
      "grad_norm": 0.5177219208661106,
      "learning_rate": 8.092467001063593e-06,
      "loss": 0.0398,
      "step": 4016
    },
    {
      "epoch": 1.077232502011263,
      "grad_norm": 0.3729813655612874,
      "learning_rate": 8.091240871155162e-06,
      "loss": 0.0395,
      "step": 4017
    },
    {
      "epoch": 1.0775006704210244,
      "grad_norm": 0.23546090939524478,
      "learning_rate": 8.090014440260814e-06,
      "loss": 0.0222,
      "step": 4018
    },
    {
      "epoch": 1.0777688388307858,
      "grad_norm": 0.34919507426610485,
      "learning_rate": 8.088787708499964e-06,
      "loss": 0.0355,
      "step": 4019
    },
    {
      "epoch": 1.078037007240547,
      "grad_norm": 0.42953376691963135,
      "learning_rate": 8.087560675992053e-06,
      "loss": 0.0221,
      "step": 4020
    },
    {
      "epoch": 1.0783051756503084,
      "grad_norm": 0.41296775415768683,
      "learning_rate": 8.086333342856561e-06,
      "loss": 0.0317,
      "step": 4021
    },
    {
      "epoch": 1.0785733440600698,
      "grad_norm": 0.30368758400624407,
      "learning_rate": 8.08510570921298e-06,
      "loss": 0.0274,
      "step": 4022
    },
    {
      "epoch": 1.078841512469831,
      "grad_norm": 0.3544727514387742,
      "learning_rate": 8.083877775180851e-06,
      "loss": 0.0257,
      "step": 4023
    },
    {
      "epoch": 1.0791096808795924,
      "grad_norm": 0.28199482312265,
      "learning_rate": 8.08264954087973e-06,
      "loss": 0.0261,
      "step": 4024
    },
    {
      "epoch": 1.0793778492893538,
      "grad_norm": 0.3426246650758932,
      "learning_rate": 8.081421006429205e-06,
      "loss": 0.0372,
      "step": 4025
    },
    {
      "epoch": 1.079646017699115,
      "grad_norm": 0.3440385616870937,
      "learning_rate": 8.0801921719489e-06,
      "loss": 0.0241,
      "step": 4026
    },
    {
      "epoch": 1.0799141861088764,
      "grad_norm": 0.30631584242231974,
      "learning_rate": 8.078963037558461e-06,
      "loss": 0.0278,
      "step": 4027
    },
    {
      "epoch": 1.0801823545186378,
      "grad_norm": 0.23438062454742695,
      "learning_rate": 8.077733603377565e-06,
      "loss": 0.0217,
      "step": 4028
    },
    {
      "epoch": 1.080450522928399,
      "grad_norm": 0.2534919756097813,
      "learning_rate": 8.076503869525919e-06,
      "loss": 0.0257,
      "step": 4029
    },
    {
      "epoch": 1.0807186913381603,
      "grad_norm": 0.2536373087260301,
      "learning_rate": 8.07527383612326e-06,
      "loss": 0.0209,
      "step": 4030
    },
    {
      "epoch": 1.0809868597479217,
      "grad_norm": 0.44563928327614527,
      "learning_rate": 8.074043503289349e-06,
      "loss": 0.0363,
      "step": 4031
    },
    {
      "epoch": 1.081255028157683,
      "grad_norm": 0.34378620511302005,
      "learning_rate": 8.072812871143987e-06,
      "loss": 0.0282,
      "step": 4032
    },
    {
      "epoch": 1.0815231965674443,
      "grad_norm": 0.30719081251400204,
      "learning_rate": 8.071581939806991e-06,
      "loss": 0.0342,
      "step": 4033
    },
    {
      "epoch": 1.0817913649772057,
      "grad_norm": 0.2635018480871943,
      "learning_rate": 8.07035070939822e-06,
      "loss": 0.0303,
      "step": 4034
    },
    {
      "epoch": 1.082059533386967,
      "grad_norm": 0.2991762134067907,
      "learning_rate": 8.06911918003755e-06,
      "loss": 0.0255,
      "step": 4035
    },
    {
      "epoch": 1.0823277017967283,
      "grad_norm": 0.5953932006014163,
      "learning_rate": 8.067887351844895e-06,
      "loss": 0.0347,
      "step": 4036
    },
    {
      "epoch": 1.0825958702064897,
      "grad_norm": 0.26123024138162865,
      "learning_rate": 8.066655224940193e-06,
      "loss": 0.0227,
      "step": 4037
    },
    {
      "epoch": 1.082864038616251,
      "grad_norm": 0.25532011411982336,
      "learning_rate": 8.065422799443412e-06,
      "loss": 0.0212,
      "step": 4038
    },
    {
      "epoch": 1.0831322070260123,
      "grad_norm": 0.2595992096164906,
      "learning_rate": 8.064190075474553e-06,
      "loss": 0.0216,
      "step": 4039
    },
    {
      "epoch": 1.0834003754357737,
      "grad_norm": 0.3168796294216165,
      "learning_rate": 8.062957053153641e-06,
      "loss": 0.0235,
      "step": 4040
    },
    {
      "epoch": 1.083668543845535,
      "grad_norm": 0.31844684525508077,
      "learning_rate": 8.061723732600733e-06,
      "loss": 0.0323,
      "step": 4041
    },
    {
      "epoch": 1.0839367122552963,
      "grad_norm": 0.3486797883559125,
      "learning_rate": 8.060490113935916e-06,
      "loss": 0.027,
      "step": 4042
    },
    {
      "epoch": 1.0842048806650577,
      "grad_norm": 0.26238202735324084,
      "learning_rate": 8.0592561972793e-06,
      "loss": 0.0234,
      "step": 4043
    },
    {
      "epoch": 1.084473049074819,
      "grad_norm": 0.25955016335902786,
      "learning_rate": 8.05802198275103e-06,
      "loss": 0.0239,
      "step": 4044
    },
    {
      "epoch": 1.0847412174845803,
      "grad_norm": 0.3794019931453181,
      "learning_rate": 8.056787470471279e-06,
      "loss": 0.0317,
      "step": 4045
    },
    {
      "epoch": 1.0850093858943417,
      "grad_norm": 0.27177213588586335,
      "learning_rate": 8.055552660560246e-06,
      "loss": 0.0265,
      "step": 4046
    },
    {
      "epoch": 1.0852775543041029,
      "grad_norm": 0.2719606560556994,
      "learning_rate": 8.054317553138164e-06,
      "loss": 0.0231,
      "step": 4047
    },
    {
      "epoch": 1.0855457227138643,
      "grad_norm": 0.35985091577195283,
      "learning_rate": 8.053082148325293e-06,
      "loss": 0.024,
      "step": 4048
    },
    {
      "epoch": 1.0858138911236257,
      "grad_norm": 0.3224255478113859,
      "learning_rate": 8.051846446241916e-06,
      "loss": 0.0253,
      "step": 4049
    },
    {
      "epoch": 1.0860820595333869,
      "grad_norm": 0.2576902214272502,
      "learning_rate": 8.050610447008354e-06,
      "loss": 0.0188,
      "step": 4050
    },
    {
      "epoch": 1.0863502279431483,
      "grad_norm": 0.29356023393777875,
      "learning_rate": 8.049374150744953e-06,
      "loss": 0.0268,
      "step": 4051
    },
    {
      "epoch": 1.0866183963529097,
      "grad_norm": 0.28297118725891,
      "learning_rate": 8.048137557572087e-06,
      "loss": 0.0263,
      "step": 4052
    },
    {
      "epoch": 1.0868865647626709,
      "grad_norm": 0.3422632096740693,
      "learning_rate": 8.046900667610158e-06,
      "loss": 0.024,
      "step": 4053
    },
    {
      "epoch": 1.0871547331724323,
      "grad_norm": 0.38670197076052765,
      "learning_rate": 8.0456634809796e-06,
      "loss": 0.0378,
      "step": 4054
    },
    {
      "epoch": 1.0874229015821937,
      "grad_norm": 0.28359859242672236,
      "learning_rate": 8.044425997800875e-06,
      "loss": 0.0251,
      "step": 4055
    },
    {
      "epoch": 1.0876910699919549,
      "grad_norm": 0.42216898171422046,
      "learning_rate": 8.043188218194476e-06,
      "loss": 0.0284,
      "step": 4056
    },
    {
      "epoch": 1.0879592384017163,
      "grad_norm": 0.2782266862935743,
      "learning_rate": 8.041950142280918e-06,
      "loss": 0.0256,
      "step": 4057
    },
    {
      "epoch": 1.0882274068114777,
      "grad_norm": 0.3419740042546062,
      "learning_rate": 8.040711770180751e-06,
      "loss": 0.03,
      "step": 4058
    },
    {
      "epoch": 1.0884955752212389,
      "grad_norm": 0.30344508289312155,
      "learning_rate": 8.039473102014552e-06,
      "loss": 0.027,
      "step": 4059
    },
    {
      "epoch": 1.0887637436310003,
      "grad_norm": 0.4485768283632113,
      "learning_rate": 8.038234137902925e-06,
      "loss": 0.0281,
      "step": 4060
    },
    {
      "epoch": 1.0890319120407617,
      "grad_norm": 0.36335700574738167,
      "learning_rate": 8.036994877966509e-06,
      "loss": 0.0272,
      "step": 4061
    },
    {
      "epoch": 1.0893000804505228,
      "grad_norm": 0.25781821185777726,
      "learning_rate": 8.035755322325964e-06,
      "loss": 0.0191,
      "step": 4062
    },
    {
      "epoch": 1.0895682488602842,
      "grad_norm": 0.3929097040945974,
      "learning_rate": 8.034515471101982e-06,
      "loss": 0.0296,
      "step": 4063
    },
    {
      "epoch": 1.0898364172700457,
      "grad_norm": 0.26604554681248765,
      "learning_rate": 8.033275324415286e-06,
      "loss": 0.0213,
      "step": 4064
    },
    {
      "epoch": 1.0901045856798068,
      "grad_norm": 0.38964493018400487,
      "learning_rate": 8.032034882386625e-06,
      "loss": 0.0353,
      "step": 4065
    },
    {
      "epoch": 1.0903727540895682,
      "grad_norm": 0.3098233263762199,
      "learning_rate": 8.030794145136776e-06,
      "loss": 0.0246,
      "step": 4066
    },
    {
      "epoch": 1.0906409224993296,
      "grad_norm": 0.3479751544175198,
      "learning_rate": 8.02955311278655e-06,
      "loss": 0.033,
      "step": 4067
    },
    {
      "epoch": 1.0909090909090908,
      "grad_norm": 0.26280201990364777,
      "learning_rate": 8.028311785456778e-06,
      "loss": 0.0253,
      "step": 4068
    },
    {
      "epoch": 1.0911772593188522,
      "grad_norm": 0.3485378301828578,
      "learning_rate": 8.02707016326833e-06,
      "loss": 0.0318,
      "step": 4069
    },
    {
      "epoch": 1.0914454277286136,
      "grad_norm": 0.32296027816786327,
      "learning_rate": 8.025828246342096e-06,
      "loss": 0.028,
      "step": 4070
    },
    {
      "epoch": 1.0917135961383748,
      "grad_norm": 0.3752148376626047,
      "learning_rate": 8.024586034798998e-06,
      "loss": 0.0248,
      "step": 4071
    },
    {
      "epoch": 1.0919817645481362,
      "grad_norm": 0.2916201458555858,
      "learning_rate": 8.02334352875999e-06,
      "loss": 0.0279,
      "step": 4072
    },
    {
      "epoch": 1.0922499329578976,
      "grad_norm": 0.30133250812374757,
      "learning_rate": 8.022100728346048e-06,
      "loss": 0.0226,
      "step": 4073
    },
    {
      "epoch": 1.0925181013676588,
      "grad_norm": 0.2970744858560541,
      "learning_rate": 8.02085763367818e-06,
      "loss": 0.0236,
      "step": 4074
    },
    {
      "epoch": 1.0927862697774202,
      "grad_norm": 0.35383393670432006,
      "learning_rate": 8.019614244877429e-06,
      "loss": 0.0388,
      "step": 4075
    },
    {
      "epoch": 1.0930544381871816,
      "grad_norm": 0.19300881920988497,
      "learning_rate": 8.018370562064852e-06,
      "loss": 0.0199,
      "step": 4076
    },
    {
      "epoch": 1.0933226065969428,
      "grad_norm": 0.31978837121061177,
      "learning_rate": 8.017126585361548e-06,
      "loss": 0.0266,
      "step": 4077
    },
    {
      "epoch": 1.0935907750067042,
      "grad_norm": 0.30849703002597245,
      "learning_rate": 8.015882314888638e-06,
      "loss": 0.0299,
      "step": 4078
    },
    {
      "epoch": 1.0938589434164656,
      "grad_norm": 0.32214237088160264,
      "learning_rate": 8.014637750767276e-06,
      "loss": 0.0335,
      "step": 4079
    },
    {
      "epoch": 1.0941271118262268,
      "grad_norm": 0.27209436953459176,
      "learning_rate": 8.01339289311864e-06,
      "loss": 0.0221,
      "step": 4080
    },
    {
      "epoch": 1.0943952802359882,
      "grad_norm": 0.2563874563309693,
      "learning_rate": 8.012147742063936e-06,
      "loss": 0.026,
      "step": 4081
    },
    {
      "epoch": 1.0946634486457496,
      "grad_norm": 0.3253618024823035,
      "learning_rate": 8.010902297724404e-06,
      "loss": 0.0225,
      "step": 4082
    },
    {
      "epoch": 1.0949316170555108,
      "grad_norm": 0.34270618754671317,
      "learning_rate": 8.00965656022131e-06,
      "loss": 0.0296,
      "step": 4083
    },
    {
      "epoch": 1.0951997854652722,
      "grad_norm": 0.280581188469629,
      "learning_rate": 8.008410529675947e-06,
      "loss": 0.0277,
      "step": 4084
    },
    {
      "epoch": 1.0954679538750336,
      "grad_norm": 0.410099538397887,
      "learning_rate": 8.007164206209637e-06,
      "loss": 0.027,
      "step": 4085
    },
    {
      "epoch": 1.0957361222847948,
      "grad_norm": 0.28806532649586924,
      "learning_rate": 8.005917589943734e-06,
      "loss": 0.0277,
      "step": 4086
    },
    {
      "epoch": 1.0960042906945562,
      "grad_norm": 0.3224546854164573,
      "learning_rate": 8.004670680999615e-06,
      "loss": 0.0251,
      "step": 4087
    },
    {
      "epoch": 1.0962724591043176,
      "grad_norm": 0.2703998670068158,
      "learning_rate": 8.003423479498689e-06,
      "loss": 0.0176,
      "step": 4088
    },
    {
      "epoch": 1.0965406275140788,
      "grad_norm": 0.2318160085559615,
      "learning_rate": 8.002175985562392e-06,
      "loss": 0.0262,
      "step": 4089
    },
    {
      "epoch": 1.0968087959238402,
      "grad_norm": 0.44641462746047506,
      "learning_rate": 8.00092819931219e-06,
      "loss": 0.0263,
      "step": 4090
    },
    {
      "epoch": 1.0970769643336016,
      "grad_norm": 0.2891948519662771,
      "learning_rate": 7.999680120869578e-06,
      "loss": 0.0218,
      "step": 4091
    },
    {
      "epoch": 1.0973451327433628,
      "grad_norm": 0.2839405783514393,
      "learning_rate": 7.998431750356074e-06,
      "loss": 0.0248,
      "step": 4092
    },
    {
      "epoch": 1.0976133011531242,
      "grad_norm": 0.26092561202577574,
      "learning_rate": 7.997183087893233e-06,
      "loss": 0.0228,
      "step": 4093
    },
    {
      "epoch": 1.0978814695628856,
      "grad_norm": 0.2977862387545977,
      "learning_rate": 7.995934133602633e-06,
      "loss": 0.0293,
      "step": 4094
    },
    {
      "epoch": 1.0981496379726468,
      "grad_norm": 0.3074935005603619,
      "learning_rate": 7.994684887605877e-06,
      "loss": 0.0311,
      "step": 4095
    },
    {
      "epoch": 1.0984178063824082,
      "grad_norm": 0.2916319802357935,
      "learning_rate": 7.993435350024608e-06,
      "loss": 0.03,
      "step": 4096
    },
    {
      "epoch": 1.0986859747921696,
      "grad_norm": 0.2693427729525745,
      "learning_rate": 7.992185520980484e-06,
      "loss": 0.0217,
      "step": 4097
    },
    {
      "epoch": 1.0989541432019307,
      "grad_norm": 0.2746059920526003,
      "learning_rate": 7.9909354005952e-06,
      "loss": 0.0257,
      "step": 4098
    },
    {
      "epoch": 1.0992223116116921,
      "grad_norm": 0.30469054896273656,
      "learning_rate": 7.98968498899048e-06,
      "loss": 0.0282,
      "step": 4099
    },
    {
      "epoch": 1.0994904800214536,
      "grad_norm": 0.3051123006979975,
      "learning_rate": 7.988434286288067e-06,
      "loss": 0.0342,
      "step": 4100
    },
    {
      "epoch": 1.0997586484312147,
      "grad_norm": 0.24741231267079192,
      "learning_rate": 7.987183292609743e-06,
      "loss": 0.0185,
      "step": 4101
    },
    {
      "epoch": 1.1000268168409761,
      "grad_norm": 0.2352936467387726,
      "learning_rate": 7.985932008077311e-06,
      "loss": 0.0235,
      "step": 4102
    },
    {
      "epoch": 1.1002949852507375,
      "grad_norm": 0.29503384550294093,
      "learning_rate": 7.984680432812608e-06,
      "loss": 0.0266,
      "step": 4103
    },
    {
      "epoch": 1.1005631536604987,
      "grad_norm": 0.40973989250210113,
      "learning_rate": 7.983428566937495e-06,
      "loss": 0.0282,
      "step": 4104
    },
    {
      "epoch": 1.1008313220702601,
      "grad_norm": 0.3206852796346588,
      "learning_rate": 7.982176410573865e-06,
      "loss": 0.0321,
      "step": 4105
    },
    {
      "epoch": 1.1010994904800215,
      "grad_norm": 0.27259908659708826,
      "learning_rate": 7.980923963843634e-06,
      "loss": 0.0266,
      "step": 4106
    },
    {
      "epoch": 1.1013676588897827,
      "grad_norm": 0.2880521717002182,
      "learning_rate": 7.97967122686875e-06,
      "loss": 0.0228,
      "step": 4107
    },
    {
      "epoch": 1.1016358272995441,
      "grad_norm": 0.29752434595387006,
      "learning_rate": 7.97841819977119e-06,
      "loss": 0.0307,
      "step": 4108
    },
    {
      "epoch": 1.1019039957093055,
      "grad_norm": 0.3522697399587234,
      "learning_rate": 7.977164882672958e-06,
      "loss": 0.0293,
      "step": 4109
    },
    {
      "epoch": 1.1021721641190667,
      "grad_norm": 0.47534931460561736,
      "learning_rate": 7.975911275696086e-06,
      "loss": 0.0296,
      "step": 4110
    },
    {
      "epoch": 1.1024403325288281,
      "grad_norm": 0.2962036124613042,
      "learning_rate": 7.974657378962633e-06,
      "loss": 0.023,
      "step": 4111
    },
    {
      "epoch": 1.1027085009385895,
      "grad_norm": 0.3441166925427178,
      "learning_rate": 7.97340319259469e-06,
      "loss": 0.0332,
      "step": 4112
    },
    {
      "epoch": 1.1029766693483507,
      "grad_norm": 0.305179966381331,
      "learning_rate": 7.972148716714371e-06,
      "loss": 0.0232,
      "step": 4113
    },
    {
      "epoch": 1.103244837758112,
      "grad_norm": 0.34390807825084896,
      "learning_rate": 7.970893951443822e-06,
      "loss": 0.037,
      "step": 4114
    },
    {
      "epoch": 1.1035130061678735,
      "grad_norm": 0.35274026351897547,
      "learning_rate": 7.969638896905217e-06,
      "loss": 0.024,
      "step": 4115
    },
    {
      "epoch": 1.1037811745776347,
      "grad_norm": 0.2815630527758902,
      "learning_rate": 7.968383553220757e-06,
      "loss": 0.025,
      "step": 4116
    },
    {
      "epoch": 1.104049342987396,
      "grad_norm": 0.4146906240158036,
      "learning_rate": 7.967127920512672e-06,
      "loss": 0.0315,
      "step": 4117
    },
    {
      "epoch": 1.1043175113971575,
      "grad_norm": 0.26493684010792395,
      "learning_rate": 7.965871998903219e-06,
      "loss": 0.0213,
      "step": 4118
    },
    {
      "epoch": 1.1045856798069187,
      "grad_norm": 0.23645294919797766,
      "learning_rate": 7.964615788514683e-06,
      "loss": 0.0176,
      "step": 4119
    },
    {
      "epoch": 1.10485384821668,
      "grad_norm": 1.3698124769199465,
      "learning_rate": 7.963359289469378e-06,
      "loss": 0.0339,
      "step": 4120
    },
    {
      "epoch": 1.1051220166264415,
      "grad_norm": 0.238507584414975,
      "learning_rate": 7.962102501889647e-06,
      "loss": 0.0268,
      "step": 4121
    },
    {
      "epoch": 1.1053901850362027,
      "grad_norm": 0.41823921052502794,
      "learning_rate": 7.96084542589786e-06,
      "loss": 0.0326,
      "step": 4122
    },
    {
      "epoch": 1.105658353445964,
      "grad_norm": 0.41418909189497843,
      "learning_rate": 7.959588061616415e-06,
      "loss": 0.0346,
      "step": 4123
    },
    {
      "epoch": 1.1059265218557255,
      "grad_norm": 0.35998392424701786,
      "learning_rate": 7.958330409167737e-06,
      "loss": 0.0224,
      "step": 4124
    },
    {
      "epoch": 1.1061946902654867,
      "grad_norm": 0.2666897462502224,
      "learning_rate": 7.957072468674282e-06,
      "loss": 0.0233,
      "step": 4125
    },
    {
      "epoch": 1.106462858675248,
      "grad_norm": 0.26985241281361594,
      "learning_rate": 7.955814240258532e-06,
      "loss": 0.0187,
      "step": 4126
    },
    {
      "epoch": 1.1067310270850095,
      "grad_norm": 0.35953815483619417,
      "learning_rate": 7.954555724042996e-06,
      "loss": 0.0291,
      "step": 4127
    },
    {
      "epoch": 1.1069991954947707,
      "grad_norm": 0.41332380527038576,
      "learning_rate": 7.953296920150213e-06,
      "loss": 0.0287,
      "step": 4128
    },
    {
      "epoch": 1.107267363904532,
      "grad_norm": 0.370337189715217,
      "learning_rate": 7.952037828702752e-06,
      "loss": 0.0224,
      "step": 4129
    },
    {
      "epoch": 1.1075355323142935,
      "grad_norm": 0.4690741665195384,
      "learning_rate": 7.950778449823203e-06,
      "loss": 0.0259,
      "step": 4130
    },
    {
      "epoch": 1.1078037007240547,
      "grad_norm": 0.2835463083034622,
      "learning_rate": 7.949518783634191e-06,
      "loss": 0.0241,
      "step": 4131
    },
    {
      "epoch": 1.108071869133816,
      "grad_norm": 0.5299333996966844,
      "learning_rate": 7.948258830258366e-06,
      "loss": 0.0204,
      "step": 4132
    },
    {
      "epoch": 1.1083400375435775,
      "grad_norm": 0.4309818192664088,
      "learning_rate": 7.946998589818406e-06,
      "loss": 0.0314,
      "step": 4133
    },
    {
      "epoch": 1.1086082059533386,
      "grad_norm": 0.42446611482525437,
      "learning_rate": 7.945738062437018e-06,
      "loss": 0.0383,
      "step": 4134
    },
    {
      "epoch": 1.1088763743631,
      "grad_norm": 0.5847952278898373,
      "learning_rate": 7.944477248236934e-06,
      "loss": 0.0278,
      "step": 4135
    },
    {
      "epoch": 1.1091445427728615,
      "grad_norm": 0.28954735959107586,
      "learning_rate": 7.94321614734092e-06,
      "loss": 0.0219,
      "step": 4136
    },
    {
      "epoch": 1.1094127111826226,
      "grad_norm": 0.26011725131205254,
      "learning_rate": 7.941954759871763e-06,
      "loss": 0.0236,
      "step": 4137
    },
    {
      "epoch": 1.109680879592384,
      "grad_norm": 0.30336296060694157,
      "learning_rate": 7.940693085952281e-06,
      "loss": 0.0324,
      "step": 4138
    },
    {
      "epoch": 1.1099490480021454,
      "grad_norm": 0.2322930076356446,
      "learning_rate": 7.93943112570532e-06,
      "loss": 0.0273,
      "step": 4139
    },
    {
      "epoch": 1.1102172164119066,
      "grad_norm": 0.29883359732806297,
      "learning_rate": 7.938168879253757e-06,
      "loss": 0.0293,
      "step": 4140
    },
    {
      "epoch": 1.110485384821668,
      "grad_norm": 0.24570490921669041,
      "learning_rate": 7.93690634672049e-06,
      "loss": 0.0241,
      "step": 4141
    },
    {
      "epoch": 1.1107535532314294,
      "grad_norm": 0.3376069571332474,
      "learning_rate": 7.935643528228448e-06,
      "loss": 0.0363,
      "step": 4142
    },
    {
      "epoch": 1.1110217216411906,
      "grad_norm": 0.2663791489009323,
      "learning_rate": 7.934380423900591e-06,
      "loss": 0.0256,
      "step": 4143
    },
    {
      "epoch": 1.111289890050952,
      "grad_norm": 0.2670221037078257,
      "learning_rate": 7.933117033859903e-06,
      "loss": 0.0278,
      "step": 4144
    },
    {
      "epoch": 1.1115580584607134,
      "grad_norm": 0.3265573900272089,
      "learning_rate": 7.931853358229396e-06,
      "loss": 0.0247,
      "step": 4145
    },
    {
      "epoch": 1.1118262268704746,
      "grad_norm": 0.2944649524168005,
      "learning_rate": 7.93058939713211e-06,
      "loss": 0.0304,
      "step": 4146
    },
    {
      "epoch": 1.112094395280236,
      "grad_norm": 0.2429980744319957,
      "learning_rate": 7.929325150691116e-06,
      "loss": 0.024,
      "step": 4147
    },
    {
      "epoch": 1.1123625636899974,
      "grad_norm": 0.30571124497970364,
      "learning_rate": 7.92806061902951e-06,
      "loss": 0.0255,
      "step": 4148
    },
    {
      "epoch": 1.1126307320997586,
      "grad_norm": 0.35103142715837277,
      "learning_rate": 7.926795802270415e-06,
      "loss": 0.0306,
      "step": 4149
    },
    {
      "epoch": 1.11289890050952,
      "grad_norm": 0.34649021647851863,
      "learning_rate": 7.925530700536982e-06,
      "loss": 0.0263,
      "step": 4150
    },
    {
      "epoch": 1.1131670689192814,
      "grad_norm": 0.28368951759623695,
      "learning_rate": 7.924265313952393e-06,
      "loss": 0.0258,
      "step": 4151
    },
    {
      "epoch": 1.1134352373290426,
      "grad_norm": 0.347437968716451,
      "learning_rate": 7.922999642639854e-06,
      "loss": 0.037,
      "step": 4152
    },
    {
      "epoch": 1.113703405738804,
      "grad_norm": 0.335328594290532,
      "learning_rate": 7.921733686722602e-06,
      "loss": 0.0208,
      "step": 4153
    },
    {
      "epoch": 1.1139715741485654,
      "grad_norm": 0.49890591533761997,
      "learning_rate": 7.920467446323896e-06,
      "loss": 0.0225,
      "step": 4154
    },
    {
      "epoch": 1.1142397425583266,
      "grad_norm": 0.1888072012916062,
      "learning_rate": 7.919200921567029e-06,
      "loss": 0.0177,
      "step": 4155
    },
    {
      "epoch": 1.114507910968088,
      "grad_norm": 0.2428824590146001,
      "learning_rate": 7.917934112575318e-06,
      "loss": 0.0244,
      "step": 4156
    },
    {
      "epoch": 1.1147760793778492,
      "grad_norm": 0.6292839446651106,
      "learning_rate": 7.916667019472111e-06,
      "loss": 0.0253,
      "step": 4157
    },
    {
      "epoch": 1.1150442477876106,
      "grad_norm": 0.2797895336527246,
      "learning_rate": 7.915399642380778e-06,
      "loss": 0.022,
      "step": 4158
    },
    {
      "epoch": 1.115312416197372,
      "grad_norm": 0.3709033263265123,
      "learning_rate": 7.914131981424722e-06,
      "loss": 0.0245,
      "step": 4159
    },
    {
      "epoch": 1.1155805846071334,
      "grad_norm": 0.37350488573930757,
      "learning_rate": 7.912864036727374e-06,
      "loss": 0.0277,
      "step": 4160
    },
    {
      "epoch": 1.1158487530168946,
      "grad_norm": 0.35591135124029405,
      "learning_rate": 7.911595808412186e-06,
      "loss": 0.0293,
      "step": 4161
    },
    {
      "epoch": 1.116116921426656,
      "grad_norm": 0.3323658415209433,
      "learning_rate": 7.910327296602644e-06,
      "loss": 0.0274,
      "step": 4162
    },
    {
      "epoch": 1.1163850898364172,
      "grad_norm": 0.33674838702454474,
      "learning_rate": 7.909058501422262e-06,
      "loss": 0.0296,
      "step": 4163
    },
    {
      "epoch": 1.1166532582461786,
      "grad_norm": 0.30186607148297506,
      "learning_rate": 7.907789422994577e-06,
      "loss": 0.0361,
      "step": 4164
    },
    {
      "epoch": 1.11692142665594,
      "grad_norm": 0.32897186080708757,
      "learning_rate": 7.906520061443154e-06,
      "loss": 0.0368,
      "step": 4165
    },
    {
      "epoch": 1.1171895950657014,
      "grad_norm": 0.394905437197485,
      "learning_rate": 7.90525041689159e-06,
      "loss": 0.0277,
      "step": 4166
    },
    {
      "epoch": 1.1174577634754626,
      "grad_norm": 0.40044966362355716,
      "learning_rate": 7.903980489463507e-06,
      "loss": 0.0254,
      "step": 4167
    },
    {
      "epoch": 1.117725931885224,
      "grad_norm": 0.35285803009717864,
      "learning_rate": 7.902710279282554e-06,
      "loss": 0.0293,
      "step": 4168
    },
    {
      "epoch": 1.1179941002949851,
      "grad_norm": 0.2549185645416134,
      "learning_rate": 7.901439786472408e-06,
      "loss": 0.0209,
      "step": 4169
    },
    {
      "epoch": 1.1182622687047465,
      "grad_norm": 0.2965982372943463,
      "learning_rate": 7.90016901115677e-06,
      "loss": 0.032,
      "step": 4170
    },
    {
      "epoch": 1.118530437114508,
      "grad_norm": 0.27237683647084293,
      "learning_rate": 7.898897953459376e-06,
      "loss": 0.0232,
      "step": 4171
    },
    {
      "epoch": 1.1187986055242694,
      "grad_norm": 0.3157510762942235,
      "learning_rate": 7.897626613503988e-06,
      "loss": 0.0315,
      "step": 4172
    },
    {
      "epoch": 1.1190667739340305,
      "grad_norm": 0.3819584516368457,
      "learning_rate": 7.896354991414387e-06,
      "loss": 0.0405,
      "step": 4173
    },
    {
      "epoch": 1.119334942343792,
      "grad_norm": 0.3813124136664552,
      "learning_rate": 7.895083087314391e-06,
      "loss": 0.0289,
      "step": 4174
    },
    {
      "epoch": 1.1196031107535531,
      "grad_norm": 0.2559914631645647,
      "learning_rate": 7.89381090132784e-06,
      "loss": 0.0261,
      "step": 4175
    },
    {
      "epoch": 1.1198712791633145,
      "grad_norm": 0.3851260813916714,
      "learning_rate": 7.892538433578607e-06,
      "loss": 0.0205,
      "step": 4176
    },
    {
      "epoch": 1.120139447573076,
      "grad_norm": 0.32815564214621246,
      "learning_rate": 7.891265684190583e-06,
      "loss": 0.0326,
      "step": 4177
    },
    {
      "epoch": 1.1204076159828373,
      "grad_norm": 0.40285146442820446,
      "learning_rate": 7.889992653287698e-06,
      "loss": 0.0267,
      "step": 4178
    },
    {
      "epoch": 1.1206757843925985,
      "grad_norm": 0.2797457296293269,
      "learning_rate": 7.8887193409939e-06,
      "loss": 0.0252,
      "step": 4179
    },
    {
      "epoch": 1.12094395280236,
      "grad_norm": 0.27093535622584985,
      "learning_rate": 7.887445747433169e-06,
      "loss": 0.0224,
      "step": 4180
    },
    {
      "epoch": 1.121212121212121,
      "grad_norm": 0.3803487139516584,
      "learning_rate": 7.886171872729513e-06,
      "loss": 0.029,
      "step": 4181
    },
    {
      "epoch": 1.1214802896218825,
      "grad_norm": 0.8751256647126664,
      "learning_rate": 7.884897717006962e-06,
      "loss": 0.0249,
      "step": 4182
    },
    {
      "epoch": 1.121748458031644,
      "grad_norm": 0.3298758675712611,
      "learning_rate": 7.883623280389581e-06,
      "loss": 0.0307,
      "step": 4183
    },
    {
      "epoch": 1.1220166264414053,
      "grad_norm": 0.26013915217697675,
      "learning_rate": 7.882348563001459e-06,
      "loss": 0.0193,
      "step": 4184
    },
    {
      "epoch": 1.1222847948511665,
      "grad_norm": 0.31239487604244787,
      "learning_rate": 7.881073564966706e-06,
      "loss": 0.0314,
      "step": 4185
    },
    {
      "epoch": 1.122552963260928,
      "grad_norm": 0.29606126443292674,
      "learning_rate": 7.879798286409471e-06,
      "loss": 0.0256,
      "step": 4186
    },
    {
      "epoch": 1.122821131670689,
      "grad_norm": 0.25036759598590835,
      "learning_rate": 7.878522727453924e-06,
      "loss": 0.021,
      "step": 4187
    },
    {
      "epoch": 1.1230893000804505,
      "grad_norm": 0.34537980384278044,
      "learning_rate": 7.87724688822426e-06,
      "loss": 0.0244,
      "step": 4188
    },
    {
      "epoch": 1.123357468490212,
      "grad_norm": 0.3848214154466663,
      "learning_rate": 7.875970768844703e-06,
      "loss": 0.024,
      "step": 4189
    },
    {
      "epoch": 1.123625636899973,
      "grad_norm": 0.32802128320292184,
      "learning_rate": 7.874694369439511e-06,
      "loss": 0.026,
      "step": 4190
    },
    {
      "epoch": 1.1238938053097345,
      "grad_norm": 0.29409589510706435,
      "learning_rate": 7.87341769013296e-06,
      "loss": 0.0256,
      "step": 4191
    },
    {
      "epoch": 1.1241619737194959,
      "grad_norm": 0.39505588215971155,
      "learning_rate": 7.872140731049356e-06,
      "loss": 0.0311,
      "step": 4192
    },
    {
      "epoch": 1.124430142129257,
      "grad_norm": 0.30949360256394914,
      "learning_rate": 7.870863492313035e-06,
      "loss": 0.0251,
      "step": 4193
    },
    {
      "epoch": 1.1246983105390185,
      "grad_norm": 0.4524253177076165,
      "learning_rate": 7.869585974048359e-06,
      "loss": 0.0249,
      "step": 4194
    },
    {
      "epoch": 1.1249664789487799,
      "grad_norm": 0.2907674867665826,
      "learning_rate": 7.868308176379713e-06,
      "loss": 0.0293,
      "step": 4195
    },
    {
      "epoch": 1.1252346473585413,
      "grad_norm": 0.2804988109618617,
      "learning_rate": 7.867030099431517e-06,
      "loss": 0.0238,
      "step": 4196
    },
    {
      "epoch": 1.1255028157683025,
      "grad_norm": 0.322025921058777,
      "learning_rate": 7.865751743328211e-06,
      "loss": 0.0286,
      "step": 4197
    },
    {
      "epoch": 1.1257709841780639,
      "grad_norm": 0.24311528037573407,
      "learning_rate": 7.864473108194268e-06,
      "loss": 0.0257,
      "step": 4198
    },
    {
      "epoch": 1.126039152587825,
      "grad_norm": 0.283552249667885,
      "learning_rate": 7.86319419415418e-06,
      "loss": 0.0213,
      "step": 4199
    },
    {
      "epoch": 1.1263073209975865,
      "grad_norm": 0.34624401723656195,
      "learning_rate": 7.861915001332479e-06,
      "loss": 0.0385,
      "step": 4200
    },
    {
      "epoch": 1.1265754894073479,
      "grad_norm": 0.34788546625631106,
      "learning_rate": 7.860635529853708e-06,
      "loss": 0.026,
      "step": 4201
    },
    {
      "epoch": 1.1268436578171093,
      "grad_norm": 0.2502238275382981,
      "learning_rate": 7.859355779842455e-06,
      "loss": 0.0239,
      "step": 4202
    },
    {
      "epoch": 1.1271118262268705,
      "grad_norm": 0.25643857948199306,
      "learning_rate": 7.858075751423319e-06,
      "loss": 0.024,
      "step": 4203
    },
    {
      "epoch": 1.1273799946366319,
      "grad_norm": 0.3768920264595995,
      "learning_rate": 7.856795444720934e-06,
      "loss": 0.031,
      "step": 4204
    },
    {
      "epoch": 1.127648163046393,
      "grad_norm": 0.32855784639250407,
      "learning_rate": 7.855514859859963e-06,
      "loss": 0.0313,
      "step": 4205
    },
    {
      "epoch": 1.1279163314561544,
      "grad_norm": 0.31682520285739896,
      "learning_rate": 7.854233996965091e-06,
      "loss": 0.0252,
      "step": 4206
    },
    {
      "epoch": 1.1281844998659158,
      "grad_norm": 0.30662700997080405,
      "learning_rate": 7.852952856161031e-06,
      "loss": 0.0187,
      "step": 4207
    },
    {
      "epoch": 1.128452668275677,
      "grad_norm": 0.39238111491725974,
      "learning_rate": 7.851671437572528e-06,
      "loss": 0.0235,
      "step": 4208
    },
    {
      "epoch": 1.1287208366854384,
      "grad_norm": 0.35376105569710337,
      "learning_rate": 7.850389741324347e-06,
      "loss": 0.0294,
      "step": 4209
    },
    {
      "epoch": 1.1289890050951998,
      "grad_norm": 0.511336622821663,
      "learning_rate": 7.849107767541284e-06,
      "loss": 0.0313,
      "step": 4210
    },
    {
      "epoch": 1.129257173504961,
      "grad_norm": 0.2961101797776158,
      "learning_rate": 7.847825516348164e-06,
      "loss": 0.0255,
      "step": 4211
    },
    {
      "epoch": 1.1295253419147224,
      "grad_norm": 0.4827423105268493,
      "learning_rate": 7.846542987869832e-06,
      "loss": 0.0201,
      "step": 4212
    },
    {
      "epoch": 1.1297935103244838,
      "grad_norm": 0.5004546710916487,
      "learning_rate": 7.845260182231166e-06,
      "loss": 0.0305,
      "step": 4213
    },
    {
      "epoch": 1.130061678734245,
      "grad_norm": 0.35102034172555613,
      "learning_rate": 7.843977099557075e-06,
      "loss": 0.0272,
      "step": 4214
    },
    {
      "epoch": 1.1303298471440064,
      "grad_norm": 0.3400291502830289,
      "learning_rate": 7.84269373997248e-06,
      "loss": 0.0283,
      "step": 4215
    },
    {
      "epoch": 1.1305980155537678,
      "grad_norm": 0.4477580913977664,
      "learning_rate": 7.841410103602345e-06,
      "loss": 0.052,
      "step": 4216
    },
    {
      "epoch": 1.130866183963529,
      "grad_norm": 0.5295502076277343,
      "learning_rate": 7.84012619057165e-06,
      "loss": 0.0282,
      "step": 4217
    },
    {
      "epoch": 1.1311343523732904,
      "grad_norm": 0.2537692994848566,
      "learning_rate": 7.83884200100541e-06,
      "loss": 0.0206,
      "step": 4218
    },
    {
      "epoch": 1.1314025207830518,
      "grad_norm": 0.3237797465001269,
      "learning_rate": 7.83755753502866e-06,
      "loss": 0.0286,
      "step": 4219
    },
    {
      "epoch": 1.131670689192813,
      "grad_norm": 0.2368752717365191,
      "learning_rate": 7.836272792766467e-06,
      "loss": 0.0211,
      "step": 4220
    },
    {
      "epoch": 1.1319388576025744,
      "grad_norm": 0.2620768265194458,
      "learning_rate": 7.83498777434392e-06,
      "loss": 0.0258,
      "step": 4221
    },
    {
      "epoch": 1.1322070260123358,
      "grad_norm": 0.5178884195798888,
      "learning_rate": 7.833702479886143e-06,
      "loss": 0.0266,
      "step": 4222
    },
    {
      "epoch": 1.132475194422097,
      "grad_norm": 0.24665835650074486,
      "learning_rate": 7.832416909518278e-06,
      "loss": 0.0179,
      "step": 4223
    },
    {
      "epoch": 1.1327433628318584,
      "grad_norm": 0.2936674418331212,
      "learning_rate": 7.8311310633655e-06,
      "loss": 0.0295,
      "step": 4224
    },
    {
      "epoch": 1.1330115312416198,
      "grad_norm": 0.21644312525985643,
      "learning_rate": 7.829844941553004e-06,
      "loss": 0.0204,
      "step": 4225
    },
    {
      "epoch": 1.133279699651381,
      "grad_norm": 0.6281995662689923,
      "learning_rate": 7.828558544206022e-06,
      "loss": 0.0376,
      "step": 4226
    },
    {
      "epoch": 1.1335478680611424,
      "grad_norm": 0.32516705681810687,
      "learning_rate": 7.827271871449803e-06,
      "loss": 0.0302,
      "step": 4227
    },
    {
      "epoch": 1.1338160364709038,
      "grad_norm": 0.369947823275506,
      "learning_rate": 7.825984923409627e-06,
      "loss": 0.0326,
      "step": 4228
    },
    {
      "epoch": 1.134084204880665,
      "grad_norm": 0.3660119726816325,
      "learning_rate": 7.824697700210804e-06,
      "loss": 0.0354,
      "step": 4229
    },
    {
      "epoch": 1.1343523732904264,
      "grad_norm": 0.2944373871083325,
      "learning_rate": 7.823410201978663e-06,
      "loss": 0.0276,
      "step": 4230
    },
    {
      "epoch": 1.1346205417001878,
      "grad_norm": 0.29410204219220454,
      "learning_rate": 7.822122428838567e-06,
      "loss": 0.0231,
      "step": 4231
    },
    {
      "epoch": 1.134888710109949,
      "grad_norm": 0.3461791241750489,
      "learning_rate": 7.820834380915905e-06,
      "loss": 0.0272,
      "step": 4232
    },
    {
      "epoch": 1.1351568785197104,
      "grad_norm": 0.6258491007417226,
      "learning_rate": 7.819546058336087e-06,
      "loss": 0.0258,
      "step": 4233
    },
    {
      "epoch": 1.1354250469294718,
      "grad_norm": 0.23974911076342684,
      "learning_rate": 7.818257461224556e-06,
      "loss": 0.0196,
      "step": 4234
    },
    {
      "epoch": 1.135693215339233,
      "grad_norm": 0.28849904997014614,
      "learning_rate": 7.816968589706776e-06,
      "loss": 0.0256,
      "step": 4235
    },
    {
      "epoch": 1.1359613837489944,
      "grad_norm": 0.2745885167132827,
      "learning_rate": 7.815679443908246e-06,
      "loss": 0.0204,
      "step": 4236
    },
    {
      "epoch": 1.1362295521587558,
      "grad_norm": 0.3612156690860062,
      "learning_rate": 7.814390023954482e-06,
      "loss": 0.0298,
      "step": 4237
    },
    {
      "epoch": 1.136497720568517,
      "grad_norm": 0.2634264638326276,
      "learning_rate": 7.813100329971034e-06,
      "loss": 0.0285,
      "step": 4238
    },
    {
      "epoch": 1.1367658889782783,
      "grad_norm": 0.24617403448572123,
      "learning_rate": 7.811810362083476e-06,
      "loss": 0.0188,
      "step": 4239
    },
    {
      "epoch": 1.1370340573880398,
      "grad_norm": 0.33102241663509985,
      "learning_rate": 7.810520120417408e-06,
      "loss": 0.0286,
      "step": 4240
    },
    {
      "epoch": 1.137302225797801,
      "grad_norm": 0.2654474689221968,
      "learning_rate": 7.809229605098458e-06,
      "loss": 0.0256,
      "step": 4241
    },
    {
      "epoch": 1.1375703942075623,
      "grad_norm": 0.3962137473269484,
      "learning_rate": 7.807938816252279e-06,
      "loss": 0.0324,
      "step": 4242
    },
    {
      "epoch": 1.1378385626173237,
      "grad_norm": 0.32330450061901295,
      "learning_rate": 7.806647754004553e-06,
      "loss": 0.0325,
      "step": 4243
    },
    {
      "epoch": 1.138106731027085,
      "grad_norm": 0.678484345510273,
      "learning_rate": 7.805356418480988e-06,
      "loss": 0.0329,
      "step": 4244
    },
    {
      "epoch": 1.1383748994368463,
      "grad_norm": 0.5607088944147768,
      "learning_rate": 7.804064809807315e-06,
      "loss": 0.032,
      "step": 4245
    },
    {
      "epoch": 1.1386430678466077,
      "grad_norm": 0.31684433010752583,
      "learning_rate": 7.802772928109295e-06,
      "loss": 0.0234,
      "step": 4246
    },
    {
      "epoch": 1.138911236256369,
      "grad_norm": 0.27692348125251015,
      "learning_rate": 7.801480773512717e-06,
      "loss": 0.0224,
      "step": 4247
    },
    {
      "epoch": 1.1391794046661303,
      "grad_norm": 0.36685219250353557,
      "learning_rate": 7.800188346143394e-06,
      "loss": 0.0262,
      "step": 4248
    },
    {
      "epoch": 1.1394475730758917,
      "grad_norm": 0.39491953956763987,
      "learning_rate": 7.798895646127166e-06,
      "loss": 0.0294,
      "step": 4249
    },
    {
      "epoch": 1.139715741485653,
      "grad_norm": 0.3346128325556546,
      "learning_rate": 7.797602673589902e-06,
      "loss": 0.022,
      "step": 4250
    },
    {
      "epoch": 1.1399839098954143,
      "grad_norm": 0.30610599247694353,
      "learning_rate": 7.79630942865749e-06,
      "loss": 0.0345,
      "step": 4251
    },
    {
      "epoch": 1.1402520783051757,
      "grad_norm": 0.4330202243592825,
      "learning_rate": 7.795015911455854e-06,
      "loss": 0.0316,
      "step": 4252
    },
    {
      "epoch": 1.140520246714937,
      "grad_norm": 0.28897876681884577,
      "learning_rate": 7.793722122110941e-06,
      "loss": 0.0203,
      "step": 4253
    },
    {
      "epoch": 1.1407884151246983,
      "grad_norm": 0.5172339099949775,
      "learning_rate": 7.79242806074872e-06,
      "loss": 0.0315,
      "step": 4254
    },
    {
      "epoch": 1.1410565835344597,
      "grad_norm": 0.33746761284792126,
      "learning_rate": 7.791133727495193e-06,
      "loss": 0.0266,
      "step": 4255
    },
    {
      "epoch": 1.141324751944221,
      "grad_norm": 0.2816554852681798,
      "learning_rate": 7.789839122476386e-06,
      "loss": 0.0173,
      "step": 4256
    },
    {
      "epoch": 1.1415929203539823,
      "grad_norm": 0.5013424642812735,
      "learning_rate": 7.78854424581835e-06,
      "loss": 0.0332,
      "step": 4257
    },
    {
      "epoch": 1.1418610887637437,
      "grad_norm": 0.3536130326308502,
      "learning_rate": 7.787249097647166e-06,
      "loss": 0.0306,
      "step": 4258
    },
    {
      "epoch": 1.1421292571735049,
      "grad_norm": 0.3199859963214248,
      "learning_rate": 7.785953678088934e-06,
      "loss": 0.021,
      "step": 4259
    },
    {
      "epoch": 1.1423974255832663,
      "grad_norm": 0.26422165616646365,
      "learning_rate": 7.784657987269791e-06,
      "loss": 0.0215,
      "step": 4260
    },
    {
      "epoch": 1.1426655939930277,
      "grad_norm": 0.2649532672309566,
      "learning_rate": 7.783362025315892e-06,
      "loss": 0.0216,
      "step": 4261
    },
    {
      "epoch": 1.1429337624027889,
      "grad_norm": 0.5290362084315056,
      "learning_rate": 7.782065792353424e-06,
      "loss": 0.0347,
      "step": 4262
    },
    {
      "epoch": 1.1432019308125503,
      "grad_norm": 0.2466631279295337,
      "learning_rate": 7.780769288508594e-06,
      "loss": 0.0171,
      "step": 4263
    },
    {
      "epoch": 1.1434700992223117,
      "grad_norm": 0.26756269310407144,
      "learning_rate": 7.779472513907643e-06,
      "loss": 0.0258,
      "step": 4264
    },
    {
      "epoch": 1.1437382676320729,
      "grad_norm": 0.31464546983450736,
      "learning_rate": 7.778175468676833e-06,
      "loss": 0.0259,
      "step": 4265
    },
    {
      "epoch": 1.1440064360418343,
      "grad_norm": 0.9465559199681399,
      "learning_rate": 7.77687815294245e-06,
      "loss": 0.0325,
      "step": 4266
    },
    {
      "epoch": 1.1442746044515957,
      "grad_norm": 0.29915735079750944,
      "learning_rate": 7.775580566830815e-06,
      "loss": 0.0231,
      "step": 4267
    },
    {
      "epoch": 1.1445427728613569,
      "grad_norm": 0.2986760320703412,
      "learning_rate": 7.774282710468272e-06,
      "loss": 0.0297,
      "step": 4268
    },
    {
      "epoch": 1.1448109412711183,
      "grad_norm": 0.36049965750516594,
      "learning_rate": 7.772984583981185e-06,
      "loss": 0.0365,
      "step": 4269
    },
    {
      "epoch": 1.1450791096808797,
      "grad_norm": 0.26777655277023116,
      "learning_rate": 7.771686187495951e-06,
      "loss": 0.0192,
      "step": 4270
    },
    {
      "epoch": 1.1453472780906409,
      "grad_norm": 0.37497343502051117,
      "learning_rate": 7.770387521138993e-06,
      "loss": 0.033,
      "step": 4271
    },
    {
      "epoch": 1.1456154465004023,
      "grad_norm": 0.2677634811512745,
      "learning_rate": 7.769088585036756e-06,
      "loss": 0.0193,
      "step": 4272
    },
    {
      "epoch": 1.1458836149101637,
      "grad_norm": 0.47689982404128367,
      "learning_rate": 7.767789379315715e-06,
      "loss": 0.0264,
      "step": 4273
    },
    {
      "epoch": 1.1461517833199248,
      "grad_norm": 0.3201366203193023,
      "learning_rate": 7.76648990410237e-06,
      "loss": 0.0255,
      "step": 4274
    },
    {
      "epoch": 1.1464199517296862,
      "grad_norm": 0.24855301247333264,
      "learning_rate": 7.76519015952325e-06,
      "loss": 0.0234,
      "step": 4275
    },
    {
      "epoch": 1.1466881201394477,
      "grad_norm": 0.6443500196832805,
      "learning_rate": 7.763890145704905e-06,
      "loss": 0.0444,
      "step": 4276
    },
    {
      "epoch": 1.1469562885492088,
      "grad_norm": 0.38965367623426517,
      "learning_rate": 7.762589862773915e-06,
      "loss": 0.0299,
      "step": 4277
    },
    {
      "epoch": 1.1472244569589702,
      "grad_norm": 0.3033499072089395,
      "learning_rate": 7.761289310856883e-06,
      "loss": 0.0213,
      "step": 4278
    },
    {
      "epoch": 1.1474926253687316,
      "grad_norm": 0.2439365103139888,
      "learning_rate": 7.759988490080442e-06,
      "loss": 0.021,
      "step": 4279
    },
    {
      "epoch": 1.1477607937784928,
      "grad_norm": 0.39021049452889756,
      "learning_rate": 7.758687400571249e-06,
      "loss": 0.0372,
      "step": 4280
    },
    {
      "epoch": 1.1480289621882542,
      "grad_norm": 0.41446979545727886,
      "learning_rate": 7.75738604245599e-06,
      "loss": 0.0268,
      "step": 4281
    },
    {
      "epoch": 1.1482971305980156,
      "grad_norm": 0.23676277417677782,
      "learning_rate": 7.756084415861372e-06,
      "loss": 0.0177,
      "step": 4282
    },
    {
      "epoch": 1.1485652990077768,
      "grad_norm": 0.3458059722111906,
      "learning_rate": 7.754782520914132e-06,
      "loss": 0.0262,
      "step": 4283
    },
    {
      "epoch": 1.1488334674175382,
      "grad_norm": 0.43022796192827195,
      "learning_rate": 7.753480357741031e-06,
      "loss": 0.0318,
      "step": 4284
    },
    {
      "epoch": 1.1491016358272996,
      "grad_norm": 0.3685181160512033,
      "learning_rate": 7.752177926468858e-06,
      "loss": 0.0309,
      "step": 4285
    },
    {
      "epoch": 1.1493698042370608,
      "grad_norm": 0.29869264451200384,
      "learning_rate": 7.750875227224427e-06,
      "loss": 0.0265,
      "step": 4286
    },
    {
      "epoch": 1.1496379726468222,
      "grad_norm": 0.20817491676799332,
      "learning_rate": 7.749572260134578e-06,
      "loss": 0.0195,
      "step": 4287
    },
    {
      "epoch": 1.1499061410565836,
      "grad_norm": 0.3010388268156244,
      "learning_rate": 7.74826902532618e-06,
      "loss": 0.0313,
      "step": 4288
    },
    {
      "epoch": 1.1501743094663448,
      "grad_norm": 0.29954934878178924,
      "learning_rate": 7.746965522926122e-06,
      "loss": 0.0272,
      "step": 4289
    },
    {
      "epoch": 1.1504424778761062,
      "grad_norm": 0.39997731380915796,
      "learning_rate": 7.745661753061324e-06,
      "loss": 0.0331,
      "step": 4290
    },
    {
      "epoch": 1.1507106462858676,
      "grad_norm": 0.3652876056699885,
      "learning_rate": 7.744357715858732e-06,
      "loss": 0.0269,
      "step": 4291
    },
    {
      "epoch": 1.1509788146956288,
      "grad_norm": 0.25785314584095237,
      "learning_rate": 7.743053411445314e-06,
      "loss": 0.0291,
      "step": 4292
    },
    {
      "epoch": 1.1512469831053902,
      "grad_norm": 0.368211670063683,
      "learning_rate": 7.74174883994807e-06,
      "loss": 0.0297,
      "step": 4293
    },
    {
      "epoch": 1.1515151515151516,
      "grad_norm": 0.258032882044038,
      "learning_rate": 7.740444001494019e-06,
      "loss": 0.0196,
      "step": 4294
    },
    {
      "epoch": 1.1517833199249128,
      "grad_norm": 0.28749854983365075,
      "learning_rate": 7.73913889621021e-06,
      "loss": 0.0255,
      "step": 4295
    },
    {
      "epoch": 1.1520514883346742,
      "grad_norm": 0.29263897225893865,
      "learning_rate": 7.73783352422372e-06,
      "loss": 0.0239,
      "step": 4296
    },
    {
      "epoch": 1.1523196567444356,
      "grad_norm": 0.34683086193244417,
      "learning_rate": 7.73652788566165e-06,
      "loss": 0.0345,
      "step": 4297
    },
    {
      "epoch": 1.1525878251541968,
      "grad_norm": 0.3342812025673864,
      "learning_rate": 7.735221980651123e-06,
      "loss": 0.0334,
      "step": 4298
    },
    {
      "epoch": 1.1528559935639582,
      "grad_norm": 0.6164938357874172,
      "learning_rate": 7.733915809319295e-06,
      "loss": 0.0332,
      "step": 4299
    },
    {
      "epoch": 1.1531241619737196,
      "grad_norm": 0.2654121422922887,
      "learning_rate": 7.732609371793343e-06,
      "loss": 0.0194,
      "step": 4300
    },
    {
      "epoch": 1.1533923303834808,
      "grad_norm": 0.2086328259464695,
      "learning_rate": 7.73130266820047e-06,
      "loss": 0.016,
      "step": 4301
    },
    {
      "epoch": 1.1536604987932422,
      "grad_norm": 0.4448641003453191,
      "learning_rate": 7.72999569866791e-06,
      "loss": 0.0273,
      "step": 4302
    },
    {
      "epoch": 1.1539286672030036,
      "grad_norm": 0.45066491909822026,
      "learning_rate": 7.728688463322916e-06,
      "loss": 0.0329,
      "step": 4303
    },
    {
      "epoch": 1.1541968356127648,
      "grad_norm": 0.29239175973703524,
      "learning_rate": 7.72738096229277e-06,
      "loss": 0.0228,
      "step": 4304
    },
    {
      "epoch": 1.1544650040225262,
      "grad_norm": 0.38554594269411363,
      "learning_rate": 7.72607319570478e-06,
      "loss": 0.0331,
      "step": 4305
    },
    {
      "epoch": 1.1547331724322876,
      "grad_norm": 0.3886670459018736,
      "learning_rate": 7.724765163686283e-06,
      "loss": 0.0404,
      "step": 4306
    },
    {
      "epoch": 1.1550013408420488,
      "grad_norm": 0.2082667714461019,
      "learning_rate": 7.723456866364634e-06,
      "loss": 0.0207,
      "step": 4307
    },
    {
      "epoch": 1.1552695092518102,
      "grad_norm": 0.2710229342194994,
      "learning_rate": 7.722148303867222e-06,
      "loss": 0.027,
      "step": 4308
    },
    {
      "epoch": 1.1555376776615716,
      "grad_norm": 0.3792971961133131,
      "learning_rate": 7.720839476321455e-06,
      "loss": 0.0334,
      "step": 4309
    },
    {
      "epoch": 1.1558058460713327,
      "grad_norm": 0.2832787782592128,
      "learning_rate": 7.719530383854774e-06,
      "loss": 0.0243,
      "step": 4310
    },
    {
      "epoch": 1.1560740144810941,
      "grad_norm": 0.2635144206908196,
      "learning_rate": 7.718221026594638e-06,
      "loss": 0.029,
      "step": 4311
    },
    {
      "epoch": 1.1563421828908556,
      "grad_norm": 0.30802942333593264,
      "learning_rate": 7.71691140466854e-06,
      "loss": 0.037,
      "step": 4312
    },
    {
      "epoch": 1.1566103513006167,
      "grad_norm": 0.3353185715368533,
      "learning_rate": 7.71560151820399e-06,
      "loss": 0.0235,
      "step": 4313
    },
    {
      "epoch": 1.1568785197103781,
      "grad_norm": 0.3060210952740613,
      "learning_rate": 7.71429136732853e-06,
      "loss": 0.0298,
      "step": 4314
    },
    {
      "epoch": 1.1571466881201395,
      "grad_norm": 0.39684939927741403,
      "learning_rate": 7.712980952169723e-06,
      "loss": 0.0366,
      "step": 4315
    },
    {
      "epoch": 1.1574148565299007,
      "grad_norm": 0.3570417885446842,
      "learning_rate": 7.711670272855168e-06,
      "loss": 0.0269,
      "step": 4316
    },
    {
      "epoch": 1.1576830249396621,
      "grad_norm": 0.30741797389464814,
      "learning_rate": 7.710359329512477e-06,
      "loss": 0.0255,
      "step": 4317
    },
    {
      "epoch": 1.1579511933494233,
      "grad_norm": 0.30854503925735277,
      "learning_rate": 7.709048122269294e-06,
      "loss": 0.033,
      "step": 4318
    },
    {
      "epoch": 1.1582193617591847,
      "grad_norm": 0.27007192500181315,
      "learning_rate": 7.707736651253287e-06,
      "loss": 0.0295,
      "step": 4319
    },
    {
      "epoch": 1.1584875301689461,
      "grad_norm": 0.3753033877696913,
      "learning_rate": 7.70642491659215e-06,
      "loss": 0.0274,
      "step": 4320
    },
    {
      "epoch": 1.1587556985787075,
      "grad_norm": 0.4200284521883131,
      "learning_rate": 7.705112918413604e-06,
      "loss": 0.0238,
      "step": 4321
    },
    {
      "epoch": 1.1590238669884687,
      "grad_norm": 0.3153348811269925,
      "learning_rate": 7.703800656845398e-06,
      "loss": 0.0268,
      "step": 4322
    },
    {
      "epoch": 1.1592920353982301,
      "grad_norm": 0.24869822085959417,
      "learning_rate": 7.7024881320153e-06,
      "loss": 0.0245,
      "step": 4323
    },
    {
      "epoch": 1.1595602038079913,
      "grad_norm": 0.35995238034039806,
      "learning_rate": 7.701175344051104e-06,
      "loss": 0.038,
      "step": 4324
    },
    {
      "epoch": 1.1598283722177527,
      "grad_norm": 0.2828943298263581,
      "learning_rate": 7.69986229308064e-06,
      "loss": 0.0249,
      "step": 4325
    },
    {
      "epoch": 1.160096540627514,
      "grad_norm": 0.3066586834331653,
      "learning_rate": 7.69854897923175e-06,
      "loss": 0.0221,
      "step": 4326
    },
    {
      "epoch": 1.1603647090372755,
      "grad_norm": 0.30243658433683224,
      "learning_rate": 7.697235402632313e-06,
      "loss": 0.0241,
      "step": 4327
    },
    {
      "epoch": 1.1606328774470367,
      "grad_norm": 0.2563730687665492,
      "learning_rate": 7.695921563410224e-06,
      "loss": 0.0288,
      "step": 4328
    },
    {
      "epoch": 1.160901045856798,
      "grad_norm": 0.3556467999462351,
      "learning_rate": 7.69460746169341e-06,
      "loss": 0.0267,
      "step": 4329
    },
    {
      "epoch": 1.1611692142665593,
      "grad_norm": 0.2889553548937902,
      "learning_rate": 7.693293097609823e-06,
      "loss": 0.0245,
      "step": 4330
    },
    {
      "epoch": 1.1614373826763207,
      "grad_norm": 0.29564191329218553,
      "learning_rate": 7.691978471287436e-06,
      "loss": 0.029,
      "step": 4331
    },
    {
      "epoch": 1.161705551086082,
      "grad_norm": 0.45360432971456777,
      "learning_rate": 7.690663582854252e-06,
      "loss": 0.0375,
      "step": 4332
    },
    {
      "epoch": 1.1619737194958435,
      "grad_norm": 0.27273409170922563,
      "learning_rate": 7.6893484324383e-06,
      "loss": 0.0333,
      "step": 4333
    },
    {
      "epoch": 1.1622418879056047,
      "grad_norm": 0.3239549806534504,
      "learning_rate": 7.68803302016763e-06,
      "loss": 0.0286,
      "step": 4334
    },
    {
      "epoch": 1.162510056315366,
      "grad_norm": 0.2299537341531845,
      "learning_rate": 7.686717346170323e-06,
      "loss": 0.0254,
      "step": 4335
    },
    {
      "epoch": 1.1627782247251273,
      "grad_norm": 0.3460722858313576,
      "learning_rate": 7.68540141057448e-06,
      "loss": 0.0328,
      "step": 4336
    },
    {
      "epoch": 1.1630463931348887,
      "grad_norm": 0.2444923379843955,
      "learning_rate": 7.684085213508231e-06,
      "loss": 0.0271,
      "step": 4337
    },
    {
      "epoch": 1.16331456154465,
      "grad_norm": 0.37404422052135783,
      "learning_rate": 7.682768755099731e-06,
      "loss": 0.0357,
      "step": 4338
    },
    {
      "epoch": 1.1635827299544115,
      "grad_norm": 0.2986704937362207,
      "learning_rate": 7.68145203547716e-06,
      "loss": 0.0257,
      "step": 4339
    },
    {
      "epoch": 1.1638508983641727,
      "grad_norm": 0.25364652538116367,
      "learning_rate": 7.680135054768722e-06,
      "loss": 0.0207,
      "step": 4340
    },
    {
      "epoch": 1.164119066773934,
      "grad_norm": 0.40862688252688634,
      "learning_rate": 7.678817813102652e-06,
      "loss": 0.0367,
      "step": 4341
    },
    {
      "epoch": 1.1643872351836952,
      "grad_norm": 0.2538291278555848,
      "learning_rate": 7.677500310607203e-06,
      "loss": 0.0288,
      "step": 4342
    },
    {
      "epoch": 1.1646554035934567,
      "grad_norm": 0.29825961844099835,
      "learning_rate": 7.676182547410658e-06,
      "loss": 0.0254,
      "step": 4343
    },
    {
      "epoch": 1.164923572003218,
      "grad_norm": 0.2999762143903309,
      "learning_rate": 7.674864523641321e-06,
      "loss": 0.0215,
      "step": 4344
    },
    {
      "epoch": 1.1651917404129795,
      "grad_norm": 0.31912144705543466,
      "learning_rate": 7.673546239427528e-06,
      "loss": 0.0282,
      "step": 4345
    },
    {
      "epoch": 1.1654599088227406,
      "grad_norm": 0.2816485283465777,
      "learning_rate": 7.672227694897635e-06,
      "loss": 0.0256,
      "step": 4346
    },
    {
      "epoch": 1.165728077232502,
      "grad_norm": 0.359348848690921,
      "learning_rate": 7.67090889018003e-06,
      "loss": 0.0329,
      "step": 4347
    },
    {
      "epoch": 1.1659962456422632,
      "grad_norm": 0.29257703652306855,
      "learning_rate": 7.669589825403114e-06,
      "loss": 0.0282,
      "step": 4348
    },
    {
      "epoch": 1.1662644140520246,
      "grad_norm": 0.3359729538724772,
      "learning_rate": 7.668270500695324e-06,
      "loss": 0.0295,
      "step": 4349
    },
    {
      "epoch": 1.166532582461786,
      "grad_norm": 0.40956587777007475,
      "learning_rate": 7.666950916185121e-06,
      "loss": 0.0342,
      "step": 4350
    },
    {
      "epoch": 1.1668007508715474,
      "grad_norm": 0.328078990890433,
      "learning_rate": 7.665631072000987e-06,
      "loss": 0.0216,
      "step": 4351
    },
    {
      "epoch": 1.1670689192813086,
      "grad_norm": 0.9930732297452027,
      "learning_rate": 7.664310968271434e-06,
      "loss": 0.0381,
      "step": 4352
    },
    {
      "epoch": 1.16733708769107,
      "grad_norm": 0.26823805239047616,
      "learning_rate": 7.662990605124995e-06,
      "loss": 0.0282,
      "step": 4353
    },
    {
      "epoch": 1.1676052561008312,
      "grad_norm": 0.2691268578480586,
      "learning_rate": 7.661669982690229e-06,
      "loss": 0.0208,
      "step": 4354
    },
    {
      "epoch": 1.1678734245105926,
      "grad_norm": 0.33584675795374064,
      "learning_rate": 7.660349101095726e-06,
      "loss": 0.0363,
      "step": 4355
    },
    {
      "epoch": 1.168141592920354,
      "grad_norm": 0.502758116647226,
      "learning_rate": 7.659027960470091e-06,
      "loss": 0.0249,
      "step": 4356
    },
    {
      "epoch": 1.1684097613301154,
      "grad_norm": 0.31744495919719334,
      "learning_rate": 7.657706560941966e-06,
      "loss": 0.0228,
      "step": 4357
    },
    {
      "epoch": 1.1686779297398766,
      "grad_norm": 0.2799809329660122,
      "learning_rate": 7.656384902640008e-06,
      "loss": 0.021,
      "step": 4358
    },
    {
      "epoch": 1.168946098149638,
      "grad_norm": 0.3288606356169136,
      "learning_rate": 7.655062985692905e-06,
      "loss": 0.0276,
      "step": 4359
    },
    {
      "epoch": 1.1692142665593992,
      "grad_norm": 0.36575651073610055,
      "learning_rate": 7.65374081022937e-06,
      "loss": 0.0342,
      "step": 4360
    },
    {
      "epoch": 1.1694824349691606,
      "grad_norm": 0.3518942901763917,
      "learning_rate": 7.652418376378135e-06,
      "loss": 0.0265,
      "step": 4361
    },
    {
      "epoch": 1.169750603378922,
      "grad_norm": 0.3897395421248616,
      "learning_rate": 7.651095684267965e-06,
      "loss": 0.0293,
      "step": 4362
    },
    {
      "epoch": 1.1700187717886834,
      "grad_norm": 0.3658907444140491,
      "learning_rate": 7.649772734027647e-06,
      "loss": 0.034,
      "step": 4363
    },
    {
      "epoch": 1.1702869401984446,
      "grad_norm": 0.27193956589434276,
      "learning_rate": 7.648449525785994e-06,
      "loss": 0.023,
      "step": 4364
    },
    {
      "epoch": 1.170555108608206,
      "grad_norm": 0.332709353640698,
      "learning_rate": 7.647126059671841e-06,
      "loss": 0.0235,
      "step": 4365
    },
    {
      "epoch": 1.1708232770179672,
      "grad_norm": 0.2478024448506725,
      "learning_rate": 7.645802335814051e-06,
      "loss": 0.0208,
      "step": 4366
    },
    {
      "epoch": 1.1710914454277286,
      "grad_norm": 0.2821467696834224,
      "learning_rate": 7.644478354341513e-06,
      "loss": 0.0233,
      "step": 4367
    },
    {
      "epoch": 1.17135961383749,
      "grad_norm": 0.44911085329501554,
      "learning_rate": 7.64315411538314e-06,
      "loss": 0.024,
      "step": 4368
    },
    {
      "epoch": 1.1716277822472514,
      "grad_norm": 0.25506199678025704,
      "learning_rate": 7.641829619067866e-06,
      "loss": 0.0252,
      "step": 4369
    },
    {
      "epoch": 1.1718959506570126,
      "grad_norm": 0.7530798808426693,
      "learning_rate": 7.640504865524657e-06,
      "loss": 0.0402,
      "step": 4370
    },
    {
      "epoch": 1.172164119066774,
      "grad_norm": 0.24143648447291616,
      "learning_rate": 7.639179854882499e-06,
      "loss": 0.0268,
      "step": 4371
    },
    {
      "epoch": 1.1724322874765352,
      "grad_norm": 0.3008868251377793,
      "learning_rate": 7.637854587270405e-06,
      "loss": 0.0239,
      "step": 4372
    },
    {
      "epoch": 1.1727004558862966,
      "grad_norm": 0.252131024309555,
      "learning_rate": 7.636529062817415e-06,
      "loss": 0.0238,
      "step": 4373
    },
    {
      "epoch": 1.172968624296058,
      "grad_norm": 0.25597351374811295,
      "learning_rate": 7.63520328165259e-06,
      "loss": 0.0256,
      "step": 4374
    },
    {
      "epoch": 1.1732367927058194,
      "grad_norm": 0.3256072693131043,
      "learning_rate": 7.633877243905016e-06,
      "loss": 0.0339,
      "step": 4375
    },
    {
      "epoch": 1.1735049611155806,
      "grad_norm": 0.30354561756683646,
      "learning_rate": 7.632550949703808e-06,
      "loss": 0.0291,
      "step": 4376
    },
    {
      "epoch": 1.173773129525342,
      "grad_norm": 0.374452967729698,
      "learning_rate": 7.631224399178103e-06,
      "loss": 0.028,
      "step": 4377
    },
    {
      "epoch": 1.1740412979351031,
      "grad_norm": 0.40644174553048573,
      "learning_rate": 7.629897592457066e-06,
      "loss": 0.0379,
      "step": 4378
    },
    {
      "epoch": 1.1743094663448645,
      "grad_norm": 0.26270674276326994,
      "learning_rate": 7.6285705296698805e-06,
      "loss": 0.0265,
      "step": 4379
    },
    {
      "epoch": 1.174577634754626,
      "grad_norm": 0.24152654671526466,
      "learning_rate": 7.6272432109457625e-06,
      "loss": 0.0214,
      "step": 4380
    },
    {
      "epoch": 1.1748458031643874,
      "grad_norm": 0.24944512630579888,
      "learning_rate": 7.625915636413949e-06,
      "loss": 0.0237,
      "step": 4381
    },
    {
      "epoch": 1.1751139715741485,
      "grad_norm": 0.3856902614856428,
      "learning_rate": 7.624587806203699e-06,
      "loss": 0.0345,
      "step": 4382
    },
    {
      "epoch": 1.17538213998391,
      "grad_norm": 0.3433893841318947,
      "learning_rate": 7.623259720444305e-06,
      "loss": 0.0357,
      "step": 4383
    },
    {
      "epoch": 1.1756503083936711,
      "grad_norm": 0.5225586607053644,
      "learning_rate": 7.621931379265075e-06,
      "loss": 0.0235,
      "step": 4384
    },
    {
      "epoch": 1.1759184768034325,
      "grad_norm": 0.2643009982410421,
      "learning_rate": 7.62060278279535e-06,
      "loss": 0.0238,
      "step": 4385
    },
    {
      "epoch": 1.176186645213194,
      "grad_norm": 0.4332736693601954,
      "learning_rate": 7.619273931164487e-06,
      "loss": 0.0388,
      "step": 4386
    },
    {
      "epoch": 1.1764548136229551,
      "grad_norm": 0.3554110575019689,
      "learning_rate": 7.617944824501875e-06,
      "loss": 0.032,
      "step": 4387
    },
    {
      "epoch": 1.1767229820327165,
      "grad_norm": 0.40561517274220243,
      "learning_rate": 7.616615462936927e-06,
      "loss": 0.0432,
      "step": 4388
    },
    {
      "epoch": 1.176991150442478,
      "grad_norm": 0.4039621670163445,
      "learning_rate": 7.615285846599079e-06,
      "loss": 0.0493,
      "step": 4389
    },
    {
      "epoch": 1.1772593188522391,
      "grad_norm": 0.35421278904572157,
      "learning_rate": 7.613955975617793e-06,
      "loss": 0.0224,
      "step": 4390
    },
    {
      "epoch": 1.1775274872620005,
      "grad_norm": 0.3701044926323328,
      "learning_rate": 7.612625850122551e-06,
      "loss": 0.0247,
      "step": 4391
    },
    {
      "epoch": 1.177795655671762,
      "grad_norm": 0.287184922853217,
      "learning_rate": 7.611295470242867e-06,
      "loss": 0.0262,
      "step": 4392
    },
    {
      "epoch": 1.178063824081523,
      "grad_norm": 0.222661566241732,
      "learning_rate": 7.609964836108275e-06,
      "loss": 0.0202,
      "step": 4393
    },
    {
      "epoch": 1.1783319924912845,
      "grad_norm": 0.2559458709126972,
      "learning_rate": 7.608633947848338e-06,
      "loss": 0.0251,
      "step": 4394
    },
    {
      "epoch": 1.178600160901046,
      "grad_norm": 0.37746487354964414,
      "learning_rate": 7.6073028055926375e-06,
      "loss": 0.0276,
      "step": 4395
    },
    {
      "epoch": 1.178868329310807,
      "grad_norm": 0.20469225204007865,
      "learning_rate": 7.605971409470787e-06,
      "loss": 0.0211,
      "step": 4396
    },
    {
      "epoch": 1.1791364977205685,
      "grad_norm": 0.3975510855032446,
      "learning_rate": 7.604639759612419e-06,
      "loss": 0.0392,
      "step": 4397
    },
    {
      "epoch": 1.17940466613033,
      "grad_norm": 0.25485730713009525,
      "learning_rate": 7.603307856147191e-06,
      "loss": 0.0196,
      "step": 4398
    },
    {
      "epoch": 1.179672834540091,
      "grad_norm": 0.2707617827444718,
      "learning_rate": 7.601975699204789e-06,
      "loss": 0.0228,
      "step": 4399
    },
    {
      "epoch": 1.1799410029498525,
      "grad_norm": 0.4794470341019494,
      "learning_rate": 7.600643288914922e-06,
      "loss": 0.0399,
      "step": 4400
    },
    {
      "epoch": 1.180209171359614,
      "grad_norm": 0.39914905167803033,
      "learning_rate": 7.599310625407323e-06,
      "loss": 0.0263,
      "step": 4401
    },
    {
      "epoch": 1.180477339769375,
      "grad_norm": 0.2633384193625362,
      "learning_rate": 7.59797770881175e-06,
      "loss": 0.018,
      "step": 4402
    },
    {
      "epoch": 1.1807455081791365,
      "grad_norm": 0.40450352183191624,
      "learning_rate": 7.596644539257983e-06,
      "loss": 0.0437,
      "step": 4403
    },
    {
      "epoch": 1.1810136765888979,
      "grad_norm": 0.3453283301426952,
      "learning_rate": 7.595311116875832e-06,
      "loss": 0.0253,
      "step": 4404
    },
    {
      "epoch": 1.181281844998659,
      "grad_norm": 0.2774001802931696,
      "learning_rate": 7.593977441795129e-06,
      "loss": 0.0269,
      "step": 4405
    },
    {
      "epoch": 1.1815500134084205,
      "grad_norm": 0.41555957218667217,
      "learning_rate": 7.592643514145728e-06,
      "loss": 0.0328,
      "step": 4406
    },
    {
      "epoch": 1.1818181818181819,
      "grad_norm": 0.295926535130485,
      "learning_rate": 7.591309334057511e-06,
      "loss": 0.0235,
      "step": 4407
    },
    {
      "epoch": 1.182086350227943,
      "grad_norm": 0.31182175251646144,
      "learning_rate": 7.5899749016603855e-06,
      "loss": 0.0291,
      "step": 4408
    },
    {
      "epoch": 1.1823545186377045,
      "grad_norm": 0.35616126381957625,
      "learning_rate": 7.588640217084279e-06,
      "loss": 0.0321,
      "step": 4409
    },
    {
      "epoch": 1.1826226870474659,
      "grad_norm": 0.18489561105003466,
      "learning_rate": 7.587305280459148e-06,
      "loss": 0.0188,
      "step": 4410
    },
    {
      "epoch": 1.182890855457227,
      "grad_norm": 0.3006398542060778,
      "learning_rate": 7.585970091914969e-06,
      "loss": 0.0326,
      "step": 4411
    },
    {
      "epoch": 1.1831590238669885,
      "grad_norm": 0.2562784750773555,
      "learning_rate": 7.5846346515817505e-06,
      "loss": 0.0246,
      "step": 4412
    },
    {
      "epoch": 1.1834271922767499,
      "grad_norm": 0.28981603742122397,
      "learning_rate": 7.583298959589517e-06,
      "loss": 0.0202,
      "step": 4413
    },
    {
      "epoch": 1.183695360686511,
      "grad_norm": 0.31245897061298017,
      "learning_rate": 7.581963016068322e-06,
      "loss": 0.0276,
      "step": 4414
    },
    {
      "epoch": 1.1839635290962724,
      "grad_norm": 0.5325014254939602,
      "learning_rate": 7.580626821148242e-06,
      "loss": 0.0364,
      "step": 4415
    },
    {
      "epoch": 1.1842316975060339,
      "grad_norm": 0.3605253847379521,
      "learning_rate": 7.579290374959383e-06,
      "loss": 0.0303,
      "step": 4416
    },
    {
      "epoch": 1.184499865915795,
      "grad_norm": 0.3178142240579125,
      "learning_rate": 7.577953677631866e-06,
      "loss": 0.0355,
      "step": 4417
    },
    {
      "epoch": 1.1847680343255564,
      "grad_norm": 0.30090543514016804,
      "learning_rate": 7.576616729295844e-06,
      "loss": 0.023,
      "step": 4418
    },
    {
      "epoch": 1.1850362027353178,
      "grad_norm": 0.33870155754602516,
      "learning_rate": 7.5752795300814915e-06,
      "loss": 0.0287,
      "step": 4419
    },
    {
      "epoch": 1.185304371145079,
      "grad_norm": 0.30032008142654054,
      "learning_rate": 7.573942080119009e-06,
      "loss": 0.0226,
      "step": 4420
    },
    {
      "epoch": 1.1855725395548404,
      "grad_norm": 0.28028849556239216,
      "learning_rate": 7.57260437953862e-06,
      "loss": 0.0215,
      "step": 4421
    },
    {
      "epoch": 1.1858407079646018,
      "grad_norm": 0.2904747056658521,
      "learning_rate": 7.571266428470571e-06,
      "loss": 0.0262,
      "step": 4422
    },
    {
      "epoch": 1.186108876374363,
      "grad_norm": 0.24052403319642113,
      "learning_rate": 7.569928227045138e-06,
      "loss": 0.0178,
      "step": 4423
    },
    {
      "epoch": 1.1863770447841244,
      "grad_norm": 0.272311129063299,
      "learning_rate": 7.568589775392616e-06,
      "loss": 0.0287,
      "step": 4424
    },
    {
      "epoch": 1.1866452131938858,
      "grad_norm": 0.23113936682361502,
      "learning_rate": 7.567251073643327e-06,
      "loss": 0.0221,
      "step": 4425
    },
    {
      "epoch": 1.186913381603647,
      "grad_norm": 0.4685161826481084,
      "learning_rate": 7.565912121927619e-06,
      "loss": 0.0304,
      "step": 4426
    },
    {
      "epoch": 1.1871815500134084,
      "grad_norm": 0.3322442878531938,
      "learning_rate": 7.564572920375857e-06,
      "loss": 0.0348,
      "step": 4427
    },
    {
      "epoch": 1.1874497184231698,
      "grad_norm": 0.37726724738671846,
      "learning_rate": 7.563233469118441e-06,
      "loss": 0.0322,
      "step": 4428
    },
    {
      "epoch": 1.187717886832931,
      "grad_norm": 0.45879219758321593,
      "learning_rate": 7.561893768285786e-06,
      "loss": 0.0276,
      "step": 4429
    },
    {
      "epoch": 1.1879860552426924,
      "grad_norm": 0.33563006918116495,
      "learning_rate": 7.560553818008336e-06,
      "loss": 0.0272,
      "step": 4430
    },
    {
      "epoch": 1.1882542236524538,
      "grad_norm": 0.35362311137597124,
      "learning_rate": 7.5592136184165586e-06,
      "loss": 0.0251,
      "step": 4431
    },
    {
      "epoch": 1.188522392062215,
      "grad_norm": 0.294115552078573,
      "learning_rate": 7.557873169640948e-06,
      "loss": 0.024,
      "step": 4432
    },
    {
      "epoch": 1.1887905604719764,
      "grad_norm": 0.26852106883543686,
      "learning_rate": 7.556532471812017e-06,
      "loss": 0.0187,
      "step": 4433
    },
    {
      "epoch": 1.1890587288817378,
      "grad_norm": 0.24276099576075158,
      "learning_rate": 7.555191525060306e-06,
      "loss": 0.0212,
      "step": 4434
    },
    {
      "epoch": 1.189326897291499,
      "grad_norm": 0.4012440513094016,
      "learning_rate": 7.553850329516379e-06,
      "loss": 0.0359,
      "step": 4435
    },
    {
      "epoch": 1.1895950657012604,
      "grad_norm": 0.36456691742290903,
      "learning_rate": 7.552508885310827e-06,
      "loss": 0.0272,
      "step": 4436
    },
    {
      "epoch": 1.1898632341110218,
      "grad_norm": 0.31289299973777823,
      "learning_rate": 7.551167192574262e-06,
      "loss": 0.0263,
      "step": 4437
    },
    {
      "epoch": 1.190131402520783,
      "grad_norm": 0.2828939854146892,
      "learning_rate": 7.549825251437322e-06,
      "loss": 0.0241,
      "step": 4438
    },
    {
      "epoch": 1.1903995709305444,
      "grad_norm": 0.27400089339526057,
      "learning_rate": 7.548483062030665e-06,
      "loss": 0.0221,
      "step": 4439
    },
    {
      "epoch": 1.1906677393403058,
      "grad_norm": 0.4142505118260145,
      "learning_rate": 7.547140624484979e-06,
      "loss": 0.0301,
      "step": 4440
    },
    {
      "epoch": 1.190935907750067,
      "grad_norm": 0.30637732035402343,
      "learning_rate": 7.545797938930971e-06,
      "loss": 0.0241,
      "step": 4441
    },
    {
      "epoch": 1.1912040761598284,
      "grad_norm": 0.27809901510567464,
      "learning_rate": 7.544455005499378e-06,
      "loss": 0.0214,
      "step": 4442
    },
    {
      "epoch": 1.1914722445695898,
      "grad_norm": 0.24565749813099186,
      "learning_rate": 7.543111824320956e-06,
      "loss": 0.0267,
      "step": 4443
    },
    {
      "epoch": 1.191740412979351,
      "grad_norm": 0.5846477988826597,
      "learning_rate": 7.54176839552649e-06,
      "loss": 0.0404,
      "step": 4444
    },
    {
      "epoch": 1.1920085813891124,
      "grad_norm": 0.27620285388453714,
      "learning_rate": 7.540424719246782e-06,
      "loss": 0.0251,
      "step": 4445
    },
    {
      "epoch": 1.1922767497988738,
      "grad_norm": 0.3664908823665164,
      "learning_rate": 7.5390807956126634e-06,
      "loss": 0.0291,
      "step": 4446
    },
    {
      "epoch": 1.192544918208635,
      "grad_norm": 0.40985888784764357,
      "learning_rate": 7.537736624754989e-06,
      "loss": 0.0422,
      "step": 4447
    },
    {
      "epoch": 1.1928130866183964,
      "grad_norm": 0.2398115965487242,
      "learning_rate": 7.536392206804637e-06,
      "loss": 0.0152,
      "step": 4448
    },
    {
      "epoch": 1.1930812550281578,
      "grad_norm": 0.44417929945724893,
      "learning_rate": 7.5350475418925096e-06,
      "loss": 0.0474,
      "step": 4449
    },
    {
      "epoch": 1.193349423437919,
      "grad_norm": 0.42945619165428184,
      "learning_rate": 7.533702630149535e-06,
      "loss": 0.0348,
      "step": 4450
    },
    {
      "epoch": 1.1936175918476803,
      "grad_norm": 0.28148322073536697,
      "learning_rate": 7.53235747170666e-06,
      "loss": 0.0193,
      "step": 4451
    },
    {
      "epoch": 1.1938857602574418,
      "grad_norm": 0.26205175546746634,
      "learning_rate": 7.5310120666948625e-06,
      "loss": 0.0206,
      "step": 4452
    },
    {
      "epoch": 1.194153928667203,
      "grad_norm": 0.33307252113023333,
      "learning_rate": 7.529666415245138e-06,
      "loss": 0.0328,
      "step": 4453
    },
    {
      "epoch": 1.1944220970769643,
      "grad_norm": 0.2596800382767961,
      "learning_rate": 7.528320517488513e-06,
      "loss": 0.0213,
      "step": 4454
    },
    {
      "epoch": 1.1946902654867257,
      "grad_norm": 0.3286835578712297,
      "learning_rate": 7.526974373556031e-06,
      "loss": 0.0272,
      "step": 4455
    },
    {
      "epoch": 1.194958433896487,
      "grad_norm": 0.24580953173462058,
      "learning_rate": 7.525627983578762e-06,
      "loss": 0.0215,
      "step": 4456
    },
    {
      "epoch": 1.1952266023062483,
      "grad_norm": 0.3415685635723789,
      "learning_rate": 7.5242813476878055e-06,
      "loss": 0.0345,
      "step": 4457
    },
    {
      "epoch": 1.1954947707160097,
      "grad_norm": 0.27019503769639147,
      "learning_rate": 7.522934466014272e-06,
      "loss": 0.0271,
      "step": 4458
    },
    {
      "epoch": 1.195762939125771,
      "grad_norm": 0.3123228896911459,
      "learning_rate": 7.521587338689309e-06,
      "loss": 0.0311,
      "step": 4459
    },
    {
      "epoch": 1.1960311075355323,
      "grad_norm": 0.3274366681518503,
      "learning_rate": 7.520239965844083e-06,
      "loss": 0.0356,
      "step": 4460
    },
    {
      "epoch": 1.1962992759452937,
      "grad_norm": 0.3695126843300847,
      "learning_rate": 7.518892347609781e-06,
      "loss": 0.0274,
      "step": 4461
    },
    {
      "epoch": 1.196567444355055,
      "grad_norm": 0.2725941007248382,
      "learning_rate": 7.517544484117618e-06,
      "loss": 0.0207,
      "step": 4462
    },
    {
      "epoch": 1.1968356127648163,
      "grad_norm": 0.2735553490169365,
      "learning_rate": 7.516196375498834e-06,
      "loss": 0.0252,
      "step": 4463
    },
    {
      "epoch": 1.1971037811745777,
      "grad_norm": 0.27285384194398116,
      "learning_rate": 7.514848021884689e-06,
      "loss": 0.0245,
      "step": 4464
    },
    {
      "epoch": 1.197371949584339,
      "grad_norm": 0.41638126080756105,
      "learning_rate": 7.513499423406469e-06,
      "loss": 0.034,
      "step": 4465
    },
    {
      "epoch": 1.1976401179941003,
      "grad_norm": 0.3445200035163466,
      "learning_rate": 7.512150580195483e-06,
      "loss": 0.0328,
      "step": 4466
    },
    {
      "epoch": 1.1979082864038617,
      "grad_norm": 0.4052549094238427,
      "learning_rate": 7.510801492383064e-06,
      "loss": 0.0267,
      "step": 4467
    },
    {
      "epoch": 1.198176454813623,
      "grad_norm": 0.22580605361727296,
      "learning_rate": 7.509452160100572e-06,
      "loss": 0.0248,
      "step": 4468
    },
    {
      "epoch": 1.1984446232233843,
      "grad_norm": 0.2814304603046185,
      "learning_rate": 7.508102583479384e-06,
      "loss": 0.0268,
      "step": 4469
    },
    {
      "epoch": 1.1987127916331457,
      "grad_norm": 0.22678302130361858,
      "learning_rate": 7.506752762650906e-06,
      "loss": 0.0182,
      "step": 4470
    },
    {
      "epoch": 1.1989809600429069,
      "grad_norm": 0.36248691912287667,
      "learning_rate": 7.5054026977465665e-06,
      "loss": 0.0252,
      "step": 4471
    },
    {
      "epoch": 1.1992491284526683,
      "grad_norm": 0.2996751589194444,
      "learning_rate": 7.50405238889782e-06,
      "loss": 0.0355,
      "step": 4472
    },
    {
      "epoch": 1.1995172968624297,
      "grad_norm": 0.3204499027331999,
      "learning_rate": 7.502701836236137e-06,
      "loss": 0.0344,
      "step": 4473
    },
    {
      "epoch": 1.1997854652721909,
      "grad_norm": 0.4295427047837426,
      "learning_rate": 7.501351039893026e-06,
      "loss": 0.0287,
      "step": 4474
    },
    {
      "epoch": 1.2000536336819523,
      "grad_norm": 0.2911949205971258,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.0303,
      "step": 4475
    },
    {
      "epoch": 1.2003218020917137,
      "grad_norm": 0.26688931046396713,
      "learning_rate": 7.498648716688615e-06,
      "loss": 0.0251,
      "step": 4476
    },
    {
      "epoch": 1.2005899705014749,
      "grad_norm": 0.2993977733387709,
      "learning_rate": 7.497297190090436e-06,
      "loss": 0.0271,
      "step": 4477
    },
    {
      "epoch": 1.2008581389112363,
      "grad_norm": 0.27422796698902213,
      "learning_rate": 7.49594542033706e-06,
      "loss": 0.0276,
      "step": 4478
    },
    {
      "epoch": 1.2011263073209977,
      "grad_norm": 0.26991343548144825,
      "learning_rate": 7.494593407560105e-06,
      "loss": 0.0252,
      "step": 4479
    },
    {
      "epoch": 1.2013944757307589,
      "grad_norm": 0.6588222432973042,
      "learning_rate": 7.4932411518912154e-06,
      "loss": 0.0275,
      "step": 4480
    },
    {
      "epoch": 1.2016626441405203,
      "grad_norm": 0.2982332264880145,
      "learning_rate": 7.491888653462053e-06,
      "loss": 0.0238,
      "step": 4481
    },
    {
      "epoch": 1.2019308125502817,
      "grad_norm": 0.2791956788274975,
      "learning_rate": 7.490535912404308e-06,
      "loss": 0.0181,
      "step": 4482
    },
    {
      "epoch": 1.2021989809600429,
      "grad_norm": 0.3863429670376822,
      "learning_rate": 7.489182928849692e-06,
      "loss": 0.0327,
      "step": 4483
    },
    {
      "epoch": 1.2024671493698043,
      "grad_norm": 0.43156336730434414,
      "learning_rate": 7.487829702929944e-06,
      "loss": 0.0216,
      "step": 4484
    },
    {
      "epoch": 1.2027353177795657,
      "grad_norm": 0.36499746409086237,
      "learning_rate": 7.486476234776823e-06,
      "loss": 0.031,
      "step": 4485
    },
    {
      "epoch": 1.2030034861893268,
      "grad_norm": 0.42075494464808405,
      "learning_rate": 7.4851225245221125e-06,
      "loss": 0.0238,
      "step": 4486
    },
    {
      "epoch": 1.2032716545990882,
      "grad_norm": 0.617076950599668,
      "learning_rate": 7.4837685722976204e-06,
      "loss": 0.0218,
      "step": 4487
    },
    {
      "epoch": 1.2035398230088497,
      "grad_norm": 0.345412137591531,
      "learning_rate": 7.482414378235175e-06,
      "loss": 0.0252,
      "step": 4488
    },
    {
      "epoch": 1.2038079914186108,
      "grad_norm": 0.2238204369740869,
      "learning_rate": 7.481059942466633e-06,
      "loss": 0.0217,
      "step": 4489
    },
    {
      "epoch": 1.2040761598283722,
      "grad_norm": 0.2510821633393369,
      "learning_rate": 7.4797052651238715e-06,
      "loss": 0.0184,
      "step": 4490
    },
    {
      "epoch": 1.2043443282381336,
      "grad_norm": 0.23434387586905642,
      "learning_rate": 7.4783503463387915e-06,
      "loss": 0.021,
      "step": 4491
    },
    {
      "epoch": 1.2046124966478948,
      "grad_norm": 0.2822910293440983,
      "learning_rate": 7.4769951862433185e-06,
      "loss": 0.0301,
      "step": 4492
    },
    {
      "epoch": 1.2048806650576562,
      "grad_norm": 0.46807449784524263,
      "learning_rate": 7.475639784969402e-06,
      "loss": 0.0484,
      "step": 4493
    },
    {
      "epoch": 1.2051488334674176,
      "grad_norm": 0.28437476842421794,
      "learning_rate": 7.474284142649011e-06,
      "loss": 0.0284,
      "step": 4494
    },
    {
      "epoch": 1.2054170018771788,
      "grad_norm": 0.4746364532318406,
      "learning_rate": 7.472928259414142e-06,
      "loss": 0.0253,
      "step": 4495
    },
    {
      "epoch": 1.2056851702869402,
      "grad_norm": 0.2781041061733415,
      "learning_rate": 7.471572135396814e-06,
      "loss": 0.018,
      "step": 4496
    },
    {
      "epoch": 1.2059533386967014,
      "grad_norm": 0.2564956521564478,
      "learning_rate": 7.470215770729069e-06,
      "loss": 0.0206,
      "step": 4497
    },
    {
      "epoch": 1.2062215071064628,
      "grad_norm": 0.4304370575022056,
      "learning_rate": 7.4688591655429724e-06,
      "loss": 0.0285,
      "step": 4498
    },
    {
      "epoch": 1.2064896755162242,
      "grad_norm": 0.2643710428644127,
      "learning_rate": 7.467502319970614e-06,
      "loss": 0.0239,
      "step": 4499
    },
    {
      "epoch": 1.2067578439259856,
      "grad_norm": 0.2964106533306818,
      "learning_rate": 7.466145234144106e-06,
      "loss": 0.0312,
      "step": 4500
    },
    {
      "epoch": 1.2070260123357468,
      "grad_norm": 0.42816573319818774,
      "learning_rate": 7.464787908195583e-06,
      "loss": 0.04,
      "step": 4501
    },
    {
      "epoch": 1.2072941807455082,
      "grad_norm": 0.37638317872667626,
      "learning_rate": 7.463430342257206e-06,
      "loss": 0.0191,
      "step": 4502
    },
    {
      "epoch": 1.2075623491552694,
      "grad_norm": 0.2968712282540018,
      "learning_rate": 7.462072536461158e-06,
      "loss": 0.0258,
      "step": 4503
    },
    {
      "epoch": 1.2078305175650308,
      "grad_norm": 0.3130575085906102,
      "learning_rate": 7.460714490939643e-06,
      "loss": 0.0236,
      "step": 4504
    },
    {
      "epoch": 1.2080986859747922,
      "grad_norm": 0.3744932419549456,
      "learning_rate": 7.45935620582489e-06,
      "loss": 0.0406,
      "step": 4505
    },
    {
      "epoch": 1.2083668543845536,
      "grad_norm": 0.24225816007294187,
      "learning_rate": 7.457997681249154e-06,
      "loss": 0.0265,
      "step": 4506
    },
    {
      "epoch": 1.2086350227943148,
      "grad_norm": 0.32717352290677904,
      "learning_rate": 7.456638917344709e-06,
      "loss": 0.0333,
      "step": 4507
    },
    {
      "epoch": 1.2089031912040762,
      "grad_norm": 0.2969234501248619,
      "learning_rate": 7.455279914243853e-06,
      "loss": 0.0291,
      "step": 4508
    },
    {
      "epoch": 1.2091713596138374,
      "grad_norm": 0.407933700823053,
      "learning_rate": 7.453920672078913e-06,
      "loss": 0.021,
      "step": 4509
    },
    {
      "epoch": 1.2094395280235988,
      "grad_norm": 0.3237137334269245,
      "learning_rate": 7.452561190982233e-06,
      "loss": 0.0266,
      "step": 4510
    },
    {
      "epoch": 1.2097076964333602,
      "grad_norm": 0.35074633031756647,
      "learning_rate": 7.451201471086179e-06,
      "loss": 0.0212,
      "step": 4511
    },
    {
      "epoch": 1.2099758648431216,
      "grad_norm": 0.2685436054914016,
      "learning_rate": 7.449841512523146e-06,
      "loss": 0.0205,
      "step": 4512
    },
    {
      "epoch": 1.2102440332528828,
      "grad_norm": 0.3897514979162601,
      "learning_rate": 7.448481315425549e-06,
      "loss": 0.0306,
      "step": 4513
    },
    {
      "epoch": 1.2105122016626442,
      "grad_norm": 0.24786361108061764,
      "learning_rate": 7.4471208799258285e-06,
      "loss": 0.0182,
      "step": 4514
    },
    {
      "epoch": 1.2107803700724054,
      "grad_norm": 0.3313627505795268,
      "learning_rate": 7.445760206156443e-06,
      "loss": 0.0266,
      "step": 4515
    },
    {
      "epoch": 1.2110485384821668,
      "grad_norm": 0.30752886224655723,
      "learning_rate": 7.44439929424988e-06,
      "loss": 0.0229,
      "step": 4516
    },
    {
      "epoch": 1.2113167068919282,
      "grad_norm": 0.3239327785399324,
      "learning_rate": 7.443038144338649e-06,
      "loss": 0.0335,
      "step": 4517
    },
    {
      "epoch": 1.2115848753016896,
      "grad_norm": 0.34242355856629536,
      "learning_rate": 7.4416767565552794e-06,
      "loss": 0.0258,
      "step": 4518
    },
    {
      "epoch": 1.2118530437114508,
      "grad_norm": 0.26026967410999,
      "learning_rate": 7.440315131032325e-06,
      "loss": 0.0257,
      "step": 4519
    },
    {
      "epoch": 1.2121212121212122,
      "grad_norm": 0.30302898912181625,
      "learning_rate": 7.4389532679023665e-06,
      "loss": 0.0256,
      "step": 4520
    },
    {
      "epoch": 1.2123893805309733,
      "grad_norm": 0.5469247536028212,
      "learning_rate": 7.437591167298003e-06,
      "loss": 0.0348,
      "step": 4521
    },
    {
      "epoch": 1.2126575489407347,
      "grad_norm": 0.3323778659087132,
      "learning_rate": 7.4362288293518595e-06,
      "loss": 0.0278,
      "step": 4522
    },
    {
      "epoch": 1.2129257173504961,
      "grad_norm": 0.24202459602360218,
      "learning_rate": 7.434866254196584e-06,
      "loss": 0.0205,
      "step": 4523
    },
    {
      "epoch": 1.2131938857602576,
      "grad_norm": 0.2909708979538222,
      "learning_rate": 7.433503441964844e-06,
      "loss": 0.0328,
      "step": 4524
    },
    {
      "epoch": 1.2134620541700187,
      "grad_norm": 0.29589874835613944,
      "learning_rate": 7.432140392789337e-06,
      "loss": 0.0266,
      "step": 4525
    },
    {
      "epoch": 1.2137302225797801,
      "grad_norm": 0.3191294950057547,
      "learning_rate": 7.430777106802774e-06,
      "loss": 0.0299,
      "step": 4526
    },
    {
      "epoch": 1.2139983909895413,
      "grad_norm": 0.25099079978991795,
      "learning_rate": 7.429413584137899e-06,
      "loss": 0.0215,
      "step": 4527
    },
    {
      "epoch": 1.2142665593993027,
      "grad_norm": 0.6211934731640977,
      "learning_rate": 7.428049824927475e-06,
      "loss": 0.0347,
      "step": 4528
    },
    {
      "epoch": 1.2145347278090641,
      "grad_norm": 0.3451770304916984,
      "learning_rate": 7.4266858293042844e-06,
      "loss": 0.034,
      "step": 4529
    },
    {
      "epoch": 1.2148028962188255,
      "grad_norm": 0.2620606434819824,
      "learning_rate": 7.425321597401137e-06,
      "loss": 0.0202,
      "step": 4530
    },
    {
      "epoch": 1.2150710646285867,
      "grad_norm": 0.2826778209621178,
      "learning_rate": 7.423957129350866e-06,
      "loss": 0.0221,
      "step": 4531
    },
    {
      "epoch": 1.2153392330383481,
      "grad_norm": 0.2700643306128579,
      "learning_rate": 7.422592425286323e-06,
      "loss": 0.0282,
      "step": 4532
    },
    {
      "epoch": 1.2156074014481093,
      "grad_norm": 0.3257281546326731,
      "learning_rate": 7.421227485340388e-06,
      "loss": 0.025,
      "step": 4533
    },
    {
      "epoch": 1.2158755698578707,
      "grad_norm": 0.40094345596617353,
      "learning_rate": 7.419862309645962e-06,
      "loss": 0.0363,
      "step": 4534
    },
    {
      "epoch": 1.2161437382676321,
      "grad_norm": 0.20590312002635133,
      "learning_rate": 7.418496898335967e-06,
      "loss": 0.0212,
      "step": 4535
    },
    {
      "epoch": 1.2164119066773935,
      "grad_norm": 0.27211492806021564,
      "learning_rate": 7.41713125154335e-06,
      "loss": 0.0323,
      "step": 4536
    },
    {
      "epoch": 1.2166800750871547,
      "grad_norm": 0.29660303903446456,
      "learning_rate": 7.41576536940108e-06,
      "loss": 0.0225,
      "step": 4537
    },
    {
      "epoch": 1.216948243496916,
      "grad_norm": 0.259929367511306,
      "learning_rate": 7.41439925204215e-06,
      "loss": 0.021,
      "step": 4538
    },
    {
      "epoch": 1.2172164119066773,
      "grad_norm": 0.16006148487536237,
      "learning_rate": 7.413032899599575e-06,
      "loss": 0.014,
      "step": 4539
    },
    {
      "epoch": 1.2174845803164387,
      "grad_norm": 0.3174993289379245,
      "learning_rate": 7.411666312206393e-06,
      "loss": 0.0254,
      "step": 4540
    },
    {
      "epoch": 1.2177527487262,
      "grad_norm": 0.35129032622558476,
      "learning_rate": 7.410299489995666e-06,
      "loss": 0.0232,
      "step": 4541
    },
    {
      "epoch": 1.2180209171359615,
      "grad_norm": 0.29660953603191165,
      "learning_rate": 7.408932433100476e-06,
      "loss": 0.0228,
      "step": 4542
    },
    {
      "epoch": 1.2182890855457227,
      "grad_norm": 0.28627306779046874,
      "learning_rate": 7.4075651416539294e-06,
      "loss": 0.0287,
      "step": 4543
    },
    {
      "epoch": 1.218557253955484,
      "grad_norm": 0.3630584496434604,
      "learning_rate": 7.4061976157891576e-06,
      "loss": 0.0324,
      "step": 4544
    },
    {
      "epoch": 1.2188254223652453,
      "grad_norm": 0.2644223970293549,
      "learning_rate": 7.404829855639313e-06,
      "loss": 0.016,
      "step": 4545
    },
    {
      "epoch": 1.2190935907750067,
      "grad_norm": 0.3018082301801645,
      "learning_rate": 7.40346186133757e-06,
      "loss": 0.0284,
      "step": 4546
    },
    {
      "epoch": 1.219361759184768,
      "grad_norm": 0.3992535574654829,
      "learning_rate": 7.402093633017127e-06,
      "loss": 0.041,
      "step": 4547
    },
    {
      "epoch": 1.2196299275945295,
      "grad_norm": 0.6832877437170843,
      "learning_rate": 7.400725170811202e-06,
      "loss": 0.0259,
      "step": 4548
    },
    {
      "epoch": 1.2198980960042907,
      "grad_norm": 0.3001789431762235,
      "learning_rate": 7.399356474853044e-06,
      "loss": 0.0215,
      "step": 4549
    },
    {
      "epoch": 1.220166264414052,
      "grad_norm": 0.37194912519700635,
      "learning_rate": 7.397987545275915e-06,
      "loss": 0.0347,
      "step": 4550
    },
    {
      "epoch": 1.2204344328238133,
      "grad_norm": 0.35476468429301733,
      "learning_rate": 7.3966183822131055e-06,
      "loss": 0.0297,
      "step": 4551
    },
    {
      "epoch": 1.2207026012335747,
      "grad_norm": 0.36480357319359435,
      "learning_rate": 7.395248985797926e-06,
      "loss": 0.0299,
      "step": 4552
    },
    {
      "epoch": 1.220970769643336,
      "grad_norm": 0.3462489979509227,
      "learning_rate": 7.393879356163713e-06,
      "loss": 0.0287,
      "step": 4553
    },
    {
      "epoch": 1.2212389380530975,
      "grad_norm": 0.3041588042929789,
      "learning_rate": 7.392509493443823e-06,
      "loss": 0.0218,
      "step": 4554
    },
    {
      "epoch": 1.2215071064628586,
      "grad_norm": 0.3626684397157777,
      "learning_rate": 7.391139397771634e-06,
      "loss": 0.0293,
      "step": 4555
    },
    {
      "epoch": 1.22177527487262,
      "grad_norm": 0.49884822193535594,
      "learning_rate": 7.389769069280551e-06,
      "loss": 0.0255,
      "step": 4556
    },
    {
      "epoch": 1.2220434432823812,
      "grad_norm": 0.5867673052748384,
      "learning_rate": 7.388398508103999e-06,
      "loss": 0.0359,
      "step": 4557
    },
    {
      "epoch": 1.2223116116921426,
      "grad_norm": 0.39207681714803233,
      "learning_rate": 7.3870277143754254e-06,
      "loss": 0.0294,
      "step": 4558
    },
    {
      "epoch": 1.222579780101904,
      "grad_norm": 0.23645364262797505,
      "learning_rate": 7.3856566882283e-06,
      "loss": 0.0179,
      "step": 4559
    },
    {
      "epoch": 1.2228479485116655,
      "grad_norm": 0.2672458401084975,
      "learning_rate": 7.384285429796116e-06,
      "loss": 0.028,
      "step": 4560
    },
    {
      "epoch": 1.2231161169214266,
      "grad_norm": 0.44498423871638726,
      "learning_rate": 7.382913939212392e-06,
      "loss": 0.0356,
      "step": 4561
    },
    {
      "epoch": 1.223384285331188,
      "grad_norm": 0.29408750385160326,
      "learning_rate": 7.3815422166106635e-06,
      "loss": 0.0252,
      "step": 4562
    },
    {
      "epoch": 1.2236524537409492,
      "grad_norm": 0.3337315396152775,
      "learning_rate": 7.380170262124491e-06,
      "loss": 0.0313,
      "step": 4563
    },
    {
      "epoch": 1.2239206221507106,
      "grad_norm": 0.4550946550495223,
      "learning_rate": 7.378798075887459e-06,
      "loss": 0.039,
      "step": 4564
    },
    {
      "epoch": 1.224188790560472,
      "grad_norm": 0.8652703006448639,
      "learning_rate": 7.377425658033177e-06,
      "loss": 0.0448,
      "step": 4565
    },
    {
      "epoch": 1.2244569589702332,
      "grad_norm": 0.33737324005927144,
      "learning_rate": 7.376053008695268e-06,
      "loss": 0.0197,
      "step": 4566
    },
    {
      "epoch": 1.2247251273799946,
      "grad_norm": 0.30680406156451767,
      "learning_rate": 7.374680128007387e-06,
      "loss": 0.0268,
      "step": 4567
    },
    {
      "epoch": 1.224993295789756,
      "grad_norm": 0.24723886685433172,
      "learning_rate": 7.373307016103206e-06,
      "loss": 0.0218,
      "step": 4568
    },
    {
      "epoch": 1.2252614641995172,
      "grad_norm": 0.3706334551808108,
      "learning_rate": 7.371933673116421e-06,
      "loss": 0.0202,
      "step": 4569
    },
    {
      "epoch": 1.2255296326092786,
      "grad_norm": 0.4038018117910843,
      "learning_rate": 7.370560099180752e-06,
      "loss": 0.0377,
      "step": 4570
    },
    {
      "epoch": 1.22579780101904,
      "grad_norm": 0.403778743997774,
      "learning_rate": 7.369186294429942e-06,
      "loss": 0.029,
      "step": 4571
    },
    {
      "epoch": 1.2260659694288012,
      "grad_norm": 0.2259859789392059,
      "learning_rate": 7.367812258997751e-06,
      "loss": 0.0219,
      "step": 4572
    },
    {
      "epoch": 1.2263341378385626,
      "grad_norm": 0.3607258061778454,
      "learning_rate": 7.366437993017966e-06,
      "loss": 0.0351,
      "step": 4573
    },
    {
      "epoch": 1.226602306248324,
      "grad_norm": 0.33638584466644306,
      "learning_rate": 7.365063496624396e-06,
      "loss": 0.0301,
      "step": 4574
    },
    {
      "epoch": 1.2268704746580852,
      "grad_norm": 0.35178627517505556,
      "learning_rate": 7.363688769950874e-06,
      "loss": 0.0281,
      "step": 4575
    },
    {
      "epoch": 1.2271386430678466,
      "grad_norm": 0.7159865188908692,
      "learning_rate": 7.362313813131252e-06,
      "loss": 0.0283,
      "step": 4576
    },
    {
      "epoch": 1.227406811477608,
      "grad_norm": 0.3495736027349909,
      "learning_rate": 7.360938626299406e-06,
      "loss": 0.0271,
      "step": 4577
    },
    {
      "epoch": 1.2276749798873692,
      "grad_norm": 0.3359739105792775,
      "learning_rate": 7.359563209589233e-06,
      "loss": 0.0279,
      "step": 4578
    },
    {
      "epoch": 1.2279431482971306,
      "grad_norm": 0.27424014791979473,
      "learning_rate": 7.358187563134655e-06,
      "loss": 0.0202,
      "step": 4579
    },
    {
      "epoch": 1.228211316706892,
      "grad_norm": 0.2873198249334603,
      "learning_rate": 7.356811687069615e-06,
      "loss": 0.0272,
      "step": 4580
    },
    {
      "epoch": 1.2284794851166532,
      "grad_norm": 0.43816873826968705,
      "learning_rate": 7.355435581528079e-06,
      "loss": 0.0377,
      "step": 4581
    },
    {
      "epoch": 1.2287476535264146,
      "grad_norm": 0.3038780411671199,
      "learning_rate": 7.354059246644033e-06,
      "loss": 0.0318,
      "step": 4582
    },
    {
      "epoch": 1.229015821936176,
      "grad_norm": 0.3368851360339834,
      "learning_rate": 7.35268268255149e-06,
      "loss": 0.0208,
      "step": 4583
    },
    {
      "epoch": 1.2292839903459372,
      "grad_norm": 0.29585244573474934,
      "learning_rate": 7.351305889384478e-06,
      "loss": 0.0299,
      "step": 4584
    },
    {
      "epoch": 1.2295521587556986,
      "grad_norm": 0.2547671520203688,
      "learning_rate": 7.349928867277054e-06,
      "loss": 0.0195,
      "step": 4585
    },
    {
      "epoch": 1.22982032716546,
      "grad_norm": 0.30120602150452863,
      "learning_rate": 7.348551616363297e-06,
      "loss": 0.0345,
      "step": 4586
    },
    {
      "epoch": 1.2300884955752212,
      "grad_norm": 0.37004673592601683,
      "learning_rate": 7.347174136777303e-06,
      "loss": 0.0334,
      "step": 4587
    },
    {
      "epoch": 1.2303566639849826,
      "grad_norm": 0.3331548158373684,
      "learning_rate": 7.345796428653196e-06,
      "loss": 0.0233,
      "step": 4588
    },
    {
      "epoch": 1.230624832394744,
      "grad_norm": 0.3475126666687277,
      "learning_rate": 7.3444184921251196e-06,
      "loss": 0.0356,
      "step": 4589
    },
    {
      "epoch": 1.2308930008045051,
      "grad_norm": 0.18770027717112514,
      "learning_rate": 7.343040327327235e-06,
      "loss": 0.0161,
      "step": 4590
    },
    {
      "epoch": 1.2311611692142665,
      "grad_norm": 0.33711089348844175,
      "learning_rate": 7.341661934393738e-06,
      "loss": 0.033,
      "step": 4591
    },
    {
      "epoch": 1.231429337624028,
      "grad_norm": 0.7302487037283778,
      "learning_rate": 7.340283313458833e-06,
      "loss": 0.0216,
      "step": 4592
    },
    {
      "epoch": 1.2316975060337891,
      "grad_norm": 0.30724595639708796,
      "learning_rate": 7.338904464656757e-06,
      "loss": 0.0284,
      "step": 4593
    },
    {
      "epoch": 1.2319656744435505,
      "grad_norm": 0.31163989899019334,
      "learning_rate": 7.337525388121763e-06,
      "loss": 0.0291,
      "step": 4594
    },
    {
      "epoch": 1.232233842853312,
      "grad_norm": 0.3763300405663219,
      "learning_rate": 7.336146083988127e-06,
      "loss": 0.0287,
      "step": 4595
    },
    {
      "epoch": 1.2325020112630731,
      "grad_norm": 0.2897800690428736,
      "learning_rate": 7.3347665523901504e-06,
      "loss": 0.0225,
      "step": 4596
    },
    {
      "epoch": 1.2327701796728345,
      "grad_norm": 0.3689517733787715,
      "learning_rate": 7.333386793462152e-06,
      "loss": 0.0255,
      "step": 4597
    },
    {
      "epoch": 1.233038348082596,
      "grad_norm": 0.2981821533118846,
      "learning_rate": 7.332006807338476e-06,
      "loss": 0.0315,
      "step": 4598
    },
    {
      "epoch": 1.2333065164923571,
      "grad_norm": 0.35341023508648384,
      "learning_rate": 7.33062659415349e-06,
      "loss": 0.0419,
      "step": 4599
    },
    {
      "epoch": 1.2335746849021185,
      "grad_norm": 1.3760849648356483,
      "learning_rate": 7.32924615404158e-06,
      "loss": 0.0384,
      "step": 4600
    },
    {
      "epoch": 1.23384285331188,
      "grad_norm": 0.37688422670872324,
      "learning_rate": 7.327865487137155e-06,
      "loss": 0.0296,
      "step": 4601
    },
    {
      "epoch": 1.234111021721641,
      "grad_norm": 0.32672402525907873,
      "learning_rate": 7.32648459357465e-06,
      "loss": 0.0252,
      "step": 4602
    },
    {
      "epoch": 1.2343791901314025,
      "grad_norm": 0.3413727808984585,
      "learning_rate": 7.325103473488515e-06,
      "loss": 0.0273,
      "step": 4603
    },
    {
      "epoch": 1.234647358541164,
      "grad_norm": 0.24895310750367244,
      "learning_rate": 7.323722127013228e-06,
      "loss": 0.0212,
      "step": 4604
    },
    {
      "epoch": 1.234915526950925,
      "grad_norm": 0.25689965838596335,
      "learning_rate": 7.322340554283287e-06,
      "loss": 0.0261,
      "step": 4605
    },
    {
      "epoch": 1.2351836953606865,
      "grad_norm": 0.28516955771968505,
      "learning_rate": 7.320958755433213e-06,
      "loss": 0.0257,
      "step": 4606
    },
    {
      "epoch": 1.235451863770448,
      "grad_norm": 0.6609218660796218,
      "learning_rate": 7.319576730597544e-06,
      "loss": 0.0344,
      "step": 4607
    },
    {
      "epoch": 1.235720032180209,
      "grad_norm": 0.3641844729896522,
      "learning_rate": 7.318194479910851e-06,
      "loss": 0.0253,
      "step": 4608
    },
    {
      "epoch": 1.2359882005899705,
      "grad_norm": 0.30597029063321235,
      "learning_rate": 7.316812003507714e-06,
      "loss": 0.024,
      "step": 4609
    },
    {
      "epoch": 1.236256368999732,
      "grad_norm": 0.4364275448858784,
      "learning_rate": 7.315429301522743e-06,
      "loss": 0.0256,
      "step": 4610
    },
    {
      "epoch": 1.236524537409493,
      "grad_norm": 0.2198695416951886,
      "learning_rate": 7.314046374090569e-06,
      "loss": 0.0206,
      "step": 4611
    },
    {
      "epoch": 1.2367927058192545,
      "grad_norm": 0.4380974927149838,
      "learning_rate": 7.3126632213458425e-06,
      "loss": 0.0293,
      "step": 4612
    },
    {
      "epoch": 1.237060874229016,
      "grad_norm": 0.43322391953863987,
      "learning_rate": 7.311279843423238e-06,
      "loss": 0.029,
      "step": 4613
    },
    {
      "epoch": 1.237329042638777,
      "grad_norm": 0.3036743520680998,
      "learning_rate": 7.309896240457454e-06,
      "loss": 0.0222,
      "step": 4614
    },
    {
      "epoch": 1.2375972110485385,
      "grad_norm": 0.26627289630811046,
      "learning_rate": 7.308512412583204e-06,
      "loss": 0.0235,
      "step": 4615
    },
    {
      "epoch": 1.2378653794582999,
      "grad_norm": 0.43643661559758146,
      "learning_rate": 7.307128359935229e-06,
      "loss": 0.0308,
      "step": 4616
    },
    {
      "epoch": 1.238133547868061,
      "grad_norm": 0.3114275083496713,
      "learning_rate": 7.305744082648289e-06,
      "loss": 0.0276,
      "step": 4617
    },
    {
      "epoch": 1.2384017162778225,
      "grad_norm": 0.5339547706820242,
      "learning_rate": 7.304359580857172e-06,
      "loss": 0.0376,
      "step": 4618
    },
    {
      "epoch": 1.2386698846875839,
      "grad_norm": 0.27663588284588975,
      "learning_rate": 7.30297485469668e-06,
      "loss": 0.0277,
      "step": 4619
    },
    {
      "epoch": 1.238938053097345,
      "grad_norm": 0.3652557962957018,
      "learning_rate": 7.301589904301642e-06,
      "loss": 0.03,
      "step": 4620
    },
    {
      "epoch": 1.2392062215071065,
      "grad_norm": 0.47877058991234983,
      "learning_rate": 7.3002047298069044e-06,
      "loss": 0.0395,
      "step": 4621
    },
    {
      "epoch": 1.2394743899168679,
      "grad_norm": 0.277284796560363,
      "learning_rate": 7.298819331347338e-06,
      "loss": 0.021,
      "step": 4622
    },
    {
      "epoch": 1.239742558326629,
      "grad_norm": 0.2523358160662548,
      "learning_rate": 7.297433709057837e-06,
      "loss": 0.0175,
      "step": 4623
    },
    {
      "epoch": 1.2400107267363905,
      "grad_norm": 0.22755489918793204,
      "learning_rate": 7.296047863073316e-06,
      "loss": 0.0184,
      "step": 4624
    },
    {
      "epoch": 1.2402788951461519,
      "grad_norm": 0.3388627071382217,
      "learning_rate": 7.294661793528711e-06,
      "loss": 0.0254,
      "step": 4625
    },
    {
      "epoch": 1.240547063555913,
      "grad_norm": 0.49282530997591506,
      "learning_rate": 7.2932755005589805e-06,
      "loss": 0.0261,
      "step": 4626
    },
    {
      "epoch": 1.2408152319656744,
      "grad_norm": 0.3107196778430525,
      "learning_rate": 7.2918889842991e-06,
      "loss": 0.025,
      "step": 4627
    },
    {
      "epoch": 1.2410834003754359,
      "grad_norm": 0.22908325456793002,
      "learning_rate": 7.2905022448840745e-06,
      "loss": 0.014,
      "step": 4628
    },
    {
      "epoch": 1.241351568785197,
      "grad_norm": 0.23730692454528507,
      "learning_rate": 7.289115282448929e-06,
      "loss": 0.0193,
      "step": 4629
    },
    {
      "epoch": 1.2416197371949584,
      "grad_norm": 0.3477707378928443,
      "learning_rate": 7.2877280971287036e-06,
      "loss": 0.0347,
      "step": 4630
    },
    {
      "epoch": 1.2418879056047198,
      "grad_norm": 0.2745608744036304,
      "learning_rate": 7.286340689058468e-06,
      "loss": 0.0167,
      "step": 4631
    },
    {
      "epoch": 1.242156074014481,
      "grad_norm": 0.28831988965138905,
      "learning_rate": 7.28495305837331e-06,
      "loss": 0.0264,
      "step": 4632
    },
    {
      "epoch": 1.2424242424242424,
      "grad_norm": 0.32660945354746357,
      "learning_rate": 7.283565205208339e-06,
      "loss": 0.0327,
      "step": 4633
    },
    {
      "epoch": 1.2426924108340038,
      "grad_norm": 0.32502254225438143,
      "learning_rate": 7.282177129698685e-06,
      "loss": 0.0297,
      "step": 4634
    },
    {
      "epoch": 1.242960579243765,
      "grad_norm": 0.33248051898369596,
      "learning_rate": 7.280788831979504e-06,
      "loss": 0.0371,
      "step": 4635
    },
    {
      "epoch": 1.2432287476535264,
      "grad_norm": 0.37914241985462177,
      "learning_rate": 7.279400312185972e-06,
      "loss": 0.0316,
      "step": 4636
    },
    {
      "epoch": 1.2434969160632878,
      "grad_norm": 0.2701833421937976,
      "learning_rate": 7.27801157045328e-06,
      "loss": 0.025,
      "step": 4637
    },
    {
      "epoch": 1.243765084473049,
      "grad_norm": 1.249076515219052,
      "learning_rate": 7.276622606916651e-06,
      "loss": 0.0298,
      "step": 4638
    },
    {
      "epoch": 1.2440332528828104,
      "grad_norm": 0.3235058584659483,
      "learning_rate": 7.275233421711323e-06,
      "loss": 0.0232,
      "step": 4639
    },
    {
      "epoch": 1.2443014212925718,
      "grad_norm": 0.3999710440453591,
      "learning_rate": 7.273844014972558e-06,
      "loss": 0.041,
      "step": 4640
    },
    {
      "epoch": 1.244569589702333,
      "grad_norm": 0.3393825397974193,
      "learning_rate": 7.272454386835637e-06,
      "loss": 0.026,
      "step": 4641
    },
    {
      "epoch": 1.2448377581120944,
      "grad_norm": 0.21443882538740036,
      "learning_rate": 7.2710645374358655e-06,
      "loss": 0.0177,
      "step": 4642
    },
    {
      "epoch": 1.2451059265218558,
      "grad_norm": 0.30189557592707156,
      "learning_rate": 7.269674466908569e-06,
      "loss": 0.0329,
      "step": 4643
    },
    {
      "epoch": 1.245374094931617,
      "grad_norm": 0.32027811783203153,
      "learning_rate": 7.2682841753890975e-06,
      "loss": 0.0261,
      "step": 4644
    },
    {
      "epoch": 1.2456422633413784,
      "grad_norm": 0.4002705779720046,
      "learning_rate": 7.2668936630128175e-06,
      "loss": 0.0255,
      "step": 4645
    },
    {
      "epoch": 1.2459104317511398,
      "grad_norm": 0.2987993236283025,
      "learning_rate": 7.265502929915119e-06,
      "loss": 0.0303,
      "step": 4646
    },
    {
      "epoch": 1.246178600160901,
      "grad_norm": 0.5364409399423652,
      "learning_rate": 7.264111976231416e-06,
      "loss": 0.0375,
      "step": 4647
    },
    {
      "epoch": 1.2464467685706624,
      "grad_norm": 0.31461749483185736,
      "learning_rate": 7.262720802097141e-06,
      "loss": 0.0238,
      "step": 4648
    },
    {
      "epoch": 1.2467149369804238,
      "grad_norm": 0.4286401391247154,
      "learning_rate": 7.261329407647748e-06,
      "loss": 0.0354,
      "step": 4649
    },
    {
      "epoch": 1.246983105390185,
      "grad_norm": 0.2384308556558068,
      "learning_rate": 7.259937793018715e-06,
      "loss": 0.019,
      "step": 4650
    },
    {
      "epoch": 1.2472512737999464,
      "grad_norm": 0.41438447323786604,
      "learning_rate": 7.2585459583455396e-06,
      "loss": 0.0403,
      "step": 4651
    },
    {
      "epoch": 1.2475194422097078,
      "grad_norm": 0.22818140660695477,
      "learning_rate": 7.25715390376374e-06,
      "loss": 0.0194,
      "step": 4652
    },
    {
      "epoch": 1.247787610619469,
      "grad_norm": 0.3729362434083917,
      "learning_rate": 7.255761629408857e-06,
      "loss": 0.0297,
      "step": 4653
    },
    {
      "epoch": 1.2480557790292304,
      "grad_norm": 0.32139561431725594,
      "learning_rate": 7.254369135416454e-06,
      "loss": 0.0309,
      "step": 4654
    },
    {
      "epoch": 1.2483239474389918,
      "grad_norm": 0.4147638591567828,
      "learning_rate": 7.252976421922111e-06,
      "loss": 0.0291,
      "step": 4655
    },
    {
      "epoch": 1.248592115848753,
      "grad_norm": 0.2829048146044312,
      "learning_rate": 7.251583489061438e-06,
      "loss": 0.0197,
      "step": 4656
    },
    {
      "epoch": 1.2488602842585144,
      "grad_norm": 0.33685684552610673,
      "learning_rate": 7.250190336970058e-06,
      "loss": 0.0241,
      "step": 4657
    },
    {
      "epoch": 1.2491284526682758,
      "grad_norm": 0.35808295937540635,
      "learning_rate": 7.248796965783619e-06,
      "loss": 0.0266,
      "step": 4658
    },
    {
      "epoch": 1.249396621078037,
      "grad_norm": 0.4008921060217825,
      "learning_rate": 7.247403375637789e-06,
      "loss": 0.0309,
      "step": 4659
    },
    {
      "epoch": 1.2496647894877984,
      "grad_norm": 0.27462655259806784,
      "learning_rate": 7.24600956666826e-06,
      "loss": 0.0255,
      "step": 4660
    },
    {
      "epoch": 1.2499329578975598,
      "grad_norm": 0.3264078120027483,
      "learning_rate": 7.244615539010742e-06,
      "loss": 0.0241,
      "step": 4661
    },
    {
      "epoch": 1.250201126307321,
      "grad_norm": 0.36509909705706384,
      "learning_rate": 7.24322129280097e-06,
      "loss": 0.0364,
      "step": 4662
    },
    {
      "epoch": 1.2504692947170823,
      "grad_norm": 0.4725937058839055,
      "learning_rate": 7.241826828174694e-06,
      "loss": 0.0344,
      "step": 4663
    },
    {
      "epoch": 1.2507374631268435,
      "grad_norm": 0.38102819470968535,
      "learning_rate": 7.240432145267693e-06,
      "loss": 0.0248,
      "step": 4664
    },
    {
      "epoch": 1.251005631536605,
      "grad_norm": 0.2560207691158638,
      "learning_rate": 7.239037244215761e-06,
      "loss": 0.0292,
      "step": 4665
    },
    {
      "epoch": 1.2512737999463663,
      "grad_norm": 0.2981043092727063,
      "learning_rate": 7.237642125154718e-06,
      "loss": 0.0263,
      "step": 4666
    },
    {
      "epoch": 1.2515419683561277,
      "grad_norm": 0.3361091275044608,
      "learning_rate": 7.2362467882204015e-06,
      "loss": 0.0296,
      "step": 4667
    },
    {
      "epoch": 1.251810136765889,
      "grad_norm": 0.24405378773863384,
      "learning_rate": 7.2348512335486745e-06,
      "loss": 0.0236,
      "step": 4668
    },
    {
      "epoch": 1.2520783051756503,
      "grad_norm": 0.6029162396416538,
      "learning_rate": 7.233455461275414e-06,
      "loss": 0.0279,
      "step": 4669
    },
    {
      "epoch": 1.2523464735854115,
      "grad_norm": 0.3099823442534736,
      "learning_rate": 7.232059471536524e-06,
      "loss": 0.0317,
      "step": 4670
    },
    {
      "epoch": 1.252614641995173,
      "grad_norm": 0.33266854661579565,
      "learning_rate": 7.230663264467932e-06,
      "loss": 0.0165,
      "step": 4671
    },
    {
      "epoch": 1.2528828104049343,
      "grad_norm": 0.5203541189118916,
      "learning_rate": 7.229266840205579e-06,
      "loss": 0.0386,
      "step": 4672
    },
    {
      "epoch": 1.2531509788146957,
      "grad_norm": 0.5928478618930845,
      "learning_rate": 7.227870198885432e-06,
      "loss": 0.0262,
      "step": 4673
    },
    {
      "epoch": 1.253419147224457,
      "grad_norm": 0.2593291101208747,
      "learning_rate": 7.22647334064348e-06,
      "loss": 0.0211,
      "step": 4674
    },
    {
      "epoch": 1.2536873156342183,
      "grad_norm": 0.2702138730288383,
      "learning_rate": 7.225076265615729e-06,
      "loss": 0.0239,
      "step": 4675
    },
    {
      "epoch": 1.2539554840439795,
      "grad_norm": 0.4475458546222948,
      "learning_rate": 7.223678973938208e-06,
      "loss": 0.0311,
      "step": 4676
    },
    {
      "epoch": 1.254223652453741,
      "grad_norm": 0.32122534444233874,
      "learning_rate": 7.222281465746972e-06,
      "loss": 0.0296,
      "step": 4677
    },
    {
      "epoch": 1.2544918208635023,
      "grad_norm": 0.3399582262355133,
      "learning_rate": 7.220883741178088e-06,
      "loss": 0.0317,
      "step": 4678
    },
    {
      "epoch": 1.2547599892732637,
      "grad_norm": 0.23935242514832128,
      "learning_rate": 7.219485800367651e-06,
      "loss": 0.0185,
      "step": 4679
    },
    {
      "epoch": 1.255028157683025,
      "grad_norm": 0.36079681100413075,
      "learning_rate": 7.218087643451774e-06,
      "loss": 0.0388,
      "step": 4680
    },
    {
      "epoch": 1.2552963260927863,
      "grad_norm": 0.32140297862881134,
      "learning_rate": 7.2166892705665915e-06,
      "loss": 0.0268,
      "step": 4681
    },
    {
      "epoch": 1.2555644945025475,
      "grad_norm": 0.2154650001459716,
      "learning_rate": 7.21529068184826e-06,
      "loss": 0.0216,
      "step": 4682
    },
    {
      "epoch": 1.2558326629123089,
      "grad_norm": 0.28701630817367546,
      "learning_rate": 7.213891877432957e-06,
      "loss": 0.0276,
      "step": 4683
    },
    {
      "epoch": 1.2561008313220703,
      "grad_norm": 0.43070471072192734,
      "learning_rate": 7.21249285745688e-06,
      "loss": 0.0249,
      "step": 4684
    },
    {
      "epoch": 1.2563689997318317,
      "grad_norm": 1.1330304007504517,
      "learning_rate": 7.211093622056247e-06,
      "loss": 0.0281,
      "step": 4685
    },
    {
      "epoch": 1.2566371681415929,
      "grad_norm": 0.42332563304315374,
      "learning_rate": 7.209694171367299e-06,
      "loss": 0.0256,
      "step": 4686
    },
    {
      "epoch": 1.2569053365513543,
      "grad_norm": 0.30346302512607193,
      "learning_rate": 7.2082945055262946e-06,
      "loss": 0.024,
      "step": 4687
    },
    {
      "epoch": 1.2571735049611155,
      "grad_norm": 0.283264992493608,
      "learning_rate": 7.206894624669518e-06,
      "loss": 0.0302,
      "step": 4688
    },
    {
      "epoch": 1.2574416733708769,
      "grad_norm": 0.26438175654965845,
      "learning_rate": 7.205494528933271e-06,
      "loss": 0.0244,
      "step": 4689
    },
    {
      "epoch": 1.2577098417806383,
      "grad_norm": 0.3113243744283048,
      "learning_rate": 7.204094218453878e-06,
      "loss": 0.0256,
      "step": 4690
    },
    {
      "epoch": 1.2579780101903997,
      "grad_norm": 0.2773209875687245,
      "learning_rate": 7.202693693367681e-06,
      "loss": 0.0213,
      "step": 4691
    },
    {
      "epoch": 1.2582461786001609,
      "grad_norm": 0.3050636885955443,
      "learning_rate": 7.201292953811049e-06,
      "loss": 0.0239,
      "step": 4692
    },
    {
      "epoch": 1.2585143470099223,
      "grad_norm": 0.40974868699938893,
      "learning_rate": 7.1998919999203654e-06,
      "loss": 0.0272,
      "step": 4693
    },
    {
      "epoch": 1.2587825154196834,
      "grad_norm": 0.2866837714009931,
      "learning_rate": 7.198490831832038e-06,
      "loss": 0.0222,
      "step": 4694
    },
    {
      "epoch": 1.2590506838294448,
      "grad_norm": 0.2720715218580527,
      "learning_rate": 7.197089449682495e-06,
      "loss": 0.0228,
      "step": 4695
    },
    {
      "epoch": 1.2593188522392063,
      "grad_norm": 0.2585326345434907,
      "learning_rate": 7.195687853608186e-06,
      "loss": 0.0221,
      "step": 4696
    },
    {
      "epoch": 1.2595870206489677,
      "grad_norm": 0.17888864913340874,
      "learning_rate": 7.194286043745581e-06,
      "loss": 0.0176,
      "step": 4697
    },
    {
      "epoch": 1.2598551890587288,
      "grad_norm": 0.24775708536258945,
      "learning_rate": 7.192884020231168e-06,
      "loss": 0.0234,
      "step": 4698
    },
    {
      "epoch": 1.2601233574684902,
      "grad_norm": 0.38607211094775334,
      "learning_rate": 7.191481783201462e-06,
      "loss": 0.0401,
      "step": 4699
    },
    {
      "epoch": 1.2603915258782514,
      "grad_norm": 0.36877156749304657,
      "learning_rate": 7.1900793327929905e-06,
      "loss": 0.0224,
      "step": 4700
    },
    {
      "epoch": 1.2606596942880128,
      "grad_norm": 0.28847864979838256,
      "learning_rate": 7.18867666914231e-06,
      "loss": 0.0304,
      "step": 4701
    },
    {
      "epoch": 1.2609278626977742,
      "grad_norm": 0.23471396173447606,
      "learning_rate": 7.187273792385993e-06,
      "loss": 0.0205,
      "step": 4702
    },
    {
      "epoch": 1.2611960311075356,
      "grad_norm": 0.3763317305198127,
      "learning_rate": 7.185870702660635e-06,
      "loss": 0.032,
      "step": 4703
    },
    {
      "epoch": 1.2614641995172968,
      "grad_norm": 0.3355533906189248,
      "learning_rate": 7.184467400102849e-06,
      "loss": 0.0234,
      "step": 4704
    },
    {
      "epoch": 1.2617323679270582,
      "grad_norm": 0.3416500229416669,
      "learning_rate": 7.183063884849273e-06,
      "loss": 0.0362,
      "step": 4705
    },
    {
      "epoch": 1.2620005363368194,
      "grad_norm": 0.42009334967075956,
      "learning_rate": 7.181660157036561e-06,
      "loss": 0.0433,
      "step": 4706
    },
    {
      "epoch": 1.2622687047465808,
      "grad_norm": 0.5272575632300055,
      "learning_rate": 7.180256216801392e-06,
      "loss": 0.0198,
      "step": 4707
    },
    {
      "epoch": 1.2625368731563422,
      "grad_norm": 0.3038540837713627,
      "learning_rate": 7.178852064280463e-06,
      "loss": 0.0289,
      "step": 4708
    },
    {
      "epoch": 1.2628050415661036,
      "grad_norm": 0.5595897114225376,
      "learning_rate": 7.177447699610494e-06,
      "loss": 0.0354,
      "step": 4709
    },
    {
      "epoch": 1.2630732099758648,
      "grad_norm": 0.27253750989891773,
      "learning_rate": 7.176043122928226e-06,
      "loss": 0.0313,
      "step": 4710
    },
    {
      "epoch": 1.2633413783856262,
      "grad_norm": 0.2449812345445171,
      "learning_rate": 7.174638334370414e-06,
      "loss": 0.0202,
      "step": 4711
    },
    {
      "epoch": 1.2636095467953874,
      "grad_norm": 0.24259820431462986,
      "learning_rate": 7.17323333407384e-06,
      "loss": 0.0183,
      "step": 4712
    },
    {
      "epoch": 1.2638777152051488,
      "grad_norm": 0.28687155189711966,
      "learning_rate": 7.171828122175306e-06,
      "loss": 0.0285,
      "step": 4713
    },
    {
      "epoch": 1.2641458836149102,
      "grad_norm": 0.5550916667869823,
      "learning_rate": 7.170422698811635e-06,
      "loss": 0.0342,
      "step": 4714
    },
    {
      "epoch": 1.2644140520246716,
      "grad_norm": 0.3129796676803681,
      "learning_rate": 7.169017064119669e-06,
      "loss": 0.0272,
      "step": 4715
    },
    {
      "epoch": 1.2646822204344328,
      "grad_norm": 0.23766870449714858,
      "learning_rate": 7.167611218236269e-06,
      "loss": 0.0215,
      "step": 4716
    },
    {
      "epoch": 1.2649503888441942,
      "grad_norm": 0.3050900711430389,
      "learning_rate": 7.1662051612983204e-06,
      "loss": 0.0335,
      "step": 4717
    },
    {
      "epoch": 1.2652185572539554,
      "grad_norm": 0.3561590115918739,
      "learning_rate": 7.164798893442725e-06,
      "loss": 0.03,
      "step": 4718
    },
    {
      "epoch": 1.2654867256637168,
      "grad_norm": 0.2635287448350578,
      "learning_rate": 7.163392414806409e-06,
      "loss": 0.0244,
      "step": 4719
    },
    {
      "epoch": 1.2657548940734782,
      "grad_norm": 0.21757609324070415,
      "learning_rate": 7.161985725526318e-06,
      "loss": 0.0177,
      "step": 4720
    },
    {
      "epoch": 1.2660230624832396,
      "grad_norm": 0.4295325132318656,
      "learning_rate": 7.160578825739417e-06,
      "loss": 0.0424,
      "step": 4721
    },
    {
      "epoch": 1.2662912308930008,
      "grad_norm": 0.3442957616473958,
      "learning_rate": 7.159171715582691e-06,
      "loss": 0.0247,
      "step": 4722
    },
    {
      "epoch": 1.2665593993027622,
      "grad_norm": 0.4787361140962861,
      "learning_rate": 7.15776439519315e-06,
      "loss": 0.0337,
      "step": 4723
    },
    {
      "epoch": 1.2668275677125234,
      "grad_norm": 0.2826131328346021,
      "learning_rate": 7.156356864707816e-06,
      "loss": 0.0316,
      "step": 4724
    },
    {
      "epoch": 1.2670957361222848,
      "grad_norm": 0.24194820661125024,
      "learning_rate": 7.15494912426374e-06,
      "loss": 0.0234,
      "step": 4725
    },
    {
      "epoch": 1.2673639045320462,
      "grad_norm": 0.2504561737977258,
      "learning_rate": 7.153541173997989e-06,
      "loss": 0.0215,
      "step": 4726
    },
    {
      "epoch": 1.2676320729418076,
      "grad_norm": 0.2624578576635326,
      "learning_rate": 7.152133014047651e-06,
      "loss": 0.0277,
      "step": 4727
    },
    {
      "epoch": 1.2679002413515688,
      "grad_norm": 0.20817334817586813,
      "learning_rate": 7.150724644549835e-06,
      "loss": 0.023,
      "step": 4728
    },
    {
      "epoch": 1.2681684097613302,
      "grad_norm": 0.27974233361435313,
      "learning_rate": 7.149316065641672e-06,
      "loss": 0.0196,
      "step": 4729
    },
    {
      "epoch": 1.2684365781710913,
      "grad_norm": 0.2452883345373778,
      "learning_rate": 7.147907277460309e-06,
      "loss": 0.0229,
      "step": 4730
    },
    {
      "epoch": 1.2687047465808527,
      "grad_norm": 0.22663814677122504,
      "learning_rate": 7.146498280142917e-06,
      "loss": 0.0216,
      "step": 4731
    },
    {
      "epoch": 1.2689729149906142,
      "grad_norm": 0.31882531265751635,
      "learning_rate": 7.1450890738266864e-06,
      "loss": 0.0246,
      "step": 4732
    },
    {
      "epoch": 1.2692410834003756,
      "grad_norm": 0.305931639343806,
      "learning_rate": 7.143679658648828e-06,
      "loss": 0.028,
      "step": 4733
    },
    {
      "epoch": 1.2695092518101367,
      "grad_norm": 0.22779022938319887,
      "learning_rate": 7.142270034746572e-06,
      "loss": 0.0228,
      "step": 4734
    },
    {
      "epoch": 1.2697774202198981,
      "grad_norm": 0.3377587138873126,
      "learning_rate": 7.14086020225717e-06,
      "loss": 0.0439,
      "step": 4735
    },
    {
      "epoch": 1.2700455886296593,
      "grad_norm": 0.3092986307040245,
      "learning_rate": 7.139450161317895e-06,
      "loss": 0.0215,
      "step": 4736
    },
    {
      "epoch": 1.2703137570394207,
      "grad_norm": 0.34411196779703523,
      "learning_rate": 7.138039912066037e-06,
      "loss": 0.0295,
      "step": 4737
    },
    {
      "epoch": 1.2705819254491821,
      "grad_norm": 0.2608856831695081,
      "learning_rate": 7.13662945463891e-06,
      "loss": 0.0214,
      "step": 4738
    },
    {
      "epoch": 1.2708500938589435,
      "grad_norm": 0.25474214353535524,
      "learning_rate": 7.135218789173845e-06,
      "loss": 0.0201,
      "step": 4739
    },
    {
      "epoch": 1.2711182622687047,
      "grad_norm": 0.3418477350425559,
      "learning_rate": 7.133807915808197e-06,
      "loss": 0.0271,
      "step": 4740
    },
    {
      "epoch": 1.2713864306784661,
      "grad_norm": 0.39139052752025294,
      "learning_rate": 7.132396834679336e-06,
      "loss": 0.0273,
      "step": 4741
    },
    {
      "epoch": 1.2716545990882273,
      "grad_norm": 0.27457020042631014,
      "learning_rate": 7.1309855459246554e-06,
      "loss": 0.0203,
      "step": 4742
    },
    {
      "epoch": 1.2719227674979887,
      "grad_norm": 0.3814701703593124,
      "learning_rate": 7.1295740496815715e-06,
      "loss": 0.0259,
      "step": 4743
    },
    {
      "epoch": 1.2721909359077501,
      "grad_norm": 0.30643678065442015,
      "learning_rate": 7.128162346087517e-06,
      "loss": 0.027,
      "step": 4744
    },
    {
      "epoch": 1.2724591043175115,
      "grad_norm": 0.333129129835628,
      "learning_rate": 7.1267504352799435e-06,
      "loss": 0.0261,
      "step": 4745
    },
    {
      "epoch": 1.2727272727272727,
      "grad_norm": 0.5725631708459238,
      "learning_rate": 7.125338317396326e-06,
      "loss": 0.0379,
      "step": 4746
    },
    {
      "epoch": 1.2729954411370341,
      "grad_norm": 0.4652776249926086,
      "learning_rate": 7.123925992574161e-06,
      "loss": 0.0182,
      "step": 4747
    },
    {
      "epoch": 1.2732636095467953,
      "grad_norm": 0.24973758622134376,
      "learning_rate": 7.122513460950961e-06,
      "loss": 0.0245,
      "step": 4748
    },
    {
      "epoch": 1.2735317779565567,
      "grad_norm": 0.2786046932153775,
      "learning_rate": 7.1211007226642595e-06,
      "loss": 0.0274,
      "step": 4749
    },
    {
      "epoch": 1.273799946366318,
      "grad_norm": 0.6252322990247723,
      "learning_rate": 7.119687777851612e-06,
      "loss": 0.0307,
      "step": 4750
    },
    {
      "epoch": 1.2740681147760795,
      "grad_norm": 0.34541225053432,
      "learning_rate": 7.118274626650592e-06,
      "loss": 0.0298,
      "step": 4751
    },
    {
      "epoch": 1.2743362831858407,
      "grad_norm": 0.24688049491691627,
      "learning_rate": 7.116861269198797e-06,
      "loss": 0.0258,
      "step": 4752
    },
    {
      "epoch": 1.274604451595602,
      "grad_norm": 0.36296936998738577,
      "learning_rate": 7.115447705633843e-06,
      "loss": 0.032,
      "step": 4753
    },
    {
      "epoch": 1.2748726200053633,
      "grad_norm": 0.29465779582373297,
      "learning_rate": 7.1140339360933585e-06,
      "loss": 0.0268,
      "step": 4754
    },
    {
      "epoch": 1.2751407884151247,
      "grad_norm": 0.2858302671808592,
      "learning_rate": 7.112619960715004e-06,
      "loss": 0.0247,
      "step": 4755
    },
    {
      "epoch": 1.275408956824886,
      "grad_norm": 0.978837405781421,
      "learning_rate": 7.111205779636451e-06,
      "loss": 0.0328,
      "step": 4756
    },
    {
      "epoch": 1.2756771252346475,
      "grad_norm": 0.3678102723854502,
      "learning_rate": 7.109791392995398e-06,
      "loss": 0.0377,
      "step": 4757
    },
    {
      "epoch": 1.2759452936444087,
      "grad_norm": 0.30400460572480026,
      "learning_rate": 7.108376800929557e-06,
      "loss": 0.0244,
      "step": 4758
    },
    {
      "epoch": 1.27621346205417,
      "grad_norm": 0.2862356258847062,
      "learning_rate": 7.106962003576667e-06,
      "loss": 0.0218,
      "step": 4759
    },
    {
      "epoch": 1.2764816304639313,
      "grad_norm": 0.3866617766850235,
      "learning_rate": 7.10554700107448e-06,
      "loss": 0.0332,
      "step": 4760
    },
    {
      "epoch": 1.2767497988736927,
      "grad_norm": 0.9133371131334812,
      "learning_rate": 7.104131793560769e-06,
      "loss": 0.0265,
      "step": 4761
    },
    {
      "epoch": 1.277017967283454,
      "grad_norm": 0.38227843211387863,
      "learning_rate": 7.102716381173333e-06,
      "loss": 0.0276,
      "step": 4762
    },
    {
      "epoch": 1.2772861356932155,
      "grad_norm": 0.31344339500752677,
      "learning_rate": 7.1013007640499876e-06,
      "loss": 0.0326,
      "step": 4763
    },
    {
      "epoch": 1.2775543041029767,
      "grad_norm": 0.2587135210242422,
      "learning_rate": 7.0998849423285634e-06,
      "loss": 0.0238,
      "step": 4764
    },
    {
      "epoch": 1.277822472512738,
      "grad_norm": 0.4377231954297647,
      "learning_rate": 7.098468916146919e-06,
      "loss": 0.0229,
      "step": 4765
    },
    {
      "epoch": 1.2780906409224992,
      "grad_norm": 0.32452490064993916,
      "learning_rate": 7.097052685642926e-06,
      "loss": 0.0237,
      "step": 4766
    },
    {
      "epoch": 1.2783588093322606,
      "grad_norm": 0.28134442810637317,
      "learning_rate": 7.095636250954481e-06,
      "loss": 0.021,
      "step": 4767
    },
    {
      "epoch": 1.278626977742022,
      "grad_norm": 0.36106217262579693,
      "learning_rate": 7.094219612219499e-06,
      "loss": 0.0385,
      "step": 4768
    },
    {
      "epoch": 1.2788951461517835,
      "grad_norm": 0.28043151031439695,
      "learning_rate": 7.092802769575912e-06,
      "loss": 0.025,
      "step": 4769
    },
    {
      "epoch": 1.2791633145615446,
      "grad_norm": 0.24466535471147616,
      "learning_rate": 7.091385723161677e-06,
      "loss": 0.0192,
      "step": 4770
    },
    {
      "epoch": 1.279431482971306,
      "grad_norm": 0.29842425166775094,
      "learning_rate": 7.089968473114766e-06,
      "loss": 0.0255,
      "step": 4771
    },
    {
      "epoch": 1.2796996513810672,
      "grad_norm": 0.3872752327369388,
      "learning_rate": 7.088551019573173e-06,
      "loss": 0.0235,
      "step": 4772
    },
    {
      "epoch": 1.2799678197908286,
      "grad_norm": 0.268444105498665,
      "learning_rate": 7.0871333626749124e-06,
      "loss": 0.0245,
      "step": 4773
    },
    {
      "epoch": 1.28023598820059,
      "grad_norm": 0.2841865000366025,
      "learning_rate": 7.08571550255802e-06,
      "loss": 0.0281,
      "step": 4774
    },
    {
      "epoch": 1.2805041566103512,
      "grad_norm": 0.2904410037739432,
      "learning_rate": 7.084297439360544e-06,
      "loss": 0.0272,
      "step": 4775
    },
    {
      "epoch": 1.2807723250201126,
      "grad_norm": 0.3339511697779544,
      "learning_rate": 7.082879173220559e-06,
      "loss": 0.0272,
      "step": 4776
    },
    {
      "epoch": 1.281040493429874,
      "grad_norm": 0.33707742489089143,
      "learning_rate": 7.08146070427616e-06,
      "loss": 0.027,
      "step": 4777
    },
    {
      "epoch": 1.2813086618396352,
      "grad_norm": 0.33320785171339967,
      "learning_rate": 7.0800420326654596e-06,
      "loss": 0.0235,
      "step": 4778
    },
    {
      "epoch": 1.2815768302493966,
      "grad_norm": 0.3066933512206115,
      "learning_rate": 7.078623158526588e-06,
      "loss": 0.0232,
      "step": 4779
    },
    {
      "epoch": 1.281844998659158,
      "grad_norm": 0.4262843644960294,
      "learning_rate": 7.0772040819976986e-06,
      "loss": 0.0199,
      "step": 4780
    },
    {
      "epoch": 1.2821131670689192,
      "grad_norm": 0.27367984142949514,
      "learning_rate": 7.075784803216962e-06,
      "loss": 0.0327,
      "step": 4781
    },
    {
      "epoch": 1.2823813354786806,
      "grad_norm": 0.23137538533687435,
      "learning_rate": 7.074365322322569e-06,
      "loss": 0.0233,
      "step": 4782
    },
    {
      "epoch": 1.282649503888442,
      "grad_norm": 0.2850597057186108,
      "learning_rate": 7.072945639452734e-06,
      "loss": 0.0251,
      "step": 4783
    },
    {
      "epoch": 1.2829176722982032,
      "grad_norm": 0.25481635694712884,
      "learning_rate": 7.071525754745685e-06,
      "loss": 0.0185,
      "step": 4784
    },
    {
      "epoch": 1.2831858407079646,
      "grad_norm": 0.4018834688385554,
      "learning_rate": 7.070105668339673e-06,
      "loss": 0.027,
      "step": 4785
    },
    {
      "epoch": 1.283454009117726,
      "grad_norm": 0.24589730498200643,
      "learning_rate": 7.068685380372969e-06,
      "loss": 0.0206,
      "step": 4786
    },
    {
      "epoch": 1.2837221775274872,
      "grad_norm": 0.24766180555617603,
      "learning_rate": 7.0672648909838605e-06,
      "loss": 0.0218,
      "step": 4787
    },
    {
      "epoch": 1.2839903459372486,
      "grad_norm": 0.2783606320911617,
      "learning_rate": 7.065844200310657e-06,
      "loss": 0.0238,
      "step": 4788
    },
    {
      "epoch": 1.28425851434701,
      "grad_norm": 0.2477553004410436,
      "learning_rate": 7.06442330849169e-06,
      "loss": 0.0233,
      "step": 4789
    },
    {
      "epoch": 1.2845266827567712,
      "grad_norm": 0.28577697983506045,
      "learning_rate": 7.063002215665306e-06,
      "loss": 0.0184,
      "step": 4790
    },
    {
      "epoch": 1.2847948511665326,
      "grad_norm": 0.2996777420822747,
      "learning_rate": 7.061580921969875e-06,
      "loss": 0.0303,
      "step": 4791
    },
    {
      "epoch": 1.285063019576294,
      "grad_norm": 0.31173212762765723,
      "learning_rate": 7.060159427543782e-06,
      "loss": 0.0223,
      "step": 4792
    },
    {
      "epoch": 1.2853311879860552,
      "grad_norm": 0.37916466301197804,
      "learning_rate": 7.058737732525434e-06,
      "loss": 0.0271,
      "step": 4793
    },
    {
      "epoch": 1.2855993563958166,
      "grad_norm": 0.32937563128442965,
      "learning_rate": 7.057315837053258e-06,
      "loss": 0.0346,
      "step": 4794
    },
    {
      "epoch": 1.285867524805578,
      "grad_norm": 0.24912591263883926,
      "learning_rate": 7.055893741265704e-06,
      "loss": 0.0211,
      "step": 4795
    },
    {
      "epoch": 1.2861356932153392,
      "grad_norm": 0.26247988664809824,
      "learning_rate": 7.054471445301233e-06,
      "loss": 0.0232,
      "step": 4796
    },
    {
      "epoch": 1.2864038616251006,
      "grad_norm": 0.22760702276672298,
      "learning_rate": 7.053048949298332e-06,
      "loss": 0.0207,
      "step": 4797
    },
    {
      "epoch": 1.286672030034862,
      "grad_norm": 0.36051326131093536,
      "learning_rate": 7.051626253395506e-06,
      "loss": 0.0313,
      "step": 4798
    },
    {
      "epoch": 1.2869401984446232,
      "grad_norm": 0.30355714404803946,
      "learning_rate": 7.0502033577312775e-06,
      "loss": 0.0298,
      "step": 4799
    },
    {
      "epoch": 1.2872083668543846,
      "grad_norm": 0.277884193966726,
      "learning_rate": 7.0487802624441925e-06,
      "loss": 0.0239,
      "step": 4800
    },
    {
      "epoch": 1.287476535264146,
      "grad_norm": 0.2892237300536667,
      "learning_rate": 7.0473569676728134e-06,
      "loss": 0.0257,
      "step": 4801
    },
    {
      "epoch": 1.2877447036739071,
      "grad_norm": 0.2660815314578183,
      "learning_rate": 7.045933473555721e-06,
      "loss": 0.0249,
      "step": 4802
    },
    {
      "epoch": 1.2880128720836685,
      "grad_norm": 0.5281454350174769,
      "learning_rate": 7.044509780231517e-06,
      "loss": 0.035,
      "step": 4803
    },
    {
      "epoch": 1.28828104049343,
      "grad_norm": 0.33086197563769065,
      "learning_rate": 7.043085887838825e-06,
      "loss": 0.0286,
      "step": 4804
    },
    {
      "epoch": 1.2885492089031911,
      "grad_norm": 0.2447160752965827,
      "learning_rate": 7.041661796516283e-06,
      "loss": 0.0257,
      "step": 4805
    },
    {
      "epoch": 1.2888173773129525,
      "grad_norm": 0.21796824603750267,
      "learning_rate": 7.040237506402554e-06,
      "loss": 0.0232,
      "step": 4806
    },
    {
      "epoch": 1.289085545722714,
      "grad_norm": 0.27237410007160057,
      "learning_rate": 7.038813017636317e-06,
      "loss": 0.0352,
      "step": 4807
    },
    {
      "epoch": 1.2893537141324751,
      "grad_norm": 0.24921526916866615,
      "learning_rate": 7.0373883303562675e-06,
      "loss": 0.0242,
      "step": 4808
    },
    {
      "epoch": 1.2896218825422365,
      "grad_norm": 0.39418199912321283,
      "learning_rate": 7.035963444701125e-06,
      "loss": 0.0348,
      "step": 4809
    },
    {
      "epoch": 1.289890050951998,
      "grad_norm": 0.2047648246584857,
      "learning_rate": 7.034538360809628e-06,
      "loss": 0.0161,
      "step": 4810
    },
    {
      "epoch": 1.2901582193617591,
      "grad_norm": 0.29033463282902877,
      "learning_rate": 7.0331130788205325e-06,
      "loss": 0.0274,
      "step": 4811
    },
    {
      "epoch": 1.2904263877715205,
      "grad_norm": 0.2928477146104081,
      "learning_rate": 7.031687598872614e-06,
      "loss": 0.0207,
      "step": 4812
    },
    {
      "epoch": 1.290694556181282,
      "grad_norm": 0.28932778345600285,
      "learning_rate": 7.03026192110467e-06,
      "loss": 0.0242,
      "step": 4813
    },
    {
      "epoch": 1.290962724591043,
      "grad_norm": 0.42108012916388293,
      "learning_rate": 7.028836045655509e-06,
      "loss": 0.0284,
      "step": 4814
    },
    {
      "epoch": 1.2912308930008045,
      "grad_norm": 0.35370972371715004,
      "learning_rate": 7.027409972663972e-06,
      "loss": 0.0289,
      "step": 4815
    },
    {
      "epoch": 1.291499061410566,
      "grad_norm": 0.3058040764888948,
      "learning_rate": 7.025983702268907e-06,
      "loss": 0.0208,
      "step": 4816
    },
    {
      "epoch": 1.291767229820327,
      "grad_norm": 0.3268480988852991,
      "learning_rate": 7.024557234609189e-06,
      "loss": 0.0355,
      "step": 4817
    },
    {
      "epoch": 1.2920353982300885,
      "grad_norm": 0.3528283882842599,
      "learning_rate": 7.023130569823707e-06,
      "loss": 0.0207,
      "step": 4818
    },
    {
      "epoch": 1.2923035666398497,
      "grad_norm": 0.43276314773533986,
      "learning_rate": 7.021703708051375e-06,
      "loss": 0.0251,
      "step": 4819
    },
    {
      "epoch": 1.292571735049611,
      "grad_norm": 0.25370146613199024,
      "learning_rate": 7.020276649431117e-06,
      "loss": 0.0184,
      "step": 4820
    },
    {
      "epoch": 1.2928399034593725,
      "grad_norm": 0.31952983852406536,
      "learning_rate": 7.018849394101888e-06,
      "loss": 0.015,
      "step": 4821
    },
    {
      "epoch": 1.293108071869134,
      "grad_norm": 0.3064726012251518,
      "learning_rate": 7.017421942202651e-06,
      "loss": 0.0289,
      "step": 4822
    },
    {
      "epoch": 1.293376240278895,
      "grad_norm": 0.297163986002648,
      "learning_rate": 7.015994293872398e-06,
      "loss": 0.024,
      "step": 4823
    },
    {
      "epoch": 1.2936444086886565,
      "grad_norm": 0.3367729279902189,
      "learning_rate": 7.014566449250132e-06,
      "loss": 0.0361,
      "step": 4824
    },
    {
      "epoch": 1.2939125770984177,
      "grad_norm": 0.28463481177564703,
      "learning_rate": 7.01313840847488e-06,
      "loss": 0.021,
      "step": 4825
    },
    {
      "epoch": 1.294180745508179,
      "grad_norm": 0.31739381243300907,
      "learning_rate": 7.011710171685685e-06,
      "loss": 0.0261,
      "step": 4826
    },
    {
      "epoch": 1.2944489139179405,
      "grad_norm": 0.26269494074434335,
      "learning_rate": 7.010281739021612e-06,
      "loss": 0.0254,
      "step": 4827
    },
    {
      "epoch": 1.2947170823277019,
      "grad_norm": 0.3114438789220039,
      "learning_rate": 7.008853110621744e-06,
      "loss": 0.0242,
      "step": 4828
    },
    {
      "epoch": 1.294985250737463,
      "grad_norm": 0.2765408180816259,
      "learning_rate": 7.007424286625181e-06,
      "loss": 0.0246,
      "step": 4829
    },
    {
      "epoch": 1.2952534191472245,
      "grad_norm": 0.31736257463918727,
      "learning_rate": 7.005995267171045e-06,
      "loss": 0.0201,
      "step": 4830
    },
    {
      "epoch": 1.2955215875569857,
      "grad_norm": 0.4419308689434764,
      "learning_rate": 7.004566052398476e-06,
      "loss": 0.0199,
      "step": 4831
    },
    {
      "epoch": 1.295789755966747,
      "grad_norm": 0.28619185258921215,
      "learning_rate": 7.003136642446632e-06,
      "loss": 0.0198,
      "step": 4832
    },
    {
      "epoch": 1.2960579243765085,
      "grad_norm": 0.24417459606865535,
      "learning_rate": 7.001707037454693e-06,
      "loss": 0.023,
      "step": 4833
    },
    {
      "epoch": 1.2963260927862699,
      "grad_norm": 0.2782980026858033,
      "learning_rate": 7.000277237561852e-06,
      "loss": 0.0241,
      "step": 4834
    },
    {
      "epoch": 1.296594261196031,
      "grad_norm": 0.24062819725398713,
      "learning_rate": 6.998847242907327e-06,
      "loss": 0.0244,
      "step": 4835
    },
    {
      "epoch": 1.2968624296057925,
      "grad_norm": 0.28502146271686457,
      "learning_rate": 6.997417053630354e-06,
      "loss": 0.0284,
      "step": 4836
    },
    {
      "epoch": 1.2971305980155536,
      "grad_norm": 0.4186347837248692,
      "learning_rate": 6.995986669870185e-06,
      "loss": 0.0337,
      "step": 4837
    },
    {
      "epoch": 1.297398766425315,
      "grad_norm": 0.4421422998790123,
      "learning_rate": 6.994556091766094e-06,
      "loss": 0.0292,
      "step": 4838
    },
    {
      "epoch": 1.2976669348350764,
      "grad_norm": 0.31127006541185676,
      "learning_rate": 6.993125319457371e-06,
      "loss": 0.0234,
      "step": 4839
    },
    {
      "epoch": 1.2979351032448379,
      "grad_norm": 0.38932711669783254,
      "learning_rate": 6.991694353083327e-06,
      "loss": 0.023,
      "step": 4840
    },
    {
      "epoch": 1.298203271654599,
      "grad_norm": 0.25418419706694056,
      "learning_rate": 6.990263192783293e-06,
      "loss": 0.025,
      "step": 4841
    },
    {
      "epoch": 1.2984714400643604,
      "grad_norm": 0.30014918209266206,
      "learning_rate": 6.988831838696614e-06,
      "loss": 0.0341,
      "step": 4842
    },
    {
      "epoch": 1.2987396084741216,
      "grad_norm": 0.20695649062811497,
      "learning_rate": 6.987400290962661e-06,
      "loss": 0.0184,
      "step": 4843
    },
    {
      "epoch": 1.299007776883883,
      "grad_norm": 0.3194793690667802,
      "learning_rate": 6.985968549720817e-06,
      "loss": 0.0238,
      "step": 4844
    },
    {
      "epoch": 1.2992759452936444,
      "grad_norm": 0.38394850866529884,
      "learning_rate": 6.984536615110489e-06,
      "loss": 0.0248,
      "step": 4845
    },
    {
      "epoch": 1.2995441137034058,
      "grad_norm": 0.2834082241347149,
      "learning_rate": 6.983104487271099e-06,
      "loss": 0.0252,
      "step": 4846
    },
    {
      "epoch": 1.299812282113167,
      "grad_norm": 0.21470481357818075,
      "learning_rate": 6.981672166342088e-06,
      "loss": 0.0215,
      "step": 4847
    },
    {
      "epoch": 1.3000804505229284,
      "grad_norm": 0.22390070392440534,
      "learning_rate": 6.980239652462921e-06,
      "loss": 0.0203,
      "step": 4848
    },
    {
      "epoch": 1.3003486189326896,
      "grad_norm": 0.25660961695523843,
      "learning_rate": 6.978806945773078e-06,
      "loss": 0.0236,
      "step": 4849
    },
    {
      "epoch": 1.300616787342451,
      "grad_norm": 0.27222541444735276,
      "learning_rate": 6.977374046412056e-06,
      "loss": 0.0286,
      "step": 4850
    },
    {
      "epoch": 1.3008849557522124,
      "grad_norm": 0.810680475093095,
      "learning_rate": 6.975940954519372e-06,
      "loss": 0.0356,
      "step": 4851
    },
    {
      "epoch": 1.3011531241619738,
      "grad_norm": 0.2744870102858102,
      "learning_rate": 6.974507670234563e-06,
      "loss": 0.0293,
      "step": 4852
    },
    {
      "epoch": 1.301421292571735,
      "grad_norm": 0.20740687727674478,
      "learning_rate": 6.9730741936971844e-06,
      "loss": 0.0177,
      "step": 4853
    },
    {
      "epoch": 1.3016894609814964,
      "grad_norm": 0.26604847943190046,
      "learning_rate": 6.97164052504681e-06,
      "loss": 0.0238,
      "step": 4854
    },
    {
      "epoch": 1.3019576293912576,
      "grad_norm": 0.433969866889402,
      "learning_rate": 6.970206664423034e-06,
      "loss": 0.0193,
      "step": 4855
    },
    {
      "epoch": 1.302225797801019,
      "grad_norm": 0.2929816513194438,
      "learning_rate": 6.968772611965464e-06,
      "loss": 0.0285,
      "step": 4856
    },
    {
      "epoch": 1.3024939662107804,
      "grad_norm": 0.2465491294944875,
      "learning_rate": 6.967338367813732e-06,
      "loss": 0.0266,
      "step": 4857
    },
    {
      "epoch": 1.3027621346205418,
      "grad_norm": 0.23244167731757573,
      "learning_rate": 6.965903932107488e-06,
      "loss": 0.0219,
      "step": 4858
    },
    {
      "epoch": 1.303030303030303,
      "grad_norm": 0.584477894418307,
      "learning_rate": 6.964469304986394e-06,
      "loss": 0.0425,
      "step": 4859
    },
    {
      "epoch": 1.3032984714400644,
      "grad_norm": 1.03064282865046,
      "learning_rate": 6.963034486590142e-06,
      "loss": 0.0305,
      "step": 4860
    },
    {
      "epoch": 1.3035666398498256,
      "grad_norm": 0.2492180453331263,
      "learning_rate": 6.9615994770584335e-06,
      "loss": 0.0209,
      "step": 4861
    },
    {
      "epoch": 1.303834808259587,
      "grad_norm": 0.323259893857889,
      "learning_rate": 6.9601642765309925e-06,
      "loss": 0.0218,
      "step": 4862
    },
    {
      "epoch": 1.3041029766693484,
      "grad_norm": 0.3059534688000298,
      "learning_rate": 6.958728885147559e-06,
      "loss": 0.0361,
      "step": 4863
    },
    {
      "epoch": 1.3043711450791098,
      "grad_norm": 0.2697071480144899,
      "learning_rate": 6.957293303047894e-06,
      "loss": 0.0251,
      "step": 4864
    },
    {
      "epoch": 1.304639313488871,
      "grad_norm": 0.7178545448976204,
      "learning_rate": 6.955857530371777e-06,
      "loss": 0.0391,
      "step": 4865
    },
    {
      "epoch": 1.3049074818986324,
      "grad_norm": 0.29814923602884263,
      "learning_rate": 6.954421567259004e-06,
      "loss": 0.0297,
      "step": 4866
    },
    {
      "epoch": 1.3051756503083936,
      "grad_norm": 0.20432820919085684,
      "learning_rate": 6.952985413849393e-06,
      "loss": 0.0171,
      "step": 4867
    },
    {
      "epoch": 1.305443818718155,
      "grad_norm": 0.4457507368037086,
      "learning_rate": 6.951549070282778e-06,
      "loss": 0.042,
      "step": 4868
    },
    {
      "epoch": 1.3057119871279164,
      "grad_norm": 0.32892171814189264,
      "learning_rate": 6.95011253669901e-06,
      "loss": 0.0379,
      "step": 4869
    },
    {
      "epoch": 1.3059801555376778,
      "grad_norm": 0.23763358060658146,
      "learning_rate": 6.948675813237963e-06,
      "loss": 0.0186,
      "step": 4870
    },
    {
      "epoch": 1.306248323947439,
      "grad_norm": 0.29450754689905423,
      "learning_rate": 6.9472389000395245e-06,
      "loss": 0.024,
      "step": 4871
    },
    {
      "epoch": 1.3065164923572004,
      "grad_norm": 0.41974692404292063,
      "learning_rate": 6.945801797243604e-06,
      "loss": 0.0203,
      "step": 4872
    },
    {
      "epoch": 1.3067846607669615,
      "grad_norm": 0.3398192797197089,
      "learning_rate": 6.94436450499013e-06,
      "loss": 0.0245,
      "step": 4873
    },
    {
      "epoch": 1.307052829176723,
      "grad_norm": 0.2937893670893608,
      "learning_rate": 6.9429270234190445e-06,
      "loss": 0.0258,
      "step": 4874
    },
    {
      "epoch": 1.3073209975864843,
      "grad_norm": 0.3502837389274702,
      "learning_rate": 6.941489352670315e-06,
      "loss": 0.0294,
      "step": 4875
    },
    {
      "epoch": 1.3075891659962458,
      "grad_norm": 0.29098717465034046,
      "learning_rate": 6.94005149288392e-06,
      "loss": 0.0253,
      "step": 4876
    },
    {
      "epoch": 1.307857334406007,
      "grad_norm": 0.461387272545984,
      "learning_rate": 6.938613444199863e-06,
      "loss": 0.0281,
      "step": 4877
    },
    {
      "epoch": 1.3081255028157683,
      "grad_norm": 0.4530933206181577,
      "learning_rate": 6.937175206758162e-06,
      "loss": 0.0352,
      "step": 4878
    },
    {
      "epoch": 1.3083936712255295,
      "grad_norm": 0.32241022629790644,
      "learning_rate": 6.935736780698852e-06,
      "loss": 0.0294,
      "step": 4879
    },
    {
      "epoch": 1.308661839635291,
      "grad_norm": 0.2906357516500285,
      "learning_rate": 6.934298166161994e-06,
      "loss": 0.0235,
      "step": 4880
    },
    {
      "epoch": 1.3089300080450523,
      "grad_norm": 1.0161698066131566,
      "learning_rate": 6.932859363287658e-06,
      "loss": 0.0263,
      "step": 4881
    },
    {
      "epoch": 1.3091981764548137,
      "grad_norm": 0.30313904851121987,
      "learning_rate": 6.931420372215937e-06,
      "loss": 0.025,
      "step": 4882
    },
    {
      "epoch": 1.309466344864575,
      "grad_norm": 0.2533592012422497,
      "learning_rate": 6.9299811930869424e-06,
      "loss": 0.0204,
      "step": 4883
    },
    {
      "epoch": 1.3097345132743363,
      "grad_norm": 0.4603647696553685,
      "learning_rate": 6.928541826040802e-06,
      "loss": 0.0278,
      "step": 4884
    },
    {
      "epoch": 1.3100026816840975,
      "grad_norm": 0.41317225891731385,
      "learning_rate": 6.927102271217665e-06,
      "loss": 0.0342,
      "step": 4885
    },
    {
      "epoch": 1.310270850093859,
      "grad_norm": 0.39563307264148795,
      "learning_rate": 6.925662528757697e-06,
      "loss": 0.027,
      "step": 4886
    },
    {
      "epoch": 1.3105390185036203,
      "grad_norm": 0.33441073436271396,
      "learning_rate": 6.92422259880108e-06,
      "loss": 0.039,
      "step": 4887
    },
    {
      "epoch": 1.3108071869133817,
      "grad_norm": 0.2699155256210237,
      "learning_rate": 6.922782481488018e-06,
      "loss": 0.0237,
      "step": 4888
    },
    {
      "epoch": 1.311075355323143,
      "grad_norm": 0.304075316409533,
      "learning_rate": 6.921342176958731e-06,
      "loss": 0.0275,
      "step": 4889
    },
    {
      "epoch": 1.3113435237329043,
      "grad_norm": 0.30637682322971943,
      "learning_rate": 6.919901685353456e-06,
      "loss": 0.0262,
      "step": 4890
    },
    {
      "epoch": 1.3116116921426655,
      "grad_norm": 0.3259472298533724,
      "learning_rate": 6.918461006812451e-06,
      "loss": 0.0218,
      "step": 4891
    },
    {
      "epoch": 1.311879860552427,
      "grad_norm": 0.3099504722722003,
      "learning_rate": 6.917020141475993e-06,
      "loss": 0.0243,
      "step": 4892
    },
    {
      "epoch": 1.3121480289621883,
      "grad_norm": 0.297677554356472,
      "learning_rate": 6.915579089484373e-06,
      "loss": 0.0218,
      "step": 4893
    },
    {
      "epoch": 1.3124161973719497,
      "grad_norm": 0.2747069678237856,
      "learning_rate": 6.914137850977902e-06,
      "loss": 0.0233,
      "step": 4894
    },
    {
      "epoch": 1.3126843657817109,
      "grad_norm": 0.6239502987725213,
      "learning_rate": 6.91269642609691e-06,
      "loss": 0.0304,
      "step": 4895
    },
    {
      "epoch": 1.3129525341914723,
      "grad_norm": 0.25087415344624836,
      "learning_rate": 6.911254814981745e-06,
      "loss": 0.024,
      "step": 4896
    },
    {
      "epoch": 1.3132207026012335,
      "grad_norm": 0.38573212091099346,
      "learning_rate": 6.909813017772774e-06,
      "loss": 0.0273,
      "step": 4897
    },
    {
      "epoch": 1.3134888710109949,
      "grad_norm": 0.2707328617433765,
      "learning_rate": 6.908371034610381e-06,
      "loss": 0.0249,
      "step": 4898
    },
    {
      "epoch": 1.3137570394207563,
      "grad_norm": 0.22384755211100074,
      "learning_rate": 6.9069288656349654e-06,
      "loss": 0.0244,
      "step": 4899
    },
    {
      "epoch": 1.3140252078305177,
      "grad_norm": 0.2420820051508984,
      "learning_rate": 6.905486510986947e-06,
      "loss": 0.0207,
      "step": 4900
    },
    {
      "epoch": 1.3142933762402789,
      "grad_norm": 0.34842368181048305,
      "learning_rate": 6.904043970806769e-06,
      "loss": 0.0231,
      "step": 4901
    },
    {
      "epoch": 1.3145615446500403,
      "grad_norm": 0.41518191732115567,
      "learning_rate": 6.902601245234883e-06,
      "loss": 0.0285,
      "step": 4902
    },
    {
      "epoch": 1.3148297130598015,
      "grad_norm": 0.19781740071146248,
      "learning_rate": 6.901158334411767e-06,
      "loss": 0.0177,
      "step": 4903
    },
    {
      "epoch": 1.3150978814695629,
      "grad_norm": 0.2886746586953782,
      "learning_rate": 6.899715238477911e-06,
      "loss": 0.0299,
      "step": 4904
    },
    {
      "epoch": 1.3153660498793243,
      "grad_norm": 0.3185701982662068,
      "learning_rate": 6.8982719575738245e-06,
      "loss": 0.0259,
      "step": 4905
    },
    {
      "epoch": 1.3156342182890857,
      "grad_norm": 0.34932236817918927,
      "learning_rate": 6.896828491840037e-06,
      "loss": 0.0436,
      "step": 4906
    },
    {
      "epoch": 1.3159023866988468,
      "grad_norm": 0.26564991458347226,
      "learning_rate": 6.895384841417097e-06,
      "loss": 0.0183,
      "step": 4907
    },
    {
      "epoch": 1.3161705551086083,
      "grad_norm": 0.723695264267591,
      "learning_rate": 6.893941006445567e-06,
      "loss": 0.0201,
      "step": 4908
    },
    {
      "epoch": 1.3164387235183694,
      "grad_norm": 0.38622558453268685,
      "learning_rate": 6.892496987066029e-06,
      "loss": 0.0209,
      "step": 4909
    },
    {
      "epoch": 1.3167068919281308,
      "grad_norm": 0.25010421752593626,
      "learning_rate": 6.891052783419084e-06,
      "loss": 0.0231,
      "step": 4910
    },
    {
      "epoch": 1.3169750603378922,
      "grad_norm": 0.3222507198631701,
      "learning_rate": 6.8896083956453495e-06,
      "loss": 0.0261,
      "step": 4911
    },
    {
      "epoch": 1.3172432287476536,
      "grad_norm": 0.3208914690053035,
      "learning_rate": 6.888163823885463e-06,
      "loss": 0.0187,
      "step": 4912
    },
    {
      "epoch": 1.3175113971574148,
      "grad_norm": 0.4486125192031171,
      "learning_rate": 6.886719068280076e-06,
      "loss": 0.0351,
      "step": 4913
    },
    {
      "epoch": 1.3177795655671762,
      "grad_norm": 0.5377007631316899,
      "learning_rate": 6.8852741289698656e-06,
      "loss": 0.0328,
      "step": 4914
    },
    {
      "epoch": 1.3180477339769374,
      "grad_norm": 0.24357504741930416,
      "learning_rate": 6.883829006095517e-06,
      "loss": 0.0186,
      "step": 4915
    },
    {
      "epoch": 1.3183159023866988,
      "grad_norm": 0.38088382229333384,
      "learning_rate": 6.882383699797739e-06,
      "loss": 0.0324,
      "step": 4916
    },
    {
      "epoch": 1.3185840707964602,
      "grad_norm": 0.3394026464087262,
      "learning_rate": 6.880938210217258e-06,
      "loss": 0.0308,
      "step": 4917
    },
    {
      "epoch": 1.3188522392062216,
      "grad_norm": 0.29869246423031914,
      "learning_rate": 6.879492537494818e-06,
      "loss": 0.0281,
      "step": 4918
    },
    {
      "epoch": 1.3191204076159828,
      "grad_norm": 0.2598201693530205,
      "learning_rate": 6.87804668177118e-06,
      "loss": 0.0269,
      "step": 4919
    },
    {
      "epoch": 1.3193885760257442,
      "grad_norm": 0.3646353847608352,
      "learning_rate": 6.876600643187122e-06,
      "loss": 0.0293,
      "step": 4920
    },
    {
      "epoch": 1.3196567444355054,
      "grad_norm": 0.49146531165977025,
      "learning_rate": 6.875154421883441e-06,
      "loss": 0.0233,
      "step": 4921
    },
    {
      "epoch": 1.3199249128452668,
      "grad_norm": 0.216996068430698,
      "learning_rate": 6.873708018000953e-06,
      "loss": 0.0193,
      "step": 4922
    },
    {
      "epoch": 1.3201930812550282,
      "grad_norm": 0.300845406352333,
      "learning_rate": 6.87226143168049e-06,
      "loss": 0.0224,
      "step": 4923
    },
    {
      "epoch": 1.3204612496647896,
      "grad_norm": 0.2947035824608659,
      "learning_rate": 6.870814663062902e-06,
      "loss": 0.0322,
      "step": 4924
    },
    {
      "epoch": 1.3207294180745508,
      "grad_norm": 0.276578861351353,
      "learning_rate": 6.869367712289058e-06,
      "loss": 0.0218,
      "step": 4925
    },
    {
      "epoch": 1.3209975864843122,
      "grad_norm": 0.31121376375453613,
      "learning_rate": 6.867920579499842e-06,
      "loss": 0.0346,
      "step": 4926
    },
    {
      "epoch": 1.3212657548940734,
      "grad_norm": 0.6949039369640413,
      "learning_rate": 6.866473264836158e-06,
      "loss": 0.0245,
      "step": 4927
    },
    {
      "epoch": 1.3215339233038348,
      "grad_norm": 0.30704677991537715,
      "learning_rate": 6.865025768438929e-06,
      "loss": 0.032,
      "step": 4928
    },
    {
      "epoch": 1.3218020917135962,
      "grad_norm": 0.371050280348174,
      "learning_rate": 6.8635780904490915e-06,
      "loss": 0.0399,
      "step": 4929
    },
    {
      "epoch": 1.3220702601233576,
      "grad_norm": 0.3013306753539082,
      "learning_rate": 6.862130231007603e-06,
      "loss": 0.0281,
      "step": 4930
    },
    {
      "epoch": 1.3223384285331188,
      "grad_norm": 0.33183434083088414,
      "learning_rate": 6.860682190255437e-06,
      "loss": 0.0305,
      "step": 4931
    },
    {
      "epoch": 1.3226065969428802,
      "grad_norm": 0.2310559487393208,
      "learning_rate": 6.8592339683335886e-06,
      "loss": 0.0238,
      "step": 4932
    },
    {
      "epoch": 1.3228747653526414,
      "grad_norm": 0.36213826723591064,
      "learning_rate": 6.857785565383062e-06,
      "loss": 0.0375,
      "step": 4933
    },
    {
      "epoch": 1.3231429337624028,
      "grad_norm": 0.25825111427282366,
      "learning_rate": 6.856336981544889e-06,
      "loss": 0.0251,
      "step": 4934
    },
    {
      "epoch": 1.3234111021721642,
      "grad_norm": 0.28790813158603973,
      "learning_rate": 6.8548882169601125e-06,
      "loss": 0.0228,
      "step": 4935
    },
    {
      "epoch": 1.3236792705819256,
      "grad_norm": 0.3759520759105815,
      "learning_rate": 6.853439271769795e-06,
      "loss": 0.0239,
      "step": 4936
    },
    {
      "epoch": 1.3239474389916868,
      "grad_norm": 0.37512221341993857,
      "learning_rate": 6.851990146115016e-06,
      "loss": 0.0281,
      "step": 4937
    },
    {
      "epoch": 1.3242156074014482,
      "grad_norm": 0.31777879749238375,
      "learning_rate": 6.850540840136872e-06,
      "loss": 0.0229,
      "step": 4938
    },
    {
      "epoch": 1.3244837758112094,
      "grad_norm": 0.17559967255584505,
      "learning_rate": 6.849091353976481e-06,
      "loss": 0.0145,
      "step": 4939
    },
    {
      "epoch": 1.3247519442209708,
      "grad_norm": 0.3259172329794027,
      "learning_rate": 6.8476416877749754e-06,
      "loss": 0.0327,
      "step": 4940
    },
    {
      "epoch": 1.3250201126307322,
      "grad_norm": 0.2907116942161748,
      "learning_rate": 6.846191841673503e-06,
      "loss": 0.0329,
      "step": 4941
    },
    {
      "epoch": 1.3252882810404936,
      "grad_norm": 0.25019848212929,
      "learning_rate": 6.844741815813232e-06,
      "loss": 0.0218,
      "step": 4942
    },
    {
      "epoch": 1.3255564494502547,
      "grad_norm": 0.258364293260714,
      "learning_rate": 6.843291610335347e-06,
      "loss": 0.0233,
      "step": 4943
    },
    {
      "epoch": 1.3258246178600162,
      "grad_norm": 0.3563590288885514,
      "learning_rate": 6.841841225381053e-06,
      "loss": 0.0146,
      "step": 4944
    },
    {
      "epoch": 1.3260927862697773,
      "grad_norm": 0.3420528503957618,
      "learning_rate": 6.840390661091569e-06,
      "loss": 0.022,
      "step": 4945
    },
    {
      "epoch": 1.3263609546795387,
      "grad_norm": 0.41597688452723464,
      "learning_rate": 6.838939917608134e-06,
      "loss": 0.027,
      "step": 4946
    },
    {
      "epoch": 1.3266291230893001,
      "grad_norm": 0.2802760455006701,
      "learning_rate": 6.837488995071999e-06,
      "loss": 0.0209,
      "step": 4947
    },
    {
      "epoch": 1.3268972914990615,
      "grad_norm": 0.4047945261340898,
      "learning_rate": 6.836037893624439e-06,
      "loss": 0.0292,
      "step": 4948
    },
    {
      "epoch": 1.3271654599088227,
      "grad_norm": 0.30388158283999817,
      "learning_rate": 6.8345866134067454e-06,
      "loss": 0.0252,
      "step": 4949
    },
    {
      "epoch": 1.3274336283185841,
      "grad_norm": 0.2265875521441694,
      "learning_rate": 6.8331351545602235e-06,
      "loss": 0.0165,
      "step": 4950
    },
    {
      "epoch": 1.3277017967283453,
      "grad_norm": 0.37970021307672275,
      "learning_rate": 6.831683517226197e-06,
      "loss": 0.0331,
      "step": 4951
    },
    {
      "epoch": 1.3279699651381067,
      "grad_norm": 0.2763960190806142,
      "learning_rate": 6.830231701546013e-06,
      "loss": 0.0297,
      "step": 4952
    },
    {
      "epoch": 1.3282381335478681,
      "grad_norm": 0.29118190580593933,
      "learning_rate": 6.8287797076610244e-06,
      "loss": 0.0301,
      "step": 4953
    },
    {
      "epoch": 1.3285063019576293,
      "grad_norm": 0.20484567452707492,
      "learning_rate": 6.827327535712611e-06,
      "loss": 0.0179,
      "step": 4954
    },
    {
      "epoch": 1.3287744703673907,
      "grad_norm": 0.34291610280074086,
      "learning_rate": 6.825875185842168e-06,
      "loss": 0.0292,
      "step": 4955
    },
    {
      "epoch": 1.3290426387771521,
      "grad_norm": 0.28051924251426985,
      "learning_rate": 6.824422658191105e-06,
      "loss": 0.0203,
      "step": 4956
    },
    {
      "epoch": 1.3293108071869133,
      "grad_norm": 0.3801734970185959,
      "learning_rate": 6.822969952900852e-06,
      "loss": 0.0418,
      "step": 4957
    },
    {
      "epoch": 1.3295789755966747,
      "grad_norm": 0.3048450416270671,
      "learning_rate": 6.821517070112855e-06,
      "loss": 0.0275,
      "step": 4958
    },
    {
      "epoch": 1.329847144006436,
      "grad_norm": 0.2976863771417503,
      "learning_rate": 6.820064009968577e-06,
      "loss": 0.0262,
      "step": 4959
    },
    {
      "epoch": 1.3301153124161973,
      "grad_norm": 0.29337800063249153,
      "learning_rate": 6.818610772609498e-06,
      "loss": 0.0328,
      "step": 4960
    },
    {
      "epoch": 1.3303834808259587,
      "grad_norm": 0.28890172261264035,
      "learning_rate": 6.817157358177116e-06,
      "loss": 0.0301,
      "step": 4961
    },
    {
      "epoch": 1.33065164923572,
      "grad_norm": 0.31893099145563,
      "learning_rate": 6.815703766812947e-06,
      "loss": 0.0293,
      "step": 4962
    },
    {
      "epoch": 1.3309198176454813,
      "grad_norm": 0.5345769181535043,
      "learning_rate": 6.814249998658522e-06,
      "loss": 0.0233,
      "step": 4963
    },
    {
      "epoch": 1.3311879860552427,
      "grad_norm": 0.3467316316980195,
      "learning_rate": 6.8127960538553925e-06,
      "loss": 0.0275,
      "step": 4964
    },
    {
      "epoch": 1.331456154465004,
      "grad_norm": 0.22988188256488568,
      "learning_rate": 6.811341932545124e-06,
      "loss": 0.0259,
      "step": 4965
    },
    {
      "epoch": 1.3317243228747653,
      "grad_norm": 0.2787744974049355,
      "learning_rate": 6.8098876348693e-06,
      "loss": 0.0274,
      "step": 4966
    },
    {
      "epoch": 1.3319924912845267,
      "grad_norm": 0.27861630160438033,
      "learning_rate": 6.808433160969522e-06,
      "loss": 0.0263,
      "step": 4967
    },
    {
      "epoch": 1.332260659694288,
      "grad_norm": 0.3944310291759222,
      "learning_rate": 6.806978510987409e-06,
      "loss": 0.0257,
      "step": 4968
    },
    {
      "epoch": 1.3325288281040493,
      "grad_norm": 0.2868600141692782,
      "learning_rate": 6.805523685064595e-06,
      "loss": 0.0246,
      "step": 4969
    },
    {
      "epoch": 1.3327969965138107,
      "grad_norm": 0.3465793229177891,
      "learning_rate": 6.804068683342733e-06,
      "loss": 0.0291,
      "step": 4970
    },
    {
      "epoch": 1.333065164923572,
      "grad_norm": 0.28155141041966797,
      "learning_rate": 6.802613505963496e-06,
      "loss": 0.0257,
      "step": 4971
    },
    {
      "epoch": 1.3333333333333333,
      "grad_norm": 0.41168379309458086,
      "learning_rate": 6.801158153068565e-06,
      "loss": 0.0304,
      "step": 4972
    },
    {
      "epoch": 1.3336015017430947,
      "grad_norm": 0.34263938319443127,
      "learning_rate": 6.799702624799648e-06,
      "loss": 0.0313,
      "step": 4973
    },
    {
      "epoch": 1.333869670152856,
      "grad_norm": 0.29541717819596386,
      "learning_rate": 6.798246921298464e-06,
      "loss": 0.0235,
      "step": 4974
    },
    {
      "epoch": 1.3341378385626173,
      "grad_norm": 0.25406444689351265,
      "learning_rate": 6.796791042706751e-06,
      "loss": 0.0224,
      "step": 4975
    },
    {
      "epoch": 1.3344060069723787,
      "grad_norm": 0.25384342543674693,
      "learning_rate": 6.795334989166264e-06,
      "loss": 0.0264,
      "step": 4976
    },
    {
      "epoch": 1.33467417538214,
      "grad_norm": 0.2567595734509517,
      "learning_rate": 6.793878760818777e-06,
      "loss": 0.0236,
      "step": 4977
    },
    {
      "epoch": 1.3349423437919012,
      "grad_norm": 0.32582796434854466,
      "learning_rate": 6.792422357806078e-06,
      "loss": 0.0308,
      "step": 4978
    },
    {
      "epoch": 1.3352105122016626,
      "grad_norm": 0.31064508302798366,
      "learning_rate": 6.790965780269971e-06,
      "loss": 0.0312,
      "step": 4979
    },
    {
      "epoch": 1.335478680611424,
      "grad_norm": 0.4741136829846071,
      "learning_rate": 6.789509028352284e-06,
      "loss": 0.024,
      "step": 4980
    },
    {
      "epoch": 1.3357468490211852,
      "grad_norm": 0.2688833349750044,
      "learning_rate": 6.78805210219485e-06,
      "loss": 0.0254,
      "step": 4981
    },
    {
      "epoch": 1.3360150174309466,
      "grad_norm": 0.2606080248073186,
      "learning_rate": 6.786595001939532e-06,
      "loss": 0.0229,
      "step": 4982
    },
    {
      "epoch": 1.336283185840708,
      "grad_norm": 0.3916911675434629,
      "learning_rate": 6.7851377277282025e-06,
      "loss": 0.0312,
      "step": 4983
    },
    {
      "epoch": 1.3365513542504692,
      "grad_norm": 0.3213175131824702,
      "learning_rate": 6.783680279702751e-06,
      "loss": 0.0254,
      "step": 4984
    },
    {
      "epoch": 1.3368195226602306,
      "grad_norm": 1.5219608260964959,
      "learning_rate": 6.782222658005088e-06,
      "loss": 0.034,
      "step": 4985
    },
    {
      "epoch": 1.337087691069992,
      "grad_norm": 0.28192536345419233,
      "learning_rate": 6.780764862777134e-06,
      "loss": 0.0205,
      "step": 4986
    },
    {
      "epoch": 1.3373558594797532,
      "grad_norm": 0.2576976473293279,
      "learning_rate": 6.779306894160834e-06,
      "loss": 0.0298,
      "step": 4987
    },
    {
      "epoch": 1.3376240278895146,
      "grad_norm": 0.2448813601073709,
      "learning_rate": 6.777848752298146e-06,
      "loss": 0.0213,
      "step": 4988
    },
    {
      "epoch": 1.337892196299276,
      "grad_norm": 0.21347433366229293,
      "learning_rate": 6.776390437331047e-06,
      "loss": 0.0218,
      "step": 4989
    },
    {
      "epoch": 1.3381603647090372,
      "grad_norm": 0.2865009017799368,
      "learning_rate": 6.774931949401524e-06,
      "loss": 0.0274,
      "step": 4990
    },
    {
      "epoch": 1.3384285331187986,
      "grad_norm": 0.3584167033879194,
      "learning_rate": 6.773473288651591e-06,
      "loss": 0.029,
      "step": 4991
    },
    {
      "epoch": 1.33869670152856,
      "grad_norm": 0.3428683375564989,
      "learning_rate": 6.772014455223271e-06,
      "loss": 0.0189,
      "step": 4992
    },
    {
      "epoch": 1.3389648699383212,
      "grad_norm": 0.43496325510121936,
      "learning_rate": 6.770555449258609e-06,
      "loss": 0.0308,
      "step": 4993
    },
    {
      "epoch": 1.3392330383480826,
      "grad_norm": 0.3300736792516876,
      "learning_rate": 6.7690962708996645e-06,
      "loss": 0.0319,
      "step": 4994
    },
    {
      "epoch": 1.339501206757844,
      "grad_norm": 0.41014148844154846,
      "learning_rate": 6.767636920288514e-06,
      "loss": 0.0275,
      "step": 4995
    },
    {
      "epoch": 1.3397693751676052,
      "grad_norm": 0.2818767179010466,
      "learning_rate": 6.766177397567246e-06,
      "loss": 0.0227,
      "step": 4996
    },
    {
      "epoch": 1.3400375435773666,
      "grad_norm": 0.5531159672986595,
      "learning_rate": 6.764717702877977e-06,
      "loss": 0.0307,
      "step": 4997
    },
    {
      "epoch": 1.3403057119871278,
      "grad_norm": 0.28902114755061614,
      "learning_rate": 6.763257836362829e-06,
      "loss": 0.0293,
      "step": 4998
    },
    {
      "epoch": 1.3405738803968892,
      "grad_norm": 0.3377310159301404,
      "learning_rate": 6.7617977981639475e-06,
      "loss": 0.0251,
      "step": 4999
    },
    {
      "epoch": 1.3408420488066506,
      "grad_norm": 0.25864389816161665,
      "learning_rate": 6.760337588423491e-06,
      "loss": 0.0255,
      "step": 5000
    },
    {
      "epoch": 1.341110217216412,
      "grad_norm": 0.2794139874470606,
      "learning_rate": 6.758877207283638e-06,
      "loss": 0.0233,
      "step": 5001
    },
    {
      "epoch": 1.3413783856261732,
      "grad_norm": 0.27218271806755717,
      "learning_rate": 6.7574166548865804e-06,
      "loss": 0.0228,
      "step": 5002
    },
    {
      "epoch": 1.3416465540359346,
      "grad_norm": 0.2867806012271734,
      "learning_rate": 6.75595593137453e-06,
      "loss": 0.0198,
      "step": 5003
    },
    {
      "epoch": 1.3419147224456958,
      "grad_norm": 0.2133792441275049,
      "learning_rate": 6.754495036889711e-06,
      "loss": 0.0226,
      "step": 5004
    },
    {
      "epoch": 1.3421828908554572,
      "grad_norm": 0.2932134827669461,
      "learning_rate": 6.75303397157437e-06,
      "loss": 0.029,
      "step": 5005
    },
    {
      "epoch": 1.3424510592652186,
      "grad_norm": 0.28411200108564105,
      "learning_rate": 6.751572735570764e-06,
      "loss": 0.0238,
      "step": 5006
    },
    {
      "epoch": 1.34271922767498,
      "grad_norm": 0.34661277051320044,
      "learning_rate": 6.7501113290211715e-06,
      "loss": 0.0329,
      "step": 5007
    },
    {
      "epoch": 1.3429873960847412,
      "grad_norm": 0.386897762415121,
      "learning_rate": 6.7486497520678864e-06,
      "loss": 0.035,
      "step": 5008
    },
    {
      "epoch": 1.3432555644945026,
      "grad_norm": 0.23075269916957455,
      "learning_rate": 6.747188004853216e-06,
      "loss": 0.0274,
      "step": 5009
    },
    {
      "epoch": 1.3435237329042637,
      "grad_norm": 0.2465671540890511,
      "learning_rate": 6.745726087519489e-06,
      "loss": 0.0278,
      "step": 5010
    },
    {
      "epoch": 1.3437919013140252,
      "grad_norm": 0.30167976493206206,
      "learning_rate": 6.7442640002090475e-06,
      "loss": 0.027,
      "step": 5011
    },
    {
      "epoch": 1.3440600697237866,
      "grad_norm": 0.5502437565311378,
      "learning_rate": 6.742801743064251e-06,
      "loss": 0.0273,
      "step": 5012
    },
    {
      "epoch": 1.344328238133548,
      "grad_norm": 0.3252126651516387,
      "learning_rate": 6.741339316227477e-06,
      "loss": 0.0225,
      "step": 5013
    },
    {
      "epoch": 1.3445964065433091,
      "grad_norm": 0.2553623964033162,
      "learning_rate": 6.739876719841115e-06,
      "loss": 0.0194,
      "step": 5014
    },
    {
      "epoch": 1.3448645749530705,
      "grad_norm": 0.38137221270285687,
      "learning_rate": 6.738413954047578e-06,
      "loss": 0.0277,
      "step": 5015
    },
    {
      "epoch": 1.3451327433628317,
      "grad_norm": 0.4981605090645751,
      "learning_rate": 6.736951018989289e-06,
      "loss": 0.0273,
      "step": 5016
    },
    {
      "epoch": 1.3454009117725931,
      "grad_norm": 0.27367997014703416,
      "learning_rate": 6.7354879148086904e-06,
      "loss": 0.0255,
      "step": 5017
    },
    {
      "epoch": 1.3456690801823545,
      "grad_norm": 0.3641032804386525,
      "learning_rate": 6.73402464164824e-06,
      "loss": 0.023,
      "step": 5018
    },
    {
      "epoch": 1.345937248592116,
      "grad_norm": 0.3156356141243637,
      "learning_rate": 6.732561199650417e-06,
      "loss": 0.034,
      "step": 5019
    },
    {
      "epoch": 1.3462054170018771,
      "grad_norm": 0.2805549338188656,
      "learning_rate": 6.731097588957708e-06,
      "loss": 0.0208,
      "step": 5020
    },
    {
      "epoch": 1.3464735854116385,
      "grad_norm": 0.19540225284971263,
      "learning_rate": 6.729633809712623e-06,
      "loss": 0.0157,
      "step": 5021
    },
    {
      "epoch": 1.3467417538213997,
      "grad_norm": 0.25919978245756614,
      "learning_rate": 6.728169862057687e-06,
      "loss": 0.0233,
      "step": 5022
    },
    {
      "epoch": 1.3470099222311611,
      "grad_norm": 0.2887090582260099,
      "learning_rate": 6.726705746135438e-06,
      "loss": 0.0298,
      "step": 5023
    },
    {
      "epoch": 1.3472780906409225,
      "grad_norm": 0.3042642899693252,
      "learning_rate": 6.725241462088434e-06,
      "loss": 0.0306,
      "step": 5024
    },
    {
      "epoch": 1.347546259050684,
      "grad_norm": 0.29991265501159897,
      "learning_rate": 6.723777010059253e-06,
      "loss": 0.022,
      "step": 5025
    },
    {
      "epoch": 1.347814427460445,
      "grad_norm": 0.2394799332537473,
      "learning_rate": 6.7223123901904785e-06,
      "loss": 0.0212,
      "step": 5026
    },
    {
      "epoch": 1.3480825958702065,
      "grad_norm": 0.24444992425515327,
      "learning_rate": 6.72084760262472e-06,
      "loss": 0.02,
      "step": 5027
    },
    {
      "epoch": 1.3483507642799677,
      "grad_norm": 0.3206941383547955,
      "learning_rate": 6.719382647504598e-06,
      "loss": 0.0274,
      "step": 5028
    },
    {
      "epoch": 1.348618932689729,
      "grad_norm": 0.35043114424613836,
      "learning_rate": 6.717917524972752e-06,
      "loss": 0.0359,
      "step": 5029
    },
    {
      "epoch": 1.3488871010994905,
      "grad_norm": 0.3900173049273321,
      "learning_rate": 6.71645223517184e-06,
      "loss": 0.0314,
      "step": 5030
    },
    {
      "epoch": 1.349155269509252,
      "grad_norm": 0.23186204819174158,
      "learning_rate": 6.71498677824453e-06,
      "loss": 0.0245,
      "step": 5031
    },
    {
      "epoch": 1.349423437919013,
      "grad_norm": 0.302822575564756,
      "learning_rate": 6.713521154333511e-06,
      "loss": 0.0302,
      "step": 5032
    },
    {
      "epoch": 1.3496916063287745,
      "grad_norm": 0.3158300891581313,
      "learning_rate": 6.712055363581485e-06,
      "loss": 0.0291,
      "step": 5033
    },
    {
      "epoch": 1.3499597747385357,
      "grad_norm": 0.30361361557921546,
      "learning_rate": 6.710589406131174e-06,
      "loss": 0.019,
      "step": 5034
    },
    {
      "epoch": 1.350227943148297,
      "grad_norm": 0.28954453278253367,
      "learning_rate": 6.7091232821253145e-06,
      "loss": 0.0244,
      "step": 5035
    },
    {
      "epoch": 1.3504961115580585,
      "grad_norm": 0.20660424090738022,
      "learning_rate": 6.707656991706659e-06,
      "loss": 0.0234,
      "step": 5036
    },
    {
      "epoch": 1.35076427996782,
      "grad_norm": 0.28698718794474076,
      "learning_rate": 6.706190535017977e-06,
      "loss": 0.023,
      "step": 5037
    },
    {
      "epoch": 1.351032448377581,
      "grad_norm": 0.49103456427724895,
      "learning_rate": 6.704723912202051e-06,
      "loss": 0.0317,
      "step": 5038
    },
    {
      "epoch": 1.3513006167873425,
      "grad_norm": 0.2869390023519569,
      "learning_rate": 6.703257123401683e-06,
      "loss": 0.0209,
      "step": 5039
    },
    {
      "epoch": 1.3515687851971037,
      "grad_norm": 0.28363044008807875,
      "learning_rate": 6.701790168759692e-06,
      "loss": 0.0226,
      "step": 5040
    },
    {
      "epoch": 1.351836953606865,
      "grad_norm": 0.2972507729032398,
      "learning_rate": 6.700323048418911e-06,
      "loss": 0.024,
      "step": 5041
    },
    {
      "epoch": 1.3521051220166265,
      "grad_norm": 0.6309358069669206,
      "learning_rate": 6.698855762522187e-06,
      "loss": 0.0296,
      "step": 5042
    },
    {
      "epoch": 1.3523732904263879,
      "grad_norm": 0.32758714252897797,
      "learning_rate": 6.69738831121239e-06,
      "loss": 0.0214,
      "step": 5043
    },
    {
      "epoch": 1.352641458836149,
      "grad_norm": 0.27862056562215365,
      "learning_rate": 6.695920694632399e-06,
      "loss": 0.0317,
      "step": 5044
    },
    {
      "epoch": 1.3529096272459105,
      "grad_norm": 0.2583153668257148,
      "learning_rate": 6.694452912925111e-06,
      "loss": 0.0223,
      "step": 5045
    },
    {
      "epoch": 1.3531777956556716,
      "grad_norm": 0.36701797301508854,
      "learning_rate": 6.692984966233443e-06,
      "loss": 0.0228,
      "step": 5046
    },
    {
      "epoch": 1.353445964065433,
      "grad_norm": 0.2713090792345699,
      "learning_rate": 6.691516854700324e-06,
      "loss": 0.0191,
      "step": 5047
    },
    {
      "epoch": 1.3537141324751945,
      "grad_norm": 0.2614201343082274,
      "learning_rate": 6.690048578468698e-06,
      "loss": 0.0187,
      "step": 5048
    },
    {
      "epoch": 1.3539823008849559,
      "grad_norm": 0.29326486334863117,
      "learning_rate": 6.68858013768153e-06,
      "loss": 0.0326,
      "step": 5049
    },
    {
      "epoch": 1.354250469294717,
      "grad_norm": 0.2952834898180767,
      "learning_rate": 6.687111532481798e-06,
      "loss": 0.0272,
      "step": 5050
    },
    {
      "epoch": 1.3545186377044784,
      "grad_norm": 0.27522460560402734,
      "learning_rate": 6.685642763012494e-06,
      "loss": 0.0196,
      "step": 5051
    },
    {
      "epoch": 1.3547868061142396,
      "grad_norm": 0.23217384018522005,
      "learning_rate": 6.68417382941663e-06,
      "loss": 0.0209,
      "step": 5052
    },
    {
      "epoch": 1.355054974524001,
      "grad_norm": 0.25616569109118725,
      "learning_rate": 6.682704731837232e-06,
      "loss": 0.017,
      "step": 5053
    },
    {
      "epoch": 1.3553231429337624,
      "grad_norm": 0.2546297647693074,
      "learning_rate": 6.681235470417341e-06,
      "loss": 0.0217,
      "step": 5054
    },
    {
      "epoch": 1.3555913113435238,
      "grad_norm": 0.3291739391909463,
      "learning_rate": 6.679766045300017e-06,
      "loss": 0.024,
      "step": 5055
    },
    {
      "epoch": 1.355859479753285,
      "grad_norm": 0.34111097541759494,
      "learning_rate": 6.6782964566283315e-06,
      "loss": 0.0297,
      "step": 5056
    },
    {
      "epoch": 1.3561276481630464,
      "grad_norm": 0.3192875922138327,
      "learning_rate": 6.676826704545376e-06,
      "loss": 0.0171,
      "step": 5057
    },
    {
      "epoch": 1.3563958165728076,
      "grad_norm": 0.3534537385551134,
      "learning_rate": 6.675356789194256e-06,
      "loss": 0.0327,
      "step": 5058
    },
    {
      "epoch": 1.356663984982569,
      "grad_norm": 0.3061867979262889,
      "learning_rate": 6.673886710718095e-06,
      "loss": 0.0268,
      "step": 5059
    },
    {
      "epoch": 1.3569321533923304,
      "grad_norm": 0.30448102013977585,
      "learning_rate": 6.672416469260028e-06,
      "loss": 0.0341,
      "step": 5060
    },
    {
      "epoch": 1.3572003218020918,
      "grad_norm": 0.28534281318190696,
      "learning_rate": 6.670946064963209e-06,
      "loss": 0.0238,
      "step": 5061
    },
    {
      "epoch": 1.357468490211853,
      "grad_norm": 0.21015625975559304,
      "learning_rate": 6.669475497970809e-06,
      "loss": 0.0223,
      "step": 5062
    },
    {
      "epoch": 1.3577366586216144,
      "grad_norm": 0.31317725447923994,
      "learning_rate": 6.668004768426011e-06,
      "loss": 0.0273,
      "step": 5063
    },
    {
      "epoch": 1.3580048270313756,
      "grad_norm": 0.2362236238183627,
      "learning_rate": 6.666533876472018e-06,
      "loss": 0.0254,
      "step": 5064
    },
    {
      "epoch": 1.358272995441137,
      "grad_norm": 0.24699963504566408,
      "learning_rate": 6.665062822252046e-06,
      "loss": 0.024,
      "step": 5065
    },
    {
      "epoch": 1.3585411638508984,
      "grad_norm": 0.326434897125347,
      "learning_rate": 6.663591605909325e-06,
      "loss": 0.0327,
      "step": 5066
    },
    {
      "epoch": 1.3588093322606598,
      "grad_norm": 0.19945791465968243,
      "learning_rate": 6.66212022758711e-06,
      "loss": 0.0293,
      "step": 5067
    },
    {
      "epoch": 1.359077500670421,
      "grad_norm": 0.2708259302081907,
      "learning_rate": 6.66064868742866e-06,
      "loss": 0.0224,
      "step": 5068
    },
    {
      "epoch": 1.3593456690801824,
      "grad_norm": 0.21620411722317956,
      "learning_rate": 6.659176985577256e-06,
      "loss": 0.0229,
      "step": 5069
    },
    {
      "epoch": 1.3596138374899436,
      "grad_norm": 0.27788741044009674,
      "learning_rate": 6.657705122176194e-06,
      "loss": 0.0241,
      "step": 5070
    },
    {
      "epoch": 1.359882005899705,
      "grad_norm": 0.18041673717614787,
      "learning_rate": 6.656233097368785e-06,
      "loss": 0.0203,
      "step": 5071
    },
    {
      "epoch": 1.3601501743094664,
      "grad_norm": 0.2991962375235837,
      "learning_rate": 6.654760911298357e-06,
      "loss": 0.0237,
      "step": 5072
    },
    {
      "epoch": 1.3604183427192278,
      "grad_norm": 0.24732961148700947,
      "learning_rate": 6.653288564108252e-06,
      "loss": 0.0354,
      "step": 5073
    },
    {
      "epoch": 1.360686511128989,
      "grad_norm": 0.2259752050825254,
      "learning_rate": 6.65181605594183e-06,
      "loss": 0.0218,
      "step": 5074
    },
    {
      "epoch": 1.3609546795387504,
      "grad_norm": 0.4170065395801382,
      "learning_rate": 6.650343386942462e-06,
      "loss": 0.0349,
      "step": 5075
    },
    {
      "epoch": 1.3612228479485116,
      "grad_norm": 0.30123259997765317,
      "learning_rate": 6.648870557253541e-06,
      "loss": 0.0344,
      "step": 5076
    },
    {
      "epoch": 1.361491016358273,
      "grad_norm": 0.2482891797060849,
      "learning_rate": 6.647397567018472e-06,
      "loss": 0.0203,
      "step": 5077
    },
    {
      "epoch": 1.3617591847680344,
      "grad_norm": 0.30567748975754927,
      "learning_rate": 6.645924416380676e-06,
      "loss": 0.0317,
      "step": 5078
    },
    {
      "epoch": 1.3620273531777958,
      "grad_norm": 0.24767753062552794,
      "learning_rate": 6.644451105483588e-06,
      "loss": 0.0218,
      "step": 5079
    },
    {
      "epoch": 1.362295521587557,
      "grad_norm": 0.23412280814432446,
      "learning_rate": 6.642977634470665e-06,
      "loss": 0.0204,
      "step": 5080
    },
    {
      "epoch": 1.3625636899973184,
      "grad_norm": 0.21496519906965245,
      "learning_rate": 6.641504003485369e-06,
      "loss": 0.0216,
      "step": 5081
    },
    {
      "epoch": 1.3628318584070795,
      "grad_norm": 0.24889239029486848,
      "learning_rate": 6.6400302126711845e-06,
      "loss": 0.0187,
      "step": 5082
    },
    {
      "epoch": 1.363100026816841,
      "grad_norm": 0.40078029198581455,
      "learning_rate": 6.638556262171615e-06,
      "loss": 0.0224,
      "step": 5083
    },
    {
      "epoch": 1.3633681952266024,
      "grad_norm": 0.2905008394101974,
      "learning_rate": 6.637082152130172e-06,
      "loss": 0.0239,
      "step": 5084
    },
    {
      "epoch": 1.3636363636363638,
      "grad_norm": 0.31030734336727234,
      "learning_rate": 6.635607882690387e-06,
      "loss": 0.0273,
      "step": 5085
    },
    {
      "epoch": 1.363904532046125,
      "grad_norm": 0.34070729252319876,
      "learning_rate": 6.634133453995805e-06,
      "loss": 0.024,
      "step": 5086
    },
    {
      "epoch": 1.3641727004558863,
      "grad_norm": 0.28896385082043063,
      "learning_rate": 6.632658866189985e-06,
      "loss": 0.0213,
      "step": 5087
    },
    {
      "epoch": 1.3644408688656475,
      "grad_norm": 0.23465225002551524,
      "learning_rate": 6.631184119416507e-06,
      "loss": 0.0193,
      "step": 5088
    },
    {
      "epoch": 1.364709037275409,
      "grad_norm": 0.34667035711358335,
      "learning_rate": 6.629709213818962e-06,
      "loss": 0.0349,
      "step": 5089
    },
    {
      "epoch": 1.3649772056851703,
      "grad_norm": 0.21791769530711644,
      "learning_rate": 6.628234149540957e-06,
      "loss": 0.0216,
      "step": 5090
    },
    {
      "epoch": 1.3652453740949317,
      "grad_norm": 0.2739570789065874,
      "learning_rate": 6.626758926726118e-06,
      "loss": 0.0275,
      "step": 5091
    },
    {
      "epoch": 1.365513542504693,
      "grad_norm": 0.21960668466380617,
      "learning_rate": 6.62528354551808e-06,
      "loss": 0.0209,
      "step": 5092
    },
    {
      "epoch": 1.3657817109144543,
      "grad_norm": 0.21993935213671614,
      "learning_rate": 6.623808006060497e-06,
      "loss": 0.0189,
      "step": 5093
    },
    {
      "epoch": 1.3660498793242155,
      "grad_norm": 0.2758263814019537,
      "learning_rate": 6.6223323084970415e-06,
      "loss": 0.0207,
      "step": 5094
    },
    {
      "epoch": 1.366318047733977,
      "grad_norm": 0.2284052009677256,
      "learning_rate": 6.620856452971395e-06,
      "loss": 0.0194,
      "step": 5095
    },
    {
      "epoch": 1.3665862161437383,
      "grad_norm": 0.2812677323751444,
      "learning_rate": 6.61938043962726e-06,
      "loss": 0.0273,
      "step": 5096
    },
    {
      "epoch": 1.3668543845534997,
      "grad_norm": 0.3668572195161334,
      "learning_rate": 6.6179042686083505e-06,
      "loss": 0.0215,
      "step": 5097
    },
    {
      "epoch": 1.367122552963261,
      "grad_norm": 0.4838162027145904,
      "learning_rate": 6.616427940058398e-06,
      "loss": 0.0258,
      "step": 5098
    },
    {
      "epoch": 1.3673907213730223,
      "grad_norm": 0.3505781867486169,
      "learning_rate": 6.614951454121148e-06,
      "loss": 0.0286,
      "step": 5099
    },
    {
      "epoch": 1.3676588897827835,
      "grad_norm": 0.28022479874431183,
      "learning_rate": 6.613474810940363e-06,
      "loss": 0.0252,
      "step": 5100
    },
    {
      "epoch": 1.367927058192545,
      "grad_norm": 0.6008600469381709,
      "learning_rate": 6.611998010659818e-06,
      "loss": 0.0271,
      "step": 5101
    },
    {
      "epoch": 1.3681952266023063,
      "grad_norm": 0.2967687825135098,
      "learning_rate": 6.610521053423308e-06,
      "loss": 0.0296,
      "step": 5102
    },
    {
      "epoch": 1.3684633950120677,
      "grad_norm": 0.21958961292984855,
      "learning_rate": 6.609043939374638e-06,
      "loss": 0.0204,
      "step": 5103
    },
    {
      "epoch": 1.368731563421829,
      "grad_norm": 0.3159632223460224,
      "learning_rate": 6.607566668657632e-06,
      "loss": 0.024,
      "step": 5104
    },
    {
      "epoch": 1.3689997318315903,
      "grad_norm": 0.30005556663436195,
      "learning_rate": 6.606089241416127e-06,
      "loss": 0.0271,
      "step": 5105
    },
    {
      "epoch": 1.3692679002413515,
      "grad_norm": 0.25760099062866715,
      "learning_rate": 6.604611657793978e-06,
      "loss": 0.0235,
      "step": 5106
    },
    {
      "epoch": 1.3695360686511129,
      "grad_norm": 0.3204697494909888,
      "learning_rate": 6.60313391793505e-06,
      "loss": 0.0295,
      "step": 5107
    },
    {
      "epoch": 1.3698042370608743,
      "grad_norm": 0.2727651145898471,
      "learning_rate": 6.60165602198323e-06,
      "loss": 0.029,
      "step": 5108
    },
    {
      "epoch": 1.3700724054706357,
      "grad_norm": 0.20786565898512394,
      "learning_rate": 6.600177970082416e-06,
      "loss": 0.0198,
      "step": 5109
    },
    {
      "epoch": 1.3703405738803969,
      "grad_norm": 0.2700042646917856,
      "learning_rate": 6.59869976237652e-06,
      "loss": 0.0184,
      "step": 5110
    },
    {
      "epoch": 1.3706087422901583,
      "grad_norm": 0.3866713630628548,
      "learning_rate": 6.597221399009474e-06,
      "loss": 0.0252,
      "step": 5111
    },
    {
      "epoch": 1.3708769106999195,
      "grad_norm": 0.28361676049979284,
      "learning_rate": 6.59574288012522e-06,
      "loss": 0.0222,
      "step": 5112
    },
    {
      "epoch": 1.3711450791096809,
      "grad_norm": 0.3996774438131774,
      "learning_rate": 6.5942642058677185e-06,
      "loss": 0.0331,
      "step": 5113
    },
    {
      "epoch": 1.3714132475194423,
      "grad_norm": 0.2227104512568194,
      "learning_rate": 6.592785376380944e-06,
      "loss": 0.0169,
      "step": 5114
    },
    {
      "epoch": 1.3716814159292037,
      "grad_norm": 0.7584157525301737,
      "learning_rate": 6.591306391808886e-06,
      "loss": 0.0286,
      "step": 5115
    },
    {
      "epoch": 1.3719495843389649,
      "grad_norm": 0.5069549852650557,
      "learning_rate": 6.589827252295551e-06,
      "loss": 0.0201,
      "step": 5116
    },
    {
      "epoch": 1.3722177527487263,
      "grad_norm": 0.28620843722007233,
      "learning_rate": 6.588347957984955e-06,
      "loss": 0.0248,
      "step": 5117
    },
    {
      "epoch": 1.3724859211584874,
      "grad_norm": 0.6020014541029284,
      "learning_rate": 6.586868509021137e-06,
      "loss": 0.0248,
      "step": 5118
    },
    {
      "epoch": 1.3727540895682488,
      "grad_norm": 0.2866972865999652,
      "learning_rate": 6.585388905548144e-06,
      "loss": 0.025,
      "step": 5119
    },
    {
      "epoch": 1.3730222579780103,
      "grad_norm": 0.3861905887095278,
      "learning_rate": 6.583909147710042e-06,
      "loss": 0.0209,
      "step": 5120
    },
    {
      "epoch": 1.3732904263877717,
      "grad_norm": 0.2416065141345552,
      "learning_rate": 6.582429235650914e-06,
      "loss": 0.0208,
      "step": 5121
    },
    {
      "epoch": 1.3735585947975328,
      "grad_norm": 0.2825162122475292,
      "learning_rate": 6.580949169514852e-06,
      "loss": 0.0254,
      "step": 5122
    },
    {
      "epoch": 1.3738267632072942,
      "grad_norm": 0.23836590871192379,
      "learning_rate": 6.579468949445965e-06,
      "loss": 0.0176,
      "step": 5123
    },
    {
      "epoch": 1.3740949316170554,
      "grad_norm": 0.2217462910104824,
      "learning_rate": 6.577988575588381e-06,
      "loss": 0.0201,
      "step": 5124
    },
    {
      "epoch": 1.3743631000268168,
      "grad_norm": 0.3829205152823053,
      "learning_rate": 6.576508048086237e-06,
      "loss": 0.0274,
      "step": 5125
    },
    {
      "epoch": 1.3746312684365782,
      "grad_norm": 0.2668024941088268,
      "learning_rate": 6.5750273670836915e-06,
      "loss": 0.0182,
      "step": 5126
    },
    {
      "epoch": 1.3748994368463396,
      "grad_norm": 0.33111035300024017,
      "learning_rate": 6.5735465327249125e-06,
      "loss": 0.0228,
      "step": 5127
    },
    {
      "epoch": 1.3751676052561008,
      "grad_norm": 0.37058450124543657,
      "learning_rate": 6.572065545154087e-06,
      "loss": 0.0366,
      "step": 5128
    },
    {
      "epoch": 1.3754357736658622,
      "grad_norm": 0.3549421173434805,
      "learning_rate": 6.570584404515412e-06,
      "loss": 0.0208,
      "step": 5129
    },
    {
      "epoch": 1.3757039420756234,
      "grad_norm": 0.2699091902228763,
      "learning_rate": 6.569103110953103e-06,
      "loss": 0.0197,
      "step": 5130
    },
    {
      "epoch": 1.3759721104853848,
      "grad_norm": 0.3762589093209011,
      "learning_rate": 6.567621664611389e-06,
      "loss": 0.0316,
      "step": 5131
    },
    {
      "epoch": 1.3762402788951462,
      "grad_norm": 0.2351667055258288,
      "learning_rate": 6.566140065634516e-06,
      "loss": 0.0243,
      "step": 5132
    },
    {
      "epoch": 1.3765084473049074,
      "grad_norm": 0.2893426605052858,
      "learning_rate": 6.5646583141667446e-06,
      "loss": 0.0259,
      "step": 5133
    },
    {
      "epoch": 1.3767766157146688,
      "grad_norm": 0.2701150698004994,
      "learning_rate": 6.563176410352347e-06,
      "loss": 0.0277,
      "step": 5134
    },
    {
      "epoch": 1.3770447841244302,
      "grad_norm": 0.3352044360474642,
      "learning_rate": 6.56169435433561e-06,
      "loss": 0.0279,
      "step": 5135
    },
    {
      "epoch": 1.3773129525341914,
      "grad_norm": 0.29523620558317343,
      "learning_rate": 6.560212146260841e-06,
      "loss": 0.0271,
      "step": 5136
    },
    {
      "epoch": 1.3775811209439528,
      "grad_norm": 0.49138982849964424,
      "learning_rate": 6.558729786272359e-06,
      "loss": 0.0333,
      "step": 5137
    },
    {
      "epoch": 1.3778492893537142,
      "grad_norm": 0.33503344411027447,
      "learning_rate": 6.5572472745144944e-06,
      "loss": 0.0275,
      "step": 5138
    },
    {
      "epoch": 1.3781174577634754,
      "grad_norm": 0.2416364582209329,
      "learning_rate": 6.555764611131599e-06,
      "loss": 0.0135,
      "step": 5139
    },
    {
      "epoch": 1.3783856261732368,
      "grad_norm": 0.26789059717793146,
      "learning_rate": 6.554281796268031e-06,
      "loss": 0.0325,
      "step": 5140
    },
    {
      "epoch": 1.3786537945829982,
      "grad_norm": 0.32317284330765095,
      "learning_rate": 6.552798830068173e-06,
      "loss": 0.0192,
      "step": 5141
    },
    {
      "epoch": 1.3789219629927594,
      "grad_norm": 0.35869690065737037,
      "learning_rate": 6.551315712676416e-06,
      "loss": 0.0353,
      "step": 5142
    },
    {
      "epoch": 1.3791901314025208,
      "grad_norm": 0.36553281445631164,
      "learning_rate": 6.549832444237165e-06,
      "loss": 0.0287,
      "step": 5143
    },
    {
      "epoch": 1.3794582998122822,
      "grad_norm": 0.30364053577404354,
      "learning_rate": 6.548349024894846e-06,
      "loss": 0.03,
      "step": 5144
    },
    {
      "epoch": 1.3797264682220434,
      "grad_norm": 0.30298155570238244,
      "learning_rate": 6.546865454793891e-06,
      "loss": 0.0256,
      "step": 5145
    },
    {
      "epoch": 1.3799946366318048,
      "grad_norm": 0.36046815855821246,
      "learning_rate": 6.545381734078754e-06,
      "loss": 0.0299,
      "step": 5146
    },
    {
      "epoch": 1.3802628050415662,
      "grad_norm": 0.27389916574120127,
      "learning_rate": 6.543897862893901e-06,
      "loss": 0.027,
      "step": 5147
    },
    {
      "epoch": 1.3805309734513274,
      "grad_norm": 0.25778395382622,
      "learning_rate": 6.542413841383813e-06,
      "loss": 0.0269,
      "step": 5148
    },
    {
      "epoch": 1.3807991418610888,
      "grad_norm": 0.316177238561347,
      "learning_rate": 6.5409296696929825e-06,
      "loss": 0.0355,
      "step": 5149
    },
    {
      "epoch": 1.3810673102708502,
      "grad_norm": 0.2877127953208787,
      "learning_rate": 6.5394453479659215e-06,
      "loss": 0.0178,
      "step": 5150
    },
    {
      "epoch": 1.3813354786806114,
      "grad_norm": 0.23668135190631837,
      "learning_rate": 6.537960876347155e-06,
      "loss": 0.0171,
      "step": 5151
    },
    {
      "epoch": 1.3816036470903728,
      "grad_norm": 0.21441462891255378,
      "learning_rate": 6.5364762549812195e-06,
      "loss": 0.0239,
      "step": 5152
    },
    {
      "epoch": 1.3818718155001342,
      "grad_norm": 0.30021628193401917,
      "learning_rate": 6.534991484012672e-06,
      "loss": 0.0266,
      "step": 5153
    },
    {
      "epoch": 1.3821399839098953,
      "grad_norm": 0.2585604144668049,
      "learning_rate": 6.533506563586079e-06,
      "loss": 0.0199,
      "step": 5154
    },
    {
      "epoch": 1.3824081523196567,
      "grad_norm": 0.4001361624400113,
      "learning_rate": 6.532021493846021e-06,
      "loss": 0.0302,
      "step": 5155
    },
    {
      "epoch": 1.3826763207294182,
      "grad_norm": 0.260373697212655,
      "learning_rate": 6.5305362749371e-06,
      "loss": 0.022,
      "step": 5156
    },
    {
      "epoch": 1.3829444891391793,
      "grad_norm": 0.37664659813509344,
      "learning_rate": 6.529050907003924e-06,
      "loss": 0.0255,
      "step": 5157
    },
    {
      "epoch": 1.3832126575489407,
      "grad_norm": 0.2806128700654724,
      "learning_rate": 6.527565390191122e-06,
      "loss": 0.0211,
      "step": 5158
    },
    {
      "epoch": 1.3834808259587021,
      "grad_norm": 0.20592256332628703,
      "learning_rate": 6.526079724643334e-06,
      "loss": 0.0168,
      "step": 5159
    },
    {
      "epoch": 1.3837489943684633,
      "grad_norm": 0.26195819425124944,
      "learning_rate": 6.524593910505214e-06,
      "loss": 0.0249,
      "step": 5160
    },
    {
      "epoch": 1.3840171627782247,
      "grad_norm": 0.32691775010066143,
      "learning_rate": 6.5231079479214325e-06,
      "loss": 0.0345,
      "step": 5161
    },
    {
      "epoch": 1.3842853311879861,
      "grad_norm": 0.21849658606767608,
      "learning_rate": 6.521621837036672e-06,
      "loss": 0.019,
      "step": 5162
    },
    {
      "epoch": 1.3845534995977473,
      "grad_norm": 0.34732717083683623,
      "learning_rate": 6.520135577995636e-06,
      "loss": 0.0373,
      "step": 5163
    },
    {
      "epoch": 1.3848216680075087,
      "grad_norm": 0.19546418682449748,
      "learning_rate": 6.518649170943035e-06,
      "loss": 0.015,
      "step": 5164
    },
    {
      "epoch": 1.3850898364172701,
      "grad_norm": 0.23112305909493605,
      "learning_rate": 6.517162616023593e-06,
      "loss": 0.0181,
      "step": 5165
    },
    {
      "epoch": 1.3853580048270313,
      "grad_norm": 0.23798546114377456,
      "learning_rate": 6.515675913382058e-06,
      "loss": 0.0218,
      "step": 5166
    },
    {
      "epoch": 1.3856261732367927,
      "grad_norm": 0.45282538591777605,
      "learning_rate": 6.514189063163182e-06,
      "loss": 0.0271,
      "step": 5167
    },
    {
      "epoch": 1.3858943416465541,
      "grad_norm": 0.2970928429852598,
      "learning_rate": 6.5127020655117355e-06,
      "loss": 0.0313,
      "step": 5168
    },
    {
      "epoch": 1.3861625100563153,
      "grad_norm": 0.23505047994155392,
      "learning_rate": 6.511214920572507e-06,
      "loss": 0.0202,
      "step": 5169
    },
    {
      "epoch": 1.3864306784660767,
      "grad_norm": 0.47194009371941376,
      "learning_rate": 6.509727628490294e-06,
      "loss": 0.0195,
      "step": 5170
    },
    {
      "epoch": 1.386698846875838,
      "grad_norm": 0.27713698068059905,
      "learning_rate": 6.508240189409911e-06,
      "loss": 0.0249,
      "step": 5171
    },
    {
      "epoch": 1.3869670152855993,
      "grad_norm": 0.19444911249917674,
      "learning_rate": 6.506752603476183e-06,
      "loss": 0.0175,
      "step": 5172
    },
    {
      "epoch": 1.3872351836953607,
      "grad_norm": 0.3400962721634422,
      "learning_rate": 6.505264870833954e-06,
      "loss": 0.0274,
      "step": 5173
    },
    {
      "epoch": 1.387503352105122,
      "grad_norm": 0.2822669872235687,
      "learning_rate": 6.503776991628082e-06,
      "loss": 0.0191,
      "step": 5174
    },
    {
      "epoch": 1.3877715205148833,
      "grad_norm": 0.22144025490236657,
      "learning_rate": 6.502288966003437e-06,
      "loss": 0.0156,
      "step": 5175
    },
    {
      "epoch": 1.3880396889246447,
      "grad_norm": 0.2796820310291233,
      "learning_rate": 6.500800794104905e-06,
      "loss": 0.0199,
      "step": 5176
    },
    {
      "epoch": 1.3883078573344059,
      "grad_norm": 0.3575901827382534,
      "learning_rate": 6.499312476077383e-06,
      "loss": 0.0279,
      "step": 5177
    },
    {
      "epoch": 1.3885760257441673,
      "grad_norm": 0.35267346214958295,
      "learning_rate": 6.497824012065784e-06,
      "loss": 0.0326,
      "step": 5178
    },
    {
      "epoch": 1.3888441941539287,
      "grad_norm": 0.28469498710799546,
      "learning_rate": 6.496335402215039e-06,
      "loss": 0.0278,
      "step": 5179
    },
    {
      "epoch": 1.38911236256369,
      "grad_norm": 0.32922009994379825,
      "learning_rate": 6.4948466466700895e-06,
      "loss": 0.0253,
      "step": 5180
    },
    {
      "epoch": 1.3893805309734513,
      "grad_norm": 0.43433414287820016,
      "learning_rate": 6.49335774557589e-06,
      "loss": 0.0241,
      "step": 5181
    },
    {
      "epoch": 1.3896486993832127,
      "grad_norm": 0.22944106146363083,
      "learning_rate": 6.4918686990774126e-06,
      "loss": 0.0171,
      "step": 5182
    },
    {
      "epoch": 1.3899168677929739,
      "grad_norm": 0.35259992801980905,
      "learning_rate": 6.49037950731964e-06,
      "loss": 0.0254,
      "step": 5183
    },
    {
      "epoch": 1.3901850362027353,
      "grad_norm": 0.37116440513520066,
      "learning_rate": 6.48889017044757e-06,
      "loss": 0.0231,
      "step": 5184
    },
    {
      "epoch": 1.3904532046124967,
      "grad_norm": 0.3408909605440221,
      "learning_rate": 6.4874006886062205e-06,
      "loss": 0.033,
      "step": 5185
    },
    {
      "epoch": 1.390721373022258,
      "grad_norm": 0.32272411590045763,
      "learning_rate": 6.4859110619406125e-06,
      "loss": 0.0275,
      "step": 5186
    },
    {
      "epoch": 1.3909895414320192,
      "grad_norm": 0.2877175369892691,
      "learning_rate": 6.48442129059579e-06,
      "loss": 0.0274,
      "step": 5187
    },
    {
      "epoch": 1.3912577098417807,
      "grad_norm": 0.20653224324990985,
      "learning_rate": 6.4829313747168075e-06,
      "loss": 0.019,
      "step": 5188
    },
    {
      "epoch": 1.3915258782515418,
      "grad_norm": 0.37871118074958016,
      "learning_rate": 6.481441314448734e-06,
      "loss": 0.0374,
      "step": 5189
    },
    {
      "epoch": 1.3917940466613032,
      "grad_norm": 0.18229658447990152,
      "learning_rate": 6.479951109936653e-06,
      "loss": 0.0144,
      "step": 5190
    },
    {
      "epoch": 1.3920622150710646,
      "grad_norm": 0.3074400030242003,
      "learning_rate": 6.4784607613256615e-06,
      "loss": 0.021,
      "step": 5191
    },
    {
      "epoch": 1.392330383480826,
      "grad_norm": 0.3109855953044543,
      "learning_rate": 6.476970268760871e-06,
      "loss": 0.0236,
      "step": 5192
    },
    {
      "epoch": 1.3925985518905872,
      "grad_norm": 0.2473880043851488,
      "learning_rate": 6.475479632387407e-06,
      "loss": 0.0179,
      "step": 5193
    },
    {
      "epoch": 1.3928667203003486,
      "grad_norm": 0.22139972218326306,
      "learning_rate": 6.473988852350408e-06,
      "loss": 0.019,
      "step": 5194
    },
    {
      "epoch": 1.3931348887101098,
      "grad_norm": 0.2159896545649916,
      "learning_rate": 6.472497928795027e-06,
      "loss": 0.0175,
      "step": 5195
    },
    {
      "epoch": 1.3934030571198712,
      "grad_norm": 0.2222372696612467,
      "learning_rate": 6.471006861866433e-06,
      "loss": 0.0226,
      "step": 5196
    },
    {
      "epoch": 1.3936712255296326,
      "grad_norm": 0.3106854918072371,
      "learning_rate": 6.469515651709808e-06,
      "loss": 0.0303,
      "step": 5197
    },
    {
      "epoch": 1.393939393939394,
      "grad_norm": 0.39325850247664007,
      "learning_rate": 6.468024298470343e-06,
      "loss": 0.0174,
      "step": 5198
    },
    {
      "epoch": 1.3942075623491552,
      "grad_norm": 0.4710431167945331,
      "learning_rate": 6.4665328022932505e-06,
      "loss": 0.0307,
      "step": 5199
    },
    {
      "epoch": 1.3944757307589166,
      "grad_norm": 0.22877078657714775,
      "learning_rate": 6.465041163323752e-06,
      "loss": 0.0181,
      "step": 5200
    },
    {
      "epoch": 1.3947438991686778,
      "grad_norm": 0.26515572725263986,
      "learning_rate": 6.463549381707087e-06,
      "loss": 0.0188,
      "step": 5201
    },
    {
      "epoch": 1.3950120675784392,
      "grad_norm": 0.27943431328503227,
      "learning_rate": 6.462057457588503e-06,
      "loss": 0.0207,
      "step": 5202
    },
    {
      "epoch": 1.3952802359882006,
      "grad_norm": 0.2785532796197089,
      "learning_rate": 6.460565391113266e-06,
      "loss": 0.0212,
      "step": 5203
    },
    {
      "epoch": 1.395548404397962,
      "grad_norm": 0.4291464036396191,
      "learning_rate": 6.459073182426654e-06,
      "loss": 0.0223,
      "step": 5204
    },
    {
      "epoch": 1.3958165728077232,
      "grad_norm": 0.27142747503558823,
      "learning_rate": 6.45758083167396e-06,
      "loss": 0.0171,
      "step": 5205
    },
    {
      "epoch": 1.3960847412174846,
      "grad_norm": 0.21253071722607556,
      "learning_rate": 6.4560883390004905e-06,
      "loss": 0.0201,
      "step": 5206
    },
    {
      "epoch": 1.3963529096272458,
      "grad_norm": 0.27426484295345344,
      "learning_rate": 6.454595704551567e-06,
      "loss": 0.023,
      "step": 5207
    },
    {
      "epoch": 1.3966210780370072,
      "grad_norm": 0.2801770565473622,
      "learning_rate": 6.45310292847252e-06,
      "loss": 0.0325,
      "step": 5208
    },
    {
      "epoch": 1.3968892464467686,
      "grad_norm": 0.3291962380521347,
      "learning_rate": 6.4516100109087e-06,
      "loss": 0.0274,
      "step": 5209
    },
    {
      "epoch": 1.39715741485653,
      "grad_norm": 0.2301296610068382,
      "learning_rate": 6.450116952005464e-06,
      "loss": 0.0207,
      "step": 5210
    },
    {
      "epoch": 1.3974255832662912,
      "grad_norm": 0.43305789779332676,
      "learning_rate": 6.448623751908193e-06,
      "loss": 0.0265,
      "step": 5211
    },
    {
      "epoch": 1.3976937516760526,
      "grad_norm": 0.6956404534281433,
      "learning_rate": 6.4471304107622725e-06,
      "loss": 0.0407,
      "step": 5212
    },
    {
      "epoch": 1.3979619200858138,
      "grad_norm": 0.349718616042425,
      "learning_rate": 6.4456369287131084e-06,
      "loss": 0.0302,
      "step": 5213
    },
    {
      "epoch": 1.3982300884955752,
      "grad_norm": 0.33105537887205494,
      "learning_rate": 6.444143305906112e-06,
      "loss": 0.0204,
      "step": 5214
    },
    {
      "epoch": 1.3984982569053366,
      "grad_norm": 0.264735580058941,
      "learning_rate": 6.4426495424867164e-06,
      "loss": 0.0224,
      "step": 5215
    },
    {
      "epoch": 1.398766425315098,
      "grad_norm": 0.35248519644064036,
      "learning_rate": 6.441155638600365e-06,
      "loss": 0.0259,
      "step": 5216
    },
    {
      "epoch": 1.3990345937248592,
      "grad_norm": 0.3713305696021956,
      "learning_rate": 6.439661594392515e-06,
      "loss": 0.0251,
      "step": 5217
    },
    {
      "epoch": 1.3993027621346206,
      "grad_norm": 0.28178920462796453,
      "learning_rate": 6.438167410008638e-06,
      "loss": 0.0212,
      "step": 5218
    },
    {
      "epoch": 1.3995709305443818,
      "grad_norm": 0.6224282570483649,
      "learning_rate": 6.436673085594221e-06,
      "loss": 0.0251,
      "step": 5219
    },
    {
      "epoch": 1.3998390989541432,
      "grad_norm": 0.2858509706628988,
      "learning_rate": 6.4351786212947565e-06,
      "loss": 0.0226,
      "step": 5220
    },
    {
      "epoch": 1.4001072673639046,
      "grad_norm": 0.2914970784605602,
      "learning_rate": 6.433684017255761e-06,
      "loss": 0.0269,
      "step": 5221
    },
    {
      "epoch": 1.400375435773666,
      "grad_norm": 0.304243040205742,
      "learning_rate": 6.4321892736227585e-06,
      "loss": 0.0392,
      "step": 5222
    },
    {
      "epoch": 1.4006436041834271,
      "grad_norm": 0.24467194358553385,
      "learning_rate": 6.43069439054129e-06,
      "loss": 0.0227,
      "step": 5223
    },
    {
      "epoch": 1.4009117725931886,
      "grad_norm": 0.23797578224929014,
      "learning_rate": 6.4291993681569065e-06,
      "loss": 0.0252,
      "step": 5224
    },
    {
      "epoch": 1.4011799410029497,
      "grad_norm": 0.23607386543837883,
      "learning_rate": 6.427704206615176e-06,
      "loss": 0.0214,
      "step": 5225
    },
    {
      "epoch": 1.4014481094127111,
      "grad_norm": 0.3347313654782488,
      "learning_rate": 6.426208906061676e-06,
      "loss": 0.0345,
      "step": 5226
    },
    {
      "epoch": 1.4017162778224725,
      "grad_norm": 0.2561400006699048,
      "learning_rate": 6.424713466642002e-06,
      "loss": 0.0189,
      "step": 5227
    },
    {
      "epoch": 1.401984446232234,
      "grad_norm": 0.3299244174657613,
      "learning_rate": 6.4232178885017606e-06,
      "loss": 0.0275,
      "step": 5228
    },
    {
      "epoch": 1.4022526146419951,
      "grad_norm": 0.3099302222251416,
      "learning_rate": 6.421722171786571e-06,
      "loss": 0.0291,
      "step": 5229
    },
    {
      "epoch": 1.4025207830517565,
      "grad_norm": 0.2866042133617756,
      "learning_rate": 6.420226316642069e-06,
      "loss": 0.0224,
      "step": 5230
    },
    {
      "epoch": 1.4027889514615177,
      "grad_norm": 0.3153110884215908,
      "learning_rate": 6.418730323213901e-06,
      "loss": 0.0228,
      "step": 5231
    },
    {
      "epoch": 1.4030571198712791,
      "grad_norm": 0.43223590951397056,
      "learning_rate": 6.417234191647728e-06,
      "loss": 0.0233,
      "step": 5232
    },
    {
      "epoch": 1.4033252882810405,
      "grad_norm": 0.2854848003877294,
      "learning_rate": 6.415737922089224e-06,
      "loss": 0.0275,
      "step": 5233
    },
    {
      "epoch": 1.403593456690802,
      "grad_norm": 0.2693254431737801,
      "learning_rate": 6.414241514684077e-06,
      "loss": 0.0238,
      "step": 5234
    },
    {
      "epoch": 1.4038616251005631,
      "grad_norm": 0.2882725970015527,
      "learning_rate": 6.4127449695779894e-06,
      "loss": 0.0199,
      "step": 5235
    },
    {
      "epoch": 1.4041297935103245,
      "grad_norm": 0.2821050582967342,
      "learning_rate": 6.411248286916675e-06,
      "loss": 0.0188,
      "step": 5236
    },
    {
      "epoch": 1.4043979619200857,
      "grad_norm": 0.4002808657214984,
      "learning_rate": 6.4097514668458614e-06,
      "loss": 0.0295,
      "step": 5237
    },
    {
      "epoch": 1.404666130329847,
      "grad_norm": 0.22753981501253998,
      "learning_rate": 6.4082545095112904e-06,
      "loss": 0.025,
      "step": 5238
    },
    {
      "epoch": 1.4049342987396085,
      "grad_norm": 0.2826434960327498,
      "learning_rate": 6.406757415058717e-06,
      "loss": 0.0252,
      "step": 5239
    },
    {
      "epoch": 1.40520246714937,
      "grad_norm": 0.3243913881501028,
      "learning_rate": 6.405260183633908e-06,
      "loss": 0.0317,
      "step": 5240
    },
    {
      "epoch": 1.405470635559131,
      "grad_norm": 0.38825089959778947,
      "learning_rate": 6.403762815382646e-06,
      "loss": 0.0351,
      "step": 5241
    },
    {
      "epoch": 1.4057388039688925,
      "grad_norm": 0.3005784717623661,
      "learning_rate": 6.402265310450726e-06,
      "loss": 0.0243,
      "step": 5242
    },
    {
      "epoch": 1.4060069723786537,
      "grad_norm": 0.2820356864044271,
      "learning_rate": 6.400767668983955e-06,
      "loss": 0.0315,
      "step": 5243
    },
    {
      "epoch": 1.406275140788415,
      "grad_norm": 0.32030767616135214,
      "learning_rate": 6.399269891128156e-06,
      "loss": 0.0242,
      "step": 5244
    },
    {
      "epoch": 1.4065433091981765,
      "grad_norm": 0.2640864191029548,
      "learning_rate": 6.397771977029161e-06,
      "loss": 0.026,
      "step": 5245
    },
    {
      "epoch": 1.406811477607938,
      "grad_norm": 0.449536447638611,
      "learning_rate": 6.396273926832821e-06,
      "loss": 0.034,
      "step": 5246
    },
    {
      "epoch": 1.407079646017699,
      "grad_norm": 0.2624313155718272,
      "learning_rate": 6.394775740684996e-06,
      "loss": 0.0265,
      "step": 5247
    },
    {
      "epoch": 1.4073478144274605,
      "grad_norm": 0.23758192821345253,
      "learning_rate": 6.39327741873156e-06,
      "loss": 0.0228,
      "step": 5248
    },
    {
      "epoch": 1.4076159828372217,
      "grad_norm": 0.2667669305291097,
      "learning_rate": 6.3917789611184e-06,
      "loss": 0.0219,
      "step": 5249
    },
    {
      "epoch": 1.407884151246983,
      "grad_norm": 0.43251985750390337,
      "learning_rate": 6.390280367991419e-06,
      "loss": 0.0296,
      "step": 5250
    },
    {
      "epoch": 1.4081523196567445,
      "grad_norm": 0.6050776657347895,
      "learning_rate": 6.388781639496529e-06,
      "loss": 0.0258,
      "step": 5251
    },
    {
      "epoch": 1.4084204880665059,
      "grad_norm": 0.3003150905683207,
      "learning_rate": 6.387282775779657e-06,
      "loss": 0.0225,
      "step": 5252
    },
    {
      "epoch": 1.408688656476267,
      "grad_norm": 0.2683298170080757,
      "learning_rate": 6.385783776986745e-06,
      "loss": 0.0198,
      "step": 5253
    },
    {
      "epoch": 1.4089568248860285,
      "grad_norm": 0.268678344864896,
      "learning_rate": 6.384284643263745e-06,
      "loss": 0.0216,
      "step": 5254
    },
    {
      "epoch": 1.4092249932957897,
      "grad_norm": 0.3731545596970325,
      "learning_rate": 6.382785374756627e-06,
      "loss": 0.032,
      "step": 5255
    },
    {
      "epoch": 1.409493161705551,
      "grad_norm": 0.3426301584764138,
      "learning_rate": 6.3812859716113675e-06,
      "loss": 0.0248,
      "step": 5256
    },
    {
      "epoch": 1.4097613301153125,
      "grad_norm": 0.285326095267181,
      "learning_rate": 6.379786433973959e-06,
      "loss": 0.0274,
      "step": 5257
    },
    {
      "epoch": 1.4100294985250739,
      "grad_norm": 0.4712938969360387,
      "learning_rate": 6.378286761990408e-06,
      "loss": 0.0202,
      "step": 5258
    },
    {
      "epoch": 1.410297666934835,
      "grad_norm": 0.4685059918605246,
      "learning_rate": 6.376786955806735e-06,
      "loss": 0.0381,
      "step": 5259
    },
    {
      "epoch": 1.4105658353445965,
      "grad_norm": 0.22491126398403954,
      "learning_rate": 6.375287015568971e-06,
      "loss": 0.0189,
      "step": 5260
    },
    {
      "epoch": 1.4108340037543576,
      "grad_norm": 0.387916355012814,
      "learning_rate": 6.373786941423165e-06,
      "loss": 0.0295,
      "step": 5261
    },
    {
      "epoch": 1.411102172164119,
      "grad_norm": 0.28576208946203746,
      "learning_rate": 6.37228673351537e-06,
      "loss": 0.021,
      "step": 5262
    },
    {
      "epoch": 1.4113703405738804,
      "grad_norm": 0.2885125546012685,
      "learning_rate": 6.3707863919916565e-06,
      "loss": 0.0221,
      "step": 5263
    },
    {
      "epoch": 1.4116385089836418,
      "grad_norm": 0.2869403262283209,
      "learning_rate": 6.369285916998113e-06,
      "loss": 0.0288,
      "step": 5264
    },
    {
      "epoch": 1.411906677393403,
      "grad_norm": 0.28287860071558796,
      "learning_rate": 6.367785308680835e-06,
      "loss": 0.0234,
      "step": 5265
    },
    {
      "epoch": 1.4121748458031644,
      "grad_norm": 0.5990039101555422,
      "learning_rate": 6.366284567185934e-06,
      "loss": 0.028,
      "step": 5266
    },
    {
      "epoch": 1.4124430142129256,
      "grad_norm": 0.39137981904099617,
      "learning_rate": 6.364783692659532e-06,
      "loss": 0.0255,
      "step": 5267
    },
    {
      "epoch": 1.412711182622687,
      "grad_norm": 0.34672424828636894,
      "learning_rate": 6.363282685247765e-06,
      "loss": 0.0347,
      "step": 5268
    },
    {
      "epoch": 1.4129793510324484,
      "grad_norm": 0.2955189419664367,
      "learning_rate": 6.36178154509678e-06,
      "loss": 0.0211,
      "step": 5269
    },
    {
      "epoch": 1.4132475194422098,
      "grad_norm": 0.30517366119817585,
      "learning_rate": 6.360280272352744e-06,
      "loss": 0.0266,
      "step": 5270
    },
    {
      "epoch": 1.413515687851971,
      "grad_norm": 0.29938145815087114,
      "learning_rate": 6.358778867161829e-06,
      "loss": 0.0251,
      "step": 5271
    },
    {
      "epoch": 1.4137838562617324,
      "grad_norm": 0.24257819976122985,
      "learning_rate": 6.357277329670223e-06,
      "loss": 0.0209,
      "step": 5272
    },
    {
      "epoch": 1.4140520246714936,
      "grad_norm": 0.23794818770066298,
      "learning_rate": 6.355775660024128e-06,
      "loss": 0.0169,
      "step": 5273
    },
    {
      "epoch": 1.414320193081255,
      "grad_norm": 0.32894626930039006,
      "learning_rate": 6.354273858369754e-06,
      "loss": 0.0231,
      "step": 5274
    },
    {
      "epoch": 1.4145883614910164,
      "grad_norm": 0.29096812876719,
      "learning_rate": 6.352771924853331e-06,
      "loss": 0.0232,
      "step": 5275
    },
    {
      "epoch": 1.4148565299007778,
      "grad_norm": 0.38797090038222903,
      "learning_rate": 6.3512698596210985e-06,
      "loss": 0.0253,
      "step": 5276
    },
    {
      "epoch": 1.415124698310539,
      "grad_norm": 0.2774409407450422,
      "learning_rate": 6.349767662819306e-06,
      "loss": 0.0226,
      "step": 5277
    },
    {
      "epoch": 1.4153928667203004,
      "grad_norm": 0.2958631893308465,
      "learning_rate": 6.34826533459422e-06,
      "loss": 0.0238,
      "step": 5278
    },
    {
      "epoch": 1.4156610351300616,
      "grad_norm": 0.2816345864410328,
      "learning_rate": 6.346762875092118e-06,
      "loss": 0.0208,
      "step": 5279
    },
    {
      "epoch": 1.415929203539823,
      "grad_norm": 0.26791417253659155,
      "learning_rate": 6.3452602844592905e-06,
      "loss": 0.0201,
      "step": 5280
    },
    {
      "epoch": 1.4161973719495844,
      "grad_norm": 0.2740727744967442,
      "learning_rate": 6.3437575628420415e-06,
      "loss": 0.0199,
      "step": 5281
    },
    {
      "epoch": 1.4164655403593458,
      "grad_norm": 0.29438001216708637,
      "learning_rate": 6.342254710386685e-06,
      "loss": 0.0271,
      "step": 5282
    },
    {
      "epoch": 1.416733708769107,
      "grad_norm": 0.33155585095414847,
      "learning_rate": 6.340751727239551e-06,
      "loss": 0.0274,
      "step": 5283
    },
    {
      "epoch": 1.4170018771788684,
      "grad_norm": 0.4094600522951844,
      "learning_rate": 6.339248613546981e-06,
      "loss": 0.0321,
      "step": 5284
    },
    {
      "epoch": 1.4172700455886296,
      "grad_norm": 0.23928278020833182,
      "learning_rate": 6.33774536945533e-06,
      "loss": 0.023,
      "step": 5285
    },
    {
      "epoch": 1.417538213998391,
      "grad_norm": 0.25189107980082537,
      "learning_rate": 6.336241995110963e-06,
      "loss": 0.0202,
      "step": 5286
    },
    {
      "epoch": 1.4178063824081524,
      "grad_norm": 0.23110726629153236,
      "learning_rate": 6.3347384906602615e-06,
      "loss": 0.0215,
      "step": 5287
    },
    {
      "epoch": 1.4180745508179138,
      "grad_norm": 0.2477437644769945,
      "learning_rate": 6.333234856249616e-06,
      "loss": 0.0214,
      "step": 5288
    },
    {
      "epoch": 1.418342719227675,
      "grad_norm": 0.2452979555821637,
      "learning_rate": 6.331731092025433e-06,
      "loss": 0.0228,
      "step": 5289
    },
    {
      "epoch": 1.4186108876374364,
      "grad_norm": 0.33282911104338675,
      "learning_rate": 6.330227198134129e-06,
      "loss": 0.0279,
      "step": 5290
    },
    {
      "epoch": 1.4188790560471976,
      "grad_norm": 0.21087370513202716,
      "learning_rate": 6.328723174722133e-06,
      "loss": 0.0196,
      "step": 5291
    },
    {
      "epoch": 1.419147224456959,
      "grad_norm": 0.6791959302081488,
      "learning_rate": 6.3272190219358905e-06,
      "loss": 0.0381,
      "step": 5292
    },
    {
      "epoch": 1.4194153928667204,
      "grad_norm": 0.29527479029750564,
      "learning_rate": 6.325714739921857e-06,
      "loss": 0.0266,
      "step": 5293
    },
    {
      "epoch": 1.4196835612764818,
      "grad_norm": 0.29092415678884226,
      "learning_rate": 6.324210328826496e-06,
      "loss": 0.0252,
      "step": 5294
    },
    {
      "epoch": 1.419951729686243,
      "grad_norm": 0.17590664031497444,
      "learning_rate": 6.322705788796293e-06,
      "loss": 0.0143,
      "step": 5295
    },
    {
      "epoch": 1.4202198980960044,
      "grad_norm": 0.1954030057545939,
      "learning_rate": 6.321201119977738e-06,
      "loss": 0.0162,
      "step": 5296
    },
    {
      "epoch": 1.4204880665057655,
      "grad_norm": 0.5147688847097033,
      "learning_rate": 6.319696322517339e-06,
      "loss": 0.0418,
      "step": 5297
    },
    {
      "epoch": 1.420756234915527,
      "grad_norm": 0.28811284829650596,
      "learning_rate": 6.318191396561613e-06,
      "loss": 0.0298,
      "step": 5298
    },
    {
      "epoch": 1.4210244033252883,
      "grad_norm": 0.3596284113754197,
      "learning_rate": 6.316686342257088e-06,
      "loss": 0.0317,
      "step": 5299
    },
    {
      "epoch": 1.4212925717350497,
      "grad_norm": 0.26525356528217353,
      "learning_rate": 6.315181159750311e-06,
      "loss": 0.0235,
      "step": 5300
    },
    {
      "epoch": 1.421560740144811,
      "grad_norm": 0.19579851057319878,
      "learning_rate": 6.313675849187836e-06,
      "loss": 0.0166,
      "step": 5301
    },
    {
      "epoch": 1.4218289085545723,
      "grad_norm": 0.23903943648080928,
      "learning_rate": 6.312170410716231e-06,
      "loss": 0.0212,
      "step": 5302
    },
    {
      "epoch": 1.4220970769643335,
      "grad_norm": 0.2662655635446794,
      "learning_rate": 6.310664844482078e-06,
      "loss": 0.022,
      "step": 5303
    },
    {
      "epoch": 1.422365245374095,
      "grad_norm": 0.296284990351937,
      "learning_rate": 6.3091591506319695e-06,
      "loss": 0.0222,
      "step": 5304
    },
    {
      "epoch": 1.4226334137838563,
      "grad_norm": 0.3513022918152601,
      "learning_rate": 6.307653329312509e-06,
      "loss": 0.024,
      "step": 5305
    },
    {
      "epoch": 1.4229015821936175,
      "grad_norm": 0.2912085737295004,
      "learning_rate": 6.306147380670313e-06,
      "loss": 0.0178,
      "step": 5306
    },
    {
      "epoch": 1.423169750603379,
      "grad_norm": 0.2431643508910642,
      "learning_rate": 6.304641304852017e-06,
      "loss": 0.0209,
      "step": 5307
    },
    {
      "epoch": 1.4234379190131403,
      "grad_norm": 0.26397467506014016,
      "learning_rate": 6.303135102004261e-06,
      "loss": 0.029,
      "step": 5308
    },
    {
      "epoch": 1.4237060874229015,
      "grad_norm": 0.2451141481956949,
      "learning_rate": 6.3016287722737015e-06,
      "loss": 0.021,
      "step": 5309
    },
    {
      "epoch": 1.423974255832663,
      "grad_norm": 0.31323563700490503,
      "learning_rate": 6.300122315807003e-06,
      "loss": 0.022,
      "step": 5310
    },
    {
      "epoch": 1.4242424242424243,
      "grad_norm": 0.2689697616355054,
      "learning_rate": 6.298615732750845e-06,
      "loss": 0.0345,
      "step": 5311
    },
    {
      "epoch": 1.4245105926521855,
      "grad_norm": 0.2515621802688394,
      "learning_rate": 6.297109023251919e-06,
      "loss": 0.0243,
      "step": 5312
    },
    {
      "epoch": 1.424778761061947,
      "grad_norm": 0.287115642051253,
      "learning_rate": 6.295602187456935e-06,
      "loss": 0.0317,
      "step": 5313
    },
    {
      "epoch": 1.4250469294717083,
      "grad_norm": 0.30084846365717655,
      "learning_rate": 6.294095225512604e-06,
      "loss": 0.0246,
      "step": 5314
    },
    {
      "epoch": 1.4253150978814695,
      "grad_norm": 0.5250332356935385,
      "learning_rate": 6.292588137565657e-06,
      "loss": 0.0161,
      "step": 5315
    },
    {
      "epoch": 1.4255832662912309,
      "grad_norm": 0.2009372908100997,
      "learning_rate": 6.291080923762836e-06,
      "loss": 0.0171,
      "step": 5316
    },
    {
      "epoch": 1.4258514347009923,
      "grad_norm": 0.25315568040806347,
      "learning_rate": 6.289573584250892e-06,
      "loss": 0.0165,
      "step": 5317
    },
    {
      "epoch": 1.4261196031107535,
      "grad_norm": 0.34238494980140943,
      "learning_rate": 6.2880661191765916e-06,
      "loss": 0.0242,
      "step": 5318
    },
    {
      "epoch": 1.4263877715205149,
      "grad_norm": 0.29577184021758646,
      "learning_rate": 6.286558528686713e-06,
      "loss": 0.0235,
      "step": 5319
    },
    {
      "epoch": 1.4266559399302763,
      "grad_norm": 0.33966007520380315,
      "learning_rate": 6.285050812928047e-06,
      "loss": 0.0239,
      "step": 5320
    },
    {
      "epoch": 1.4269241083400375,
      "grad_norm": 0.22092185642465098,
      "learning_rate": 6.283542972047394e-06,
      "loss": 0.019,
      "step": 5321
    },
    {
      "epoch": 1.4271922767497989,
      "grad_norm": 0.31381530525257817,
      "learning_rate": 6.28203500619157e-06,
      "loss": 0.0324,
      "step": 5322
    },
    {
      "epoch": 1.4274604451595603,
      "grad_norm": 0.22676412223518147,
      "learning_rate": 6.280526915507401e-06,
      "loss": 0.0187,
      "step": 5323
    },
    {
      "epoch": 1.4277286135693215,
      "grad_norm": 0.3388827535947665,
      "learning_rate": 6.279018700141727e-06,
      "loss": 0.0262,
      "step": 5324
    },
    {
      "epoch": 1.4279967819790829,
      "grad_norm": 0.2531809595948373,
      "learning_rate": 6.277510360241397e-06,
      "loss": 0.0251,
      "step": 5325
    },
    {
      "epoch": 1.4282649503888443,
      "grad_norm": 0.2514353170350491,
      "learning_rate": 6.276001895953276e-06,
      "loss": 0.017,
      "step": 5326
    },
    {
      "epoch": 1.4285331187986055,
      "grad_norm": 0.34551389676483873,
      "learning_rate": 6.274493307424237e-06,
      "loss": 0.0353,
      "step": 5327
    },
    {
      "epoch": 1.4288012872083669,
      "grad_norm": 0.27627166393599467,
      "learning_rate": 6.272984594801168e-06,
      "loss": 0.0265,
      "step": 5328
    },
    {
      "epoch": 1.4290694556181283,
      "grad_norm": 0.3529716224944672,
      "learning_rate": 6.271475758230969e-06,
      "loss": 0.0266,
      "step": 5329
    },
    {
      "epoch": 1.4293376240278894,
      "grad_norm": 0.238325768865046,
      "learning_rate": 6.2699667978605514e-06,
      "loss": 0.0202,
      "step": 5330
    },
    {
      "epoch": 1.4296057924376508,
      "grad_norm": 0.24948170084339144,
      "learning_rate": 6.268457713836839e-06,
      "loss": 0.0266,
      "step": 5331
    },
    {
      "epoch": 1.4298739608474123,
      "grad_norm": 0.24790634538282338,
      "learning_rate": 6.266948506306767e-06,
      "loss": 0.0228,
      "step": 5332
    },
    {
      "epoch": 1.4301421292571734,
      "grad_norm": 0.1966884924329599,
      "learning_rate": 6.265439175417282e-06,
      "loss": 0.0163,
      "step": 5333
    },
    {
      "epoch": 1.4304102976669348,
      "grad_norm": 0.2746508998224272,
      "learning_rate": 6.263929721315343e-06,
      "loss": 0.0257,
      "step": 5334
    },
    {
      "epoch": 1.4306784660766962,
      "grad_norm": 0.2831131848300995,
      "learning_rate": 6.262420144147924e-06,
      "loss": 0.0315,
      "step": 5335
    },
    {
      "epoch": 1.4309466344864574,
      "grad_norm": 0.27531355947121033,
      "learning_rate": 6.260910444062007e-06,
      "loss": 0.0216,
      "step": 5336
    },
    {
      "epoch": 1.4312148028962188,
      "grad_norm": 0.2284117946181391,
      "learning_rate": 6.2594006212045874e-06,
      "loss": 0.0259,
      "step": 5337
    },
    {
      "epoch": 1.4314829713059802,
      "grad_norm": 0.3635535174231931,
      "learning_rate": 6.257890675722673e-06,
      "loss": 0.0332,
      "step": 5338
    },
    {
      "epoch": 1.4317511397157414,
      "grad_norm": 0.2597549147439196,
      "learning_rate": 6.256380607763283e-06,
      "loss": 0.0301,
      "step": 5339
    },
    {
      "epoch": 1.4320193081255028,
      "grad_norm": 0.30154118391154877,
      "learning_rate": 6.25487041747345e-06,
      "loss": 0.0256,
      "step": 5340
    },
    {
      "epoch": 1.4322874765352642,
      "grad_norm": 0.25394083036931275,
      "learning_rate": 6.253360105000215e-06,
      "loss": 0.0278,
      "step": 5341
    },
    {
      "epoch": 1.4325556449450254,
      "grad_norm": 0.32571866178191444,
      "learning_rate": 6.251849670490634e-06,
      "loss": 0.0316,
      "step": 5342
    },
    {
      "epoch": 1.4328238133547868,
      "grad_norm": 0.2573370619177963,
      "learning_rate": 6.250339114091775e-06,
      "loss": 0.0208,
      "step": 5343
    },
    {
      "epoch": 1.4330919817645482,
      "grad_norm": 0.27500914381349895,
      "learning_rate": 6.248828435950714e-06,
      "loss": 0.0222,
      "step": 5344
    },
    {
      "epoch": 1.4333601501743094,
      "grad_norm": 0.34246373123089185,
      "learning_rate": 6.247317636214546e-06,
      "loss": 0.0307,
      "step": 5345
    },
    {
      "epoch": 1.4336283185840708,
      "grad_norm": 0.2243444086184642,
      "learning_rate": 6.2458067150303715e-06,
      "loss": 0.0171,
      "step": 5346
    },
    {
      "epoch": 1.4338964869938322,
      "grad_norm": 0.37676339920252094,
      "learning_rate": 6.244295672545304e-06,
      "loss": 0.0244,
      "step": 5347
    },
    {
      "epoch": 1.4341646554035934,
      "grad_norm": 0.4287692486266675,
      "learning_rate": 6.24278450890647e-06,
      "loss": 0.0254,
      "step": 5348
    },
    {
      "epoch": 1.4344328238133548,
      "grad_norm": 0.3084207538886438,
      "learning_rate": 6.241273224261007e-06,
      "loss": 0.0192,
      "step": 5349
    },
    {
      "epoch": 1.434700992223116,
      "grad_norm": 0.3092660479953778,
      "learning_rate": 6.239761818756067e-06,
      "loss": 0.0288,
      "step": 5350
    },
    {
      "epoch": 1.4349691606328774,
      "grad_norm": 0.2897819533030703,
      "learning_rate": 6.23825029253881e-06,
      "loss": 0.0301,
      "step": 5351
    },
    {
      "epoch": 1.4352373290426388,
      "grad_norm": 0.38790822378766804,
      "learning_rate": 6.236738645756412e-06,
      "loss": 0.028,
      "step": 5352
    },
    {
      "epoch": 1.4355054974524002,
      "grad_norm": 0.312685517382252,
      "learning_rate": 6.235226878556052e-06,
      "loss": 0.0223,
      "step": 5353
    },
    {
      "epoch": 1.4357736658621614,
      "grad_norm": 0.4029924728266074,
      "learning_rate": 6.233714991084931e-06,
      "loss": 0.0321,
      "step": 5354
    },
    {
      "epoch": 1.4360418342719228,
      "grad_norm": 0.2589231604220812,
      "learning_rate": 6.2322029834902565e-06,
      "loss": 0.0253,
      "step": 5355
    },
    {
      "epoch": 1.436310002681684,
      "grad_norm": 0.18906847993576714,
      "learning_rate": 6.23069085591925e-06,
      "loss": 0.0172,
      "step": 5356
    },
    {
      "epoch": 1.4365781710914454,
      "grad_norm": 0.26724489482962327,
      "learning_rate": 6.229178608519142e-06,
      "loss": 0.0223,
      "step": 5357
    },
    {
      "epoch": 1.4368463395012068,
      "grad_norm": 0.31589357740095964,
      "learning_rate": 6.227666241437177e-06,
      "loss": 0.0266,
      "step": 5358
    },
    {
      "epoch": 1.4371145079109682,
      "grad_norm": 0.23280446922185238,
      "learning_rate": 6.2261537548206084e-06,
      "loss": 0.0168,
      "step": 5359
    },
    {
      "epoch": 1.4373826763207294,
      "grad_norm": 0.24234337781626092,
      "learning_rate": 6.2246411488167035e-06,
      "loss": 0.0226,
      "step": 5360
    },
    {
      "epoch": 1.4376508447304908,
      "grad_norm": 0.2947696170753836,
      "learning_rate": 6.2231284235727426e-06,
      "loss": 0.0258,
      "step": 5361
    },
    {
      "epoch": 1.437919013140252,
      "grad_norm": 0.2898646835357457,
      "learning_rate": 6.2216155792360144e-06,
      "loss": 0.026,
      "step": 5362
    },
    {
      "epoch": 1.4381871815500133,
      "grad_norm": 0.34950747799872084,
      "learning_rate": 6.22010261595382e-06,
      "loss": 0.0287,
      "step": 5363
    },
    {
      "epoch": 1.4384553499597748,
      "grad_norm": 0.2823247912419688,
      "learning_rate": 6.218589533873474e-06,
      "loss": 0.02,
      "step": 5364
    },
    {
      "epoch": 1.4387235183695362,
      "grad_norm": 0.37375612660157514,
      "learning_rate": 6.217076333142299e-06,
      "loss": 0.0267,
      "step": 5365
    },
    {
      "epoch": 1.4389916867792973,
      "grad_norm": 0.3170426325065319,
      "learning_rate": 6.215563013907633e-06,
      "loss": 0.0259,
      "step": 5366
    },
    {
      "epoch": 1.4392598551890587,
      "grad_norm": 0.32428165033529943,
      "learning_rate": 6.214049576316824e-06,
      "loss": 0.0266,
      "step": 5367
    },
    {
      "epoch": 1.43952802359882,
      "grad_norm": 0.39350757506337875,
      "learning_rate": 6.212536020517231e-06,
      "loss": 0.0208,
      "step": 5368
    },
    {
      "epoch": 1.4397961920085813,
      "grad_norm": 0.2894035195973142,
      "learning_rate": 6.211022346656226e-06,
      "loss": 0.0254,
      "step": 5369
    },
    {
      "epoch": 1.4400643604183427,
      "grad_norm": 0.3397095198647925,
      "learning_rate": 6.209508554881189e-06,
      "loss": 0.0308,
      "step": 5370
    },
    {
      "epoch": 1.4403325288281041,
      "grad_norm": 0.3856211192662421,
      "learning_rate": 6.207994645339514e-06,
      "loss": 0.034,
      "step": 5371
    },
    {
      "epoch": 1.4406006972378653,
      "grad_norm": 0.5271103929955762,
      "learning_rate": 6.20648061817861e-06,
      "loss": 0.0444,
      "step": 5372
    },
    {
      "epoch": 1.4408688656476267,
      "grad_norm": 0.22365516816898437,
      "learning_rate": 6.20496647354589e-06,
      "loss": 0.0188,
      "step": 5373
    },
    {
      "epoch": 1.441137034057388,
      "grad_norm": 0.28126830149709386,
      "learning_rate": 6.203452211588784e-06,
      "loss": 0.0218,
      "step": 5374
    },
    {
      "epoch": 1.4414052024671493,
      "grad_norm": 0.30106082794817834,
      "learning_rate": 6.201937832454732e-06,
      "loss": 0.0266,
      "step": 5375
    },
    {
      "epoch": 1.4416733708769107,
      "grad_norm": 0.3926781714014268,
      "learning_rate": 6.200423336291184e-06,
      "loss": 0.0229,
      "step": 5376
    },
    {
      "epoch": 1.4419415392866721,
      "grad_norm": 0.35126430458160074,
      "learning_rate": 6.198908723245601e-06,
      "loss": 0.0294,
      "step": 5377
    },
    {
      "epoch": 1.4422097076964333,
      "grad_norm": 0.277438551264075,
      "learning_rate": 6.1973939934654606e-06,
      "loss": 0.026,
      "step": 5378
    },
    {
      "epoch": 1.4424778761061947,
      "grad_norm": 0.2719039380982307,
      "learning_rate": 6.195879147098246e-06,
      "loss": 0.0277,
      "step": 5379
    },
    {
      "epoch": 1.442746044515956,
      "grad_norm": 0.2488959067645933,
      "learning_rate": 6.194364184291452e-06,
      "loss": 0.0219,
      "step": 5380
    },
    {
      "epoch": 1.4430142129257173,
      "grad_norm": 0.28226131451840397,
      "learning_rate": 6.192849105192591e-06,
      "loss": 0.0257,
      "step": 5381
    },
    {
      "epoch": 1.4432823813354787,
      "grad_norm": 2.656061093999901,
      "learning_rate": 6.191333909949178e-06,
      "loss": 0.0329,
      "step": 5382
    },
    {
      "epoch": 1.44355054974524,
      "grad_norm": 0.3199529356714819,
      "learning_rate": 6.189818598708746e-06,
      "loss": 0.0265,
      "step": 5383
    },
    {
      "epoch": 1.4438187181550013,
      "grad_norm": 0.31335336995284435,
      "learning_rate": 6.188303171618835e-06,
      "loss": 0.0277,
      "step": 5384
    },
    {
      "epoch": 1.4440868865647627,
      "grad_norm": 0.2001070001128666,
      "learning_rate": 6.186787628826999e-06,
      "loss": 0.0163,
      "step": 5385
    },
    {
      "epoch": 1.4443550549745239,
      "grad_norm": 0.3590382991931159,
      "learning_rate": 6.185271970480803e-06,
      "loss": 0.0401,
      "step": 5386
    },
    {
      "epoch": 1.4446232233842853,
      "grad_norm": 0.4490398866323869,
      "learning_rate": 6.183756196727821e-06,
      "loss": 0.0276,
      "step": 5387
    },
    {
      "epoch": 1.4448913917940467,
      "grad_norm": 0.49620282453467407,
      "learning_rate": 6.1822403077156425e-06,
      "loss": 0.0281,
      "step": 5388
    },
    {
      "epoch": 1.445159560203808,
      "grad_norm": 0.3056842327650162,
      "learning_rate": 6.180724303591863e-06,
      "loss": 0.0194,
      "step": 5389
    },
    {
      "epoch": 1.4454277286135693,
      "grad_norm": 0.25076202895367433,
      "learning_rate": 6.179208184504094e-06,
      "loss": 0.0242,
      "step": 5390
    },
    {
      "epoch": 1.4456958970233307,
      "grad_norm": 0.3380564454144418,
      "learning_rate": 6.177691950599953e-06,
      "loss": 0.0288,
      "step": 5391
    },
    {
      "epoch": 1.4459640654330919,
      "grad_norm": 0.2661658336252339,
      "learning_rate": 6.176175602027075e-06,
      "loss": 0.0249,
      "step": 5392
    },
    {
      "epoch": 1.4462322338428533,
      "grad_norm": 0.30951523977936357,
      "learning_rate": 6.1746591389331e-06,
      "loss": 0.0292,
      "step": 5393
    },
    {
      "epoch": 1.4465004022526147,
      "grad_norm": 0.35093535702746337,
      "learning_rate": 6.173142561465685e-06,
      "loss": 0.0229,
      "step": 5394
    },
    {
      "epoch": 1.446768570662376,
      "grad_norm": 0.2905202282933159,
      "learning_rate": 6.171625869772493e-06,
      "loss": 0.0252,
      "step": 5395
    },
    {
      "epoch": 1.4470367390721373,
      "grad_norm": 0.28793645631755854,
      "learning_rate": 6.1701090640012e-06,
      "loss": 0.0169,
      "step": 5396
    },
    {
      "epoch": 1.4473049074818987,
      "grad_norm": 0.2918871024965385,
      "learning_rate": 6.168592144299493e-06,
      "loss": 0.0185,
      "step": 5397
    },
    {
      "epoch": 1.4475730758916598,
      "grad_norm": 0.3992701886046388,
      "learning_rate": 6.167075110815073e-06,
      "loss": 0.0236,
      "step": 5398
    },
    {
      "epoch": 1.4478412443014212,
      "grad_norm": 0.5432146900149479,
      "learning_rate": 6.165557963695648e-06,
      "loss": 0.0324,
      "step": 5399
    },
    {
      "epoch": 1.4481094127111827,
      "grad_norm": 0.22214525160955245,
      "learning_rate": 6.164040703088938e-06,
      "loss": 0.0171,
      "step": 5400
    },
    {
      "epoch": 1.448377581120944,
      "grad_norm": 0.27235723109764803,
      "learning_rate": 6.162523329142675e-06,
      "loss": 0.0177,
      "step": 5401
    },
    {
      "epoch": 1.4486457495307052,
      "grad_norm": 0.5863557922518207,
      "learning_rate": 6.161005842004603e-06,
      "loss": 0.0238,
      "step": 5402
    },
    {
      "epoch": 1.4489139179404666,
      "grad_norm": 0.3458640342663173,
      "learning_rate": 6.159488241822473e-06,
      "loss": 0.036,
      "step": 5403
    },
    {
      "epoch": 1.4491820863502278,
      "grad_norm": 0.44968478529302625,
      "learning_rate": 6.157970528744052e-06,
      "loss": 0.0386,
      "step": 5404
    },
    {
      "epoch": 1.4494502547599892,
      "grad_norm": 0.2481024175759677,
      "learning_rate": 6.156452702917115e-06,
      "loss": 0.0224,
      "step": 5405
    },
    {
      "epoch": 1.4497184231697506,
      "grad_norm": 0.2767639939105911,
      "learning_rate": 6.154934764489451e-06,
      "loss": 0.0227,
      "step": 5406
    },
    {
      "epoch": 1.449986591579512,
      "grad_norm": 0.28716338790752344,
      "learning_rate": 6.1534167136088526e-06,
      "loss": 0.0179,
      "step": 5407
    },
    {
      "epoch": 1.4502547599892732,
      "grad_norm": 0.256288954001623,
      "learning_rate": 6.15189855042313e-06,
      "loss": 0.0194,
      "step": 5408
    },
    {
      "epoch": 1.4505229283990346,
      "grad_norm": 0.33969281969695686,
      "learning_rate": 6.150380275080106e-06,
      "loss": 0.0334,
      "step": 5409
    },
    {
      "epoch": 1.4507910968087958,
      "grad_norm": 0.2768410126681915,
      "learning_rate": 6.148861887727607e-06,
      "loss": 0.0252,
      "step": 5410
    },
    {
      "epoch": 1.4510592652185572,
      "grad_norm": 0.8544553061361833,
      "learning_rate": 6.147343388513478e-06,
      "loss": 0.0331,
      "step": 5411
    },
    {
      "epoch": 1.4513274336283186,
      "grad_norm": 0.2713558423907453,
      "learning_rate": 6.145824777585569e-06,
      "loss": 0.0281,
      "step": 5412
    },
    {
      "epoch": 1.45159560203808,
      "grad_norm": 0.22835967825886871,
      "learning_rate": 6.144306055091743e-06,
      "loss": 0.0255,
      "step": 5413
    },
    {
      "epoch": 1.4518637704478412,
      "grad_norm": 0.29043065678398794,
      "learning_rate": 6.142787221179874e-06,
      "loss": 0.0245,
      "step": 5414
    },
    {
      "epoch": 1.4521319388576026,
      "grad_norm": 0.3105668017941954,
      "learning_rate": 6.141268275997848e-06,
      "loss": 0.0218,
      "step": 5415
    },
    {
      "epoch": 1.4524001072673638,
      "grad_norm": 0.20940118728885287,
      "learning_rate": 6.1397492196935605e-06,
      "loss": 0.0166,
      "step": 5416
    },
    {
      "epoch": 1.4526682756771252,
      "grad_norm": 0.27948167936148255,
      "learning_rate": 6.1382300524149175e-06,
      "loss": 0.0307,
      "step": 5417
    },
    {
      "epoch": 1.4529364440868866,
      "grad_norm": 0.2369622337780405,
      "learning_rate": 6.136710774309836e-06,
      "loss": 0.0262,
      "step": 5418
    },
    {
      "epoch": 1.453204612496648,
      "grad_norm": 0.24471813881081952,
      "learning_rate": 6.1351913855262435e-06,
      "loss": 0.0176,
      "step": 5419
    },
    {
      "epoch": 1.4534727809064092,
      "grad_norm": 0.251004682865238,
      "learning_rate": 6.13367188621208e-06,
      "loss": 0.0183,
      "step": 5420
    },
    {
      "epoch": 1.4537409493161706,
      "grad_norm": 0.36163705295938214,
      "learning_rate": 6.1321522765152955e-06,
      "loss": 0.0347,
      "step": 5421
    },
    {
      "epoch": 1.4540091177259318,
      "grad_norm": 0.3215894335482965,
      "learning_rate": 6.130632556583849e-06,
      "loss": 0.0252,
      "step": 5422
    },
    {
      "epoch": 1.4542772861356932,
      "grad_norm": 0.3460147679776799,
      "learning_rate": 6.129112726565711e-06,
      "loss": 0.0295,
      "step": 5423
    },
    {
      "epoch": 1.4545454545454546,
      "grad_norm": 0.48694449343876706,
      "learning_rate": 6.127592786608867e-06,
      "loss": 0.0321,
      "step": 5424
    },
    {
      "epoch": 1.454813622955216,
      "grad_norm": 0.37432867130095054,
      "learning_rate": 6.1260727368613045e-06,
      "loss": 0.0321,
      "step": 5425
    },
    {
      "epoch": 1.4550817913649772,
      "grad_norm": 0.23627940977878795,
      "learning_rate": 6.124552577471032e-06,
      "loss": 0.0254,
      "step": 5426
    },
    {
      "epoch": 1.4553499597747386,
      "grad_norm": 0.27414333772204913,
      "learning_rate": 6.123032308586059e-06,
      "loss": 0.0164,
      "step": 5427
    },
    {
      "epoch": 1.4556181281844998,
      "grad_norm": 0.2512431065160774,
      "learning_rate": 6.121511930354413e-06,
      "loss": 0.023,
      "step": 5428
    },
    {
      "epoch": 1.4558862965942612,
      "grad_norm": 0.481374583153334,
      "learning_rate": 6.119991442924126e-06,
      "loss": 0.0291,
      "step": 5429
    },
    {
      "epoch": 1.4561544650040226,
      "grad_norm": 0.29528617302655935,
      "learning_rate": 6.118470846443247e-06,
      "loss": 0.0186,
      "step": 5430
    },
    {
      "epoch": 1.456422633413784,
      "grad_norm": 0.2643976611022459,
      "learning_rate": 6.11695014105983e-06,
      "loss": 0.0254,
      "step": 5431
    },
    {
      "epoch": 1.4566908018235452,
      "grad_norm": 0.19463257263029335,
      "learning_rate": 6.115429326921944e-06,
      "loss": 0.0251,
      "step": 5432
    },
    {
      "epoch": 1.4569589702333066,
      "grad_norm": 0.2890816914521631,
      "learning_rate": 6.1139084041776665e-06,
      "loss": 0.0192,
      "step": 5433
    },
    {
      "epoch": 1.4572271386430677,
      "grad_norm": 0.34849557744052806,
      "learning_rate": 6.112387372975084e-06,
      "loss": 0.0287,
      "step": 5434
    },
    {
      "epoch": 1.4574953070528291,
      "grad_norm": 0.2816684641889638,
      "learning_rate": 6.110866233462296e-06,
      "loss": 0.0262,
      "step": 5435
    },
    {
      "epoch": 1.4577634754625906,
      "grad_norm": 0.3456625339070116,
      "learning_rate": 6.109344985787413e-06,
      "loss": 0.0246,
      "step": 5436
    },
    {
      "epoch": 1.458031643872352,
      "grad_norm": 0.28936340843726444,
      "learning_rate": 6.1078236300985546e-06,
      "loss": 0.021,
      "step": 5437
    },
    {
      "epoch": 1.4582998122821131,
      "grad_norm": 0.4005875167768937,
      "learning_rate": 6.10630216654385e-06,
      "loss": 0.0326,
      "step": 5438
    },
    {
      "epoch": 1.4585679806918745,
      "grad_norm": 0.2900700953682369,
      "learning_rate": 6.10478059527144e-06,
      "loss": 0.0265,
      "step": 5439
    },
    {
      "epoch": 1.4588361491016357,
      "grad_norm": 0.24161069917715003,
      "learning_rate": 6.103258916429476e-06,
      "loss": 0.0196,
      "step": 5440
    },
    {
      "epoch": 1.4591043175113971,
      "grad_norm": 0.24877202089521064,
      "learning_rate": 6.101737130166122e-06,
      "loss": 0.0218,
      "step": 5441
    },
    {
      "epoch": 1.4593724859211585,
      "grad_norm": 0.3376080151510017,
      "learning_rate": 6.100215236629549e-06,
      "loss": 0.0262,
      "step": 5442
    },
    {
      "epoch": 1.45964065433092,
      "grad_norm": 0.30126873974226254,
      "learning_rate": 6.098693235967938e-06,
      "loss": 0.0257,
      "step": 5443
    },
    {
      "epoch": 1.4599088227406811,
      "grad_norm": 0.40290923525559247,
      "learning_rate": 6.097171128329486e-06,
      "loss": 0.0301,
      "step": 5444
    },
    {
      "epoch": 1.4601769911504425,
      "grad_norm": 0.21616161716749552,
      "learning_rate": 6.095648913862391e-06,
      "loss": 0.0166,
      "step": 5445
    },
    {
      "epoch": 1.4604451595602037,
      "grad_norm": 0.2706419875647318,
      "learning_rate": 6.094126592714871e-06,
      "loss": 0.0231,
      "step": 5446
    },
    {
      "epoch": 1.4607133279699651,
      "grad_norm": 0.28580643931958816,
      "learning_rate": 6.09260416503515e-06,
      "loss": 0.0225,
      "step": 5447
    },
    {
      "epoch": 1.4609814963797265,
      "grad_norm": 0.29414059145657817,
      "learning_rate": 6.091081630971462e-06,
      "loss": 0.0248,
      "step": 5448
    },
    {
      "epoch": 1.461249664789488,
      "grad_norm": 0.3957696831297019,
      "learning_rate": 6.0895589906720535e-06,
      "loss": 0.0206,
      "step": 5449
    },
    {
      "epoch": 1.461517833199249,
      "grad_norm": 0.2622825582085395,
      "learning_rate": 6.088036244285178e-06,
      "loss": 0.0212,
      "step": 5450
    },
    {
      "epoch": 1.4617860016090105,
      "grad_norm": 0.41511455823834637,
      "learning_rate": 6.086513391959101e-06,
      "loss": 0.0263,
      "step": 5451
    },
    {
      "epoch": 1.4620541700187717,
      "grad_norm": 0.24889719476700797,
      "learning_rate": 6.0849904338421e-06,
      "loss": 0.0207,
      "step": 5452
    },
    {
      "epoch": 1.462322338428533,
      "grad_norm": 0.22872603468769862,
      "learning_rate": 6.083467370082462e-06,
      "loss": 0.0209,
      "step": 5453
    },
    {
      "epoch": 1.4625905068382945,
      "grad_norm": 0.29901523655128687,
      "learning_rate": 6.0819442008284825e-06,
      "loss": 0.0257,
      "step": 5454
    },
    {
      "epoch": 1.462858675248056,
      "grad_norm": 0.1790766409342946,
      "learning_rate": 6.0804209262284695e-06,
      "loss": 0.0158,
      "step": 5455
    },
    {
      "epoch": 1.463126843657817,
      "grad_norm": 0.2933995115633074,
      "learning_rate": 6.078897546430738e-06,
      "loss": 0.0267,
      "step": 5456
    },
    {
      "epoch": 1.4633950120675785,
      "grad_norm": 0.5929806845358147,
      "learning_rate": 6.077374061583618e-06,
      "loss": 0.03,
      "step": 5457
    },
    {
      "epoch": 1.4636631804773397,
      "grad_norm": 0.23504409614451224,
      "learning_rate": 6.075850471835445e-06,
      "loss": 0.0193,
      "step": 5458
    },
    {
      "epoch": 1.463931348887101,
      "grad_norm": 0.289579617116702,
      "learning_rate": 6.0743267773345675e-06,
      "loss": 0.0262,
      "step": 5459
    },
    {
      "epoch": 1.4641995172968625,
      "grad_norm": 0.2889018562724464,
      "learning_rate": 6.072802978229345e-06,
      "loss": 0.0242,
      "step": 5460
    },
    {
      "epoch": 1.464467685706624,
      "grad_norm": 0.22262130688063334,
      "learning_rate": 6.071279074668144e-06,
      "loss": 0.0212,
      "step": 5461
    },
    {
      "epoch": 1.464735854116385,
      "grad_norm": 0.2735253346267087,
      "learning_rate": 6.069755066799344e-06,
      "loss": 0.0253,
      "step": 5462
    },
    {
      "epoch": 1.4650040225261465,
      "grad_norm": 0.2665191785600146,
      "learning_rate": 6.068230954771334e-06,
      "loss": 0.024,
      "step": 5463
    },
    {
      "epoch": 1.4652721909359077,
      "grad_norm": 0.6899917675183308,
      "learning_rate": 6.066706738732511e-06,
      "loss": 0.0252,
      "step": 5464
    },
    {
      "epoch": 1.465540359345669,
      "grad_norm": 0.34312010097158707,
      "learning_rate": 6.065182418831286e-06,
      "loss": 0.0256,
      "step": 5465
    },
    {
      "epoch": 1.4658085277554305,
      "grad_norm": 0.5130515776071124,
      "learning_rate": 6.063657995216076e-06,
      "loss": 0.0245,
      "step": 5466
    },
    {
      "epoch": 1.4660766961651919,
      "grad_norm": 4.286859314422194,
      "learning_rate": 6.062133468035311e-06,
      "loss": 0.031,
      "step": 5467
    },
    {
      "epoch": 1.466344864574953,
      "grad_norm": 0.7128010720138436,
      "learning_rate": 6.06060883743743e-06,
      "loss": 0.0231,
      "step": 5468
    },
    {
      "epoch": 1.4666130329847145,
      "grad_norm": 0.17554618634067015,
      "learning_rate": 6.059084103570882e-06,
      "loss": 0.0132,
      "step": 5469
    },
    {
      "epoch": 1.4668812013944756,
      "grad_norm": 2.7022875264643766,
      "learning_rate": 6.057559266584128e-06,
      "loss": 0.0235,
      "step": 5470
    },
    {
      "epoch": 1.467149369804237,
      "grad_norm": 0.3911127454031509,
      "learning_rate": 6.056034326625635e-06,
      "loss": 0.0329,
      "step": 5471
    },
    {
      "epoch": 1.4674175382139985,
      "grad_norm": 0.41996558270268564,
      "learning_rate": 6.054509283843884e-06,
      "loss": 0.0345,
      "step": 5472
    },
    {
      "epoch": 1.4676857066237599,
      "grad_norm": 3.7022715726398485,
      "learning_rate": 6.052984138387364e-06,
      "loss": 0.0402,
      "step": 5473
    },
    {
      "epoch": 1.467953875033521,
      "grad_norm": 0.2835899790683567,
      "learning_rate": 6.051458890404573e-06,
      "loss": 0.0259,
      "step": 5474
    },
    {
      "epoch": 1.4682220434432824,
      "grad_norm": 0.24990812161338463,
      "learning_rate": 6.0499335400440216e-06,
      "loss": 0.0215,
      "step": 5475
    },
    {
      "epoch": 1.4684902118530436,
      "grad_norm": 0.23447965147836516,
      "learning_rate": 6.048408087454228e-06,
      "loss": 0.0231,
      "step": 5476
    },
    {
      "epoch": 1.468758380262805,
      "grad_norm": 0.8289793989645695,
      "learning_rate": 6.046882532783724e-06,
      "loss": 0.031,
      "step": 5477
    },
    {
      "epoch": 1.4690265486725664,
      "grad_norm": 0.2534794667819891,
      "learning_rate": 6.045356876181045e-06,
      "loss": 0.0289,
      "step": 5478
    },
    {
      "epoch": 1.4692947170823278,
      "grad_norm": 0.24948837704737659,
      "learning_rate": 6.043831117794746e-06,
      "loss": 0.026,
      "step": 5479
    },
    {
      "epoch": 1.469562885492089,
      "grad_norm": 0.20815076741510852,
      "learning_rate": 6.04230525777338e-06,
      "loss": 0.0175,
      "step": 5480
    },
    {
      "epoch": 1.4698310539018504,
      "grad_norm": 0.4398409332815767,
      "learning_rate": 6.040779296265519e-06,
      "loss": 0.0241,
      "step": 5481
    },
    {
      "epoch": 1.4700992223116116,
      "grad_norm": 0.23037170449282124,
      "learning_rate": 6.03925323341974e-06,
      "loss": 0.0186,
      "step": 5482
    },
    {
      "epoch": 1.470367390721373,
      "grad_norm": 0.2108856480755226,
      "learning_rate": 6.037727069384633e-06,
      "loss": 0.0152,
      "step": 5483
    },
    {
      "epoch": 1.4706355591311344,
      "grad_norm": 0.2675989913269997,
      "learning_rate": 6.036200804308798e-06,
      "loss": 0.0282,
      "step": 5484
    },
    {
      "epoch": 1.4709037275408956,
      "grad_norm": 0.3024248068631541,
      "learning_rate": 6.034674438340842e-06,
      "loss": 0.0313,
      "step": 5485
    },
    {
      "epoch": 1.471171895950657,
      "grad_norm": 0.2725665857246259,
      "learning_rate": 6.033147971629381e-06,
      "loss": 0.0254,
      "step": 5486
    },
    {
      "epoch": 1.4714400643604184,
      "grad_norm": 0.21102537000935545,
      "learning_rate": 6.031621404323046e-06,
      "loss": 0.0159,
      "step": 5487
    },
    {
      "epoch": 1.4717082327701796,
      "grad_norm": 0.3055529882076157,
      "learning_rate": 6.030094736570472e-06,
      "loss": 0.0264,
      "step": 5488
    },
    {
      "epoch": 1.471976401179941,
      "grad_norm": 0.2306060368839079,
      "learning_rate": 6.02856796852031e-06,
      "loss": 0.0185,
      "step": 5489
    },
    {
      "epoch": 1.4722445695897024,
      "grad_norm": 0.2957048053302528,
      "learning_rate": 6.027041100321216e-06,
      "loss": 0.0224,
      "step": 5490
    },
    {
      "epoch": 1.4725127379994636,
      "grad_norm": 0.22119621726835992,
      "learning_rate": 6.025514132121857e-06,
      "loss": 0.0141,
      "step": 5491
    },
    {
      "epoch": 1.472780906409225,
      "grad_norm": 0.3291149477952848,
      "learning_rate": 6.0239870640709084e-06,
      "loss": 0.0305,
      "step": 5492
    },
    {
      "epoch": 1.4730490748189864,
      "grad_norm": 0.48902882079114485,
      "learning_rate": 6.022459896317058e-06,
      "loss": 0.0341,
      "step": 5493
    },
    {
      "epoch": 1.4733172432287476,
      "grad_norm": 0.5681868465548964,
      "learning_rate": 6.020932629009003e-06,
      "loss": 0.0236,
      "step": 5494
    },
    {
      "epoch": 1.473585411638509,
      "grad_norm": 0.3127715697611482,
      "learning_rate": 6.019405262295448e-06,
      "loss": 0.0243,
      "step": 5495
    },
    {
      "epoch": 1.4738535800482704,
      "grad_norm": 0.31903810291132906,
      "learning_rate": 6.017877796325109e-06,
      "loss": 0.0233,
      "step": 5496
    },
    {
      "epoch": 1.4741217484580316,
      "grad_norm": 0.23545837119695515,
      "learning_rate": 6.016350231246713e-06,
      "loss": 0.019,
      "step": 5497
    },
    {
      "epoch": 1.474389916867793,
      "grad_norm": 0.3950414886658713,
      "learning_rate": 6.014822567208991e-06,
      "loss": 0.0265,
      "step": 5498
    },
    {
      "epoch": 1.4746580852775544,
      "grad_norm": 0.26107162474385875,
      "learning_rate": 6.013294804360689e-06,
      "loss": 0.0276,
      "step": 5499
    },
    {
      "epoch": 1.4749262536873156,
      "grad_norm": 0.2726101585297778,
      "learning_rate": 6.011766942850565e-06,
      "loss": 0.0278,
      "step": 5500
    },
    {
      "epoch": 1.475194422097077,
      "grad_norm": 0.49028381985058334,
      "learning_rate": 6.010238982827379e-06,
      "loss": 0.0233,
      "step": 5501
    },
    {
      "epoch": 1.4754625905068384,
      "grad_norm": 0.2728249185938734,
      "learning_rate": 6.008710924439906e-06,
      "loss": 0.0201,
      "step": 5502
    },
    {
      "epoch": 1.4757307589165995,
      "grad_norm": 0.3941689004766452,
      "learning_rate": 6.0071827678369275e-06,
      "loss": 0.0324,
      "step": 5503
    },
    {
      "epoch": 1.475998927326361,
      "grad_norm": 0.2826319405596248,
      "learning_rate": 6.005654513167236e-06,
      "loss": 0.0234,
      "step": 5504
    },
    {
      "epoch": 1.4762670957361224,
      "grad_norm": 0.26923757130186193,
      "learning_rate": 6.004126160579636e-06,
      "loss": 0.0285,
      "step": 5505
    },
    {
      "epoch": 1.4765352641458835,
      "grad_norm": 1.1364890871828919,
      "learning_rate": 6.002597710222937e-06,
      "loss": 0.027,
      "step": 5506
    },
    {
      "epoch": 1.476803432555645,
      "grad_norm": 0.32964184896064597,
      "learning_rate": 6.001069162245961e-06,
      "loss": 0.0233,
      "step": 5507
    },
    {
      "epoch": 1.4770716009654064,
      "grad_norm": 0.37049448029540216,
      "learning_rate": 5.99954051679754e-06,
      "loss": 0.0385,
      "step": 5508
    },
    {
      "epoch": 1.4773397693751675,
      "grad_norm": 0.3298870980573362,
      "learning_rate": 5.99801177402651e-06,
      "loss": 0.0272,
      "step": 5509
    },
    {
      "epoch": 1.477607937784929,
      "grad_norm": 0.27298646530649634,
      "learning_rate": 5.9964829340817245e-06,
      "loss": 0.0226,
      "step": 5510
    },
    {
      "epoch": 1.4778761061946903,
      "grad_norm": 0.2075410674078432,
      "learning_rate": 5.9949539971120405e-06,
      "loss": 0.0232,
      "step": 5511
    },
    {
      "epoch": 1.4781442746044515,
      "grad_norm": 0.25537790265889376,
      "learning_rate": 5.9934249632663284e-06,
      "loss": 0.022,
      "step": 5512
    },
    {
      "epoch": 1.478412443014213,
      "grad_norm": 0.2730132910531634,
      "learning_rate": 5.991895832693465e-06,
      "loss": 0.0247,
      "step": 5513
    },
    {
      "epoch": 1.4786806114239743,
      "grad_norm": 0.3094383438004259,
      "learning_rate": 5.990366605542337e-06,
      "loss": 0.0264,
      "step": 5514
    },
    {
      "epoch": 1.4789487798337355,
      "grad_norm": 0.29112896158730067,
      "learning_rate": 5.988837281961843e-06,
      "loss": 0.0238,
      "step": 5515
    },
    {
      "epoch": 1.479216948243497,
      "grad_norm": 0.29883686047691954,
      "learning_rate": 5.987307862100889e-06,
      "loss": 0.0229,
      "step": 5516
    },
    {
      "epoch": 1.4794851166532583,
      "grad_norm": 0.2555849103254526,
      "learning_rate": 5.9857783461083876e-06,
      "loss": 0.0228,
      "step": 5517
    },
    {
      "epoch": 1.4797532850630195,
      "grad_norm": 0.24502050290903835,
      "learning_rate": 5.9842487341332664e-06,
      "loss": 0.0179,
      "step": 5518
    },
    {
      "epoch": 1.480021453472781,
      "grad_norm": 0.28595293924555465,
      "learning_rate": 5.982719026324459e-06,
      "loss": 0.0247,
      "step": 5519
    },
    {
      "epoch": 1.4802896218825423,
      "grad_norm": 0.3072219668312463,
      "learning_rate": 5.98118922283091e-06,
      "loss": 0.0286,
      "step": 5520
    },
    {
      "epoch": 1.4805577902923035,
      "grad_norm": 0.2894312890121414,
      "learning_rate": 5.97965932380157e-06,
      "loss": 0.0299,
      "step": 5521
    },
    {
      "epoch": 1.480825958702065,
      "grad_norm": 0.2665255298877472,
      "learning_rate": 5.978129329385404e-06,
      "loss": 0.024,
      "step": 5522
    },
    {
      "epoch": 1.4810941271118263,
      "grad_norm": 0.2317873720728504,
      "learning_rate": 5.976599239731381e-06,
      "loss": 0.0193,
      "step": 5523
    },
    {
      "epoch": 1.4813622955215875,
      "grad_norm": 0.3793191612637848,
      "learning_rate": 5.975069054988484e-06,
      "loss": 0.0393,
      "step": 5524
    },
    {
      "epoch": 1.481630463931349,
      "grad_norm": 0.23471341090092424,
      "learning_rate": 5.9735387753057e-06,
      "loss": 0.0191,
      "step": 5525
    },
    {
      "epoch": 1.4818986323411103,
      "grad_norm": 0.20109425842193432,
      "learning_rate": 5.972008400832031e-06,
      "loss": 0.0195,
      "step": 5526
    },
    {
      "epoch": 1.4821668007508715,
      "grad_norm": 0.2399398648672672,
      "learning_rate": 5.970477931716486e-06,
      "loss": 0.0172,
      "step": 5527
    },
    {
      "epoch": 1.4824349691606329,
      "grad_norm": 0.2544584432268763,
      "learning_rate": 5.96894736810808e-06,
      "loss": 0.024,
      "step": 5528
    },
    {
      "epoch": 1.482703137570394,
      "grad_norm": 0.23286021521581438,
      "learning_rate": 5.967416710155842e-06,
      "loss": 0.0256,
      "step": 5529
    },
    {
      "epoch": 1.4829713059801555,
      "grad_norm": 0.29833414152397497,
      "learning_rate": 5.965885958008807e-06,
      "loss": 0.0241,
      "step": 5530
    },
    {
      "epoch": 1.4832394743899169,
      "grad_norm": 0.27925084043506393,
      "learning_rate": 5.964355111816019e-06,
      "loss": 0.0221,
      "step": 5531
    },
    {
      "epoch": 1.4835076427996783,
      "grad_norm": 0.22182479224317272,
      "learning_rate": 5.962824171726535e-06,
      "loss": 0.0134,
      "step": 5532
    },
    {
      "epoch": 1.4837758112094395,
      "grad_norm": 0.28203517020552366,
      "learning_rate": 5.961293137889421e-06,
      "loss": 0.0293,
      "step": 5533
    },
    {
      "epoch": 1.4840439796192009,
      "grad_norm": 0.23097937484173658,
      "learning_rate": 5.959762010453744e-06,
      "loss": 0.0153,
      "step": 5534
    },
    {
      "epoch": 1.484312148028962,
      "grad_norm": 0.40418503266438394,
      "learning_rate": 5.9582307895685876e-06,
      "loss": 0.0228,
      "step": 5535
    },
    {
      "epoch": 1.4845803164387235,
      "grad_norm": 0.3130813403330914,
      "learning_rate": 5.956699475383042e-06,
      "loss": 0.0226,
      "step": 5536
    },
    {
      "epoch": 1.4848484848484849,
      "grad_norm": 0.30087008475163485,
      "learning_rate": 5.95516806804621e-06,
      "loss": 0.0235,
      "step": 5537
    },
    {
      "epoch": 1.4851166532582463,
      "grad_norm": 0.2687948291377881,
      "learning_rate": 5.953636567707199e-06,
      "loss": 0.0194,
      "step": 5538
    },
    {
      "epoch": 1.4853848216680074,
      "grad_norm": 0.38209765853558325,
      "learning_rate": 5.952104974515128e-06,
      "loss": 0.0297,
      "step": 5539
    },
    {
      "epoch": 1.4856529900777689,
      "grad_norm": 0.2511443524149898,
      "learning_rate": 5.950573288619123e-06,
      "loss": 0.0205,
      "step": 5540
    },
    {
      "epoch": 1.48592115848753,
      "grad_norm": 0.44062224813241213,
      "learning_rate": 5.94904151016832e-06,
      "loss": 0.0325,
      "step": 5541
    },
    {
      "epoch": 1.4861893268972914,
      "grad_norm": 0.3974100542224825,
      "learning_rate": 5.947509639311865e-06,
      "loss": 0.0217,
      "step": 5542
    },
    {
      "epoch": 1.4864574953070528,
      "grad_norm": 0.27743962244039516,
      "learning_rate": 5.9459776761989115e-06,
      "loss": 0.0274,
      "step": 5543
    },
    {
      "epoch": 1.4867256637168142,
      "grad_norm": 0.2989258273654116,
      "learning_rate": 5.944445620978625e-06,
      "loss": 0.0262,
      "step": 5544
    },
    {
      "epoch": 1.4869938321265754,
      "grad_norm": 0.3357574042713301,
      "learning_rate": 5.942913473800176e-06,
      "loss": 0.0297,
      "step": 5545
    },
    {
      "epoch": 1.4872620005363368,
      "grad_norm": 0.22832780119863322,
      "learning_rate": 5.941381234812744e-06,
      "loss": 0.0216,
      "step": 5546
    },
    {
      "epoch": 1.487530168946098,
      "grad_norm": 0.2660018556288038,
      "learning_rate": 5.939848904165519e-06,
      "loss": 0.0237,
      "step": 5547
    },
    {
      "epoch": 1.4877983373558594,
      "grad_norm": 0.23718718397951402,
      "learning_rate": 5.938316482007704e-06,
      "loss": 0.0237,
      "step": 5548
    },
    {
      "epoch": 1.4880665057656208,
      "grad_norm": 0.285674650376869,
      "learning_rate": 5.936783968488503e-06,
      "loss": 0.0239,
      "step": 5549
    },
    {
      "epoch": 1.4883346741753822,
      "grad_norm": 0.2964334132191494,
      "learning_rate": 5.935251363757135e-06,
      "loss": 0.0257,
      "step": 5550
    },
    {
      "epoch": 1.4886028425851434,
      "grad_norm": 0.3013285505051643,
      "learning_rate": 5.9337186679628246e-06,
      "loss": 0.0339,
      "step": 5551
    },
    {
      "epoch": 1.4888710109949048,
      "grad_norm": 0.2467066730204887,
      "learning_rate": 5.932185881254806e-06,
      "loss": 0.0197,
      "step": 5552
    },
    {
      "epoch": 1.489139179404666,
      "grad_norm": 0.3065028837999952,
      "learning_rate": 5.930653003782323e-06,
      "loss": 0.0245,
      "step": 5553
    },
    {
      "epoch": 1.4894073478144274,
      "grad_norm": 0.27385106738870474,
      "learning_rate": 5.929120035694628e-06,
      "loss": 0.0235,
      "step": 5554
    },
    {
      "epoch": 1.4896755162241888,
      "grad_norm": 0.29229846149576494,
      "learning_rate": 5.927586977140982e-06,
      "loss": 0.0242,
      "step": 5555
    },
    {
      "epoch": 1.4899436846339502,
      "grad_norm": 0.2472474677187293,
      "learning_rate": 5.926053828270655e-06,
      "loss": 0.0204,
      "step": 5556
    },
    {
      "epoch": 1.4902118530437114,
      "grad_norm": 0.3041712883849561,
      "learning_rate": 5.9245205892329246e-06,
      "loss": 0.0202,
      "step": 5557
    },
    {
      "epoch": 1.4904800214534728,
      "grad_norm": 0.2523814381303282,
      "learning_rate": 5.92298726017708e-06,
      "loss": 0.0239,
      "step": 5558
    },
    {
      "epoch": 1.490748189863234,
      "grad_norm": 0.449100213482074,
      "learning_rate": 5.9214538412524155e-06,
      "loss": 0.0232,
      "step": 5559
    },
    {
      "epoch": 1.4910163582729954,
      "grad_norm": 0.2878046923559746,
      "learning_rate": 5.919920332608238e-06,
      "loss": 0.0253,
      "step": 5560
    },
    {
      "epoch": 1.4912845266827568,
      "grad_norm": 0.3028003999994938,
      "learning_rate": 5.918386734393858e-06,
      "loss": 0.021,
      "step": 5561
    },
    {
      "epoch": 1.4915526950925182,
      "grad_norm": 0.333529303522482,
      "learning_rate": 5.916853046758601e-06,
      "loss": 0.0281,
      "step": 5562
    },
    {
      "epoch": 1.4918208635022794,
      "grad_norm": 0.27108622047871817,
      "learning_rate": 5.915319269851798e-06,
      "loss": 0.0239,
      "step": 5563
    },
    {
      "epoch": 1.4920890319120408,
      "grad_norm": 0.34430232790234094,
      "learning_rate": 5.913785403822787e-06,
      "loss": 0.023,
      "step": 5564
    },
    {
      "epoch": 1.492357200321802,
      "grad_norm": 0.3878949200427505,
      "learning_rate": 5.912251448820917e-06,
      "loss": 0.0337,
      "step": 5565
    },
    {
      "epoch": 1.4926253687315634,
      "grad_norm": 0.27612430231844265,
      "learning_rate": 5.910717404995547e-06,
      "loss": 0.027,
      "step": 5566
    },
    {
      "epoch": 1.4928935371413248,
      "grad_norm": 0.26461288588866205,
      "learning_rate": 5.90918327249604e-06,
      "loss": 0.0209,
      "step": 5567
    },
    {
      "epoch": 1.4931617055510862,
      "grad_norm": 0.27116742805347177,
      "learning_rate": 5.907649051471771e-06,
      "loss": 0.0218,
      "step": 5568
    },
    {
      "epoch": 1.4934298739608474,
      "grad_norm": 0.2963346536338028,
      "learning_rate": 5.9061147420721236e-06,
      "loss": 0.0268,
      "step": 5569
    },
    {
      "epoch": 1.4936980423706088,
      "grad_norm": 0.24782413001509243,
      "learning_rate": 5.90458034444649e-06,
      "loss": 0.0283,
      "step": 5570
    },
    {
      "epoch": 1.49396621078037,
      "grad_norm": 0.25275120235781356,
      "learning_rate": 5.903045858744271e-06,
      "loss": 0.02,
      "step": 5571
    },
    {
      "epoch": 1.4942343791901314,
      "grad_norm": 0.27009394332073783,
      "learning_rate": 5.901511285114872e-06,
      "loss": 0.028,
      "step": 5572
    },
    {
      "epoch": 1.4945025475998928,
      "grad_norm": 0.2028918152272486,
      "learning_rate": 5.899976623707714e-06,
      "loss": 0.0197,
      "step": 5573
    },
    {
      "epoch": 1.4947707160096542,
      "grad_norm": 0.30533550779263857,
      "learning_rate": 5.898441874672221e-06,
      "loss": 0.0185,
      "step": 5574
    },
    {
      "epoch": 1.4950388844194153,
      "grad_norm": 0.2579388636731747,
      "learning_rate": 5.896907038157827e-06,
      "loss": 0.0239,
      "step": 5575
    },
    {
      "epoch": 1.4953070528291768,
      "grad_norm": 0.3906252456478026,
      "learning_rate": 5.89537211431398e-06,
      "loss": 0.0283,
      "step": 5576
    },
    {
      "epoch": 1.495575221238938,
      "grad_norm": 0.2818911707506467,
      "learning_rate": 5.893837103290124e-06,
      "loss": 0.0232,
      "step": 5577
    },
    {
      "epoch": 1.4958433896486993,
      "grad_norm": 0.33289947928629654,
      "learning_rate": 5.892302005235722e-06,
      "loss": 0.0215,
      "step": 5578
    },
    {
      "epoch": 1.4961115580584607,
      "grad_norm": 0.24732012346739682,
      "learning_rate": 5.8907668203002434e-06,
      "loss": 0.0249,
      "step": 5579
    },
    {
      "epoch": 1.4963797264682221,
      "grad_norm": 0.3905034457715331,
      "learning_rate": 5.8892315486331656e-06,
      "loss": 0.0235,
      "step": 5580
    },
    {
      "epoch": 1.4966478948779833,
      "grad_norm": 0.2378470667444129,
      "learning_rate": 5.887696190383971e-06,
      "loss": 0.021,
      "step": 5581
    },
    {
      "epoch": 1.4969160632877447,
      "grad_norm": 0.30400457357558186,
      "learning_rate": 5.886160745702157e-06,
      "loss": 0.0297,
      "step": 5582
    },
    {
      "epoch": 1.497184231697506,
      "grad_norm": 0.26324130492614334,
      "learning_rate": 5.884625214737224e-06,
      "loss": 0.0252,
      "step": 5583
    },
    {
      "epoch": 1.4974524001072673,
      "grad_norm": 0.24174751675002068,
      "learning_rate": 5.8830895976386805e-06,
      "loss": 0.0198,
      "step": 5584
    },
    {
      "epoch": 1.4977205685170287,
      "grad_norm": 0.8439808531499249,
      "learning_rate": 5.881553894556048e-06,
      "loss": 0.0202,
      "step": 5585
    },
    {
      "epoch": 1.4979887369267901,
      "grad_norm": 0.4142554070989203,
      "learning_rate": 5.880018105638854e-06,
      "loss": 0.026,
      "step": 5586
    },
    {
      "epoch": 1.4982569053365513,
      "grad_norm": 0.2762681565180677,
      "learning_rate": 5.878482231036633e-06,
      "loss": 0.0247,
      "step": 5587
    },
    {
      "epoch": 1.4985250737463127,
      "grad_norm": 0.25716566321710965,
      "learning_rate": 5.8769462708989314e-06,
      "loss": 0.0203,
      "step": 5588
    },
    {
      "epoch": 1.498793242156074,
      "grad_norm": 0.2709806663682435,
      "learning_rate": 5.875410225375296e-06,
      "loss": 0.0238,
      "step": 5589
    },
    {
      "epoch": 1.4990614105658353,
      "grad_norm": 0.42818817145258176,
      "learning_rate": 5.873874094615292e-06,
      "loss": 0.0392,
      "step": 5590
    },
    {
      "epoch": 1.4993295789755967,
      "grad_norm": 0.2681138253961643,
      "learning_rate": 5.872337878768489e-06,
      "loss": 0.0217,
      "step": 5591
    },
    {
      "epoch": 1.4995977473853581,
      "grad_norm": 0.2571445264272858,
      "learning_rate": 5.870801577984463e-06,
      "loss": 0.0203,
      "step": 5592
    },
    {
      "epoch": 1.4998659157951193,
      "grad_norm": 0.6055661381954486,
      "learning_rate": 5.869265192412799e-06,
      "loss": 0.0233,
      "step": 5593
    },
    {
      "epoch": 1.5001340842048807,
      "grad_norm": 0.315042041722011,
      "learning_rate": 5.867728722203091e-06,
      "loss": 0.0313,
      "step": 5594
    },
    {
      "epoch": 1.5004022526146419,
      "grad_norm": 0.23790638095399344,
      "learning_rate": 5.866192167504941e-06,
      "loss": 0.0239,
      "step": 5595
    },
    {
      "epoch": 1.5006704210244033,
      "grad_norm": 0.3875392155026543,
      "learning_rate": 5.864655528467958e-06,
      "loss": 0.0266,
      "step": 5596
    },
    {
      "epoch": 1.5009385894341647,
      "grad_norm": 0.38238177545662455,
      "learning_rate": 5.863118805241763e-06,
      "loss": 0.0188,
      "step": 5597
    },
    {
      "epoch": 1.501206757843926,
      "grad_norm": 0.26042034341708187,
      "learning_rate": 5.861581997975982e-06,
      "loss": 0.0227,
      "step": 5598
    },
    {
      "epoch": 1.5014749262536873,
      "grad_norm": 0.2799455269219071,
      "learning_rate": 5.860045106820248e-06,
      "loss": 0.0184,
      "step": 5599
    },
    {
      "epoch": 1.5017430946634487,
      "grad_norm": 0.24066719478373227,
      "learning_rate": 5.858508131924205e-06,
      "loss": 0.0203,
      "step": 5600
    },
    {
      "epoch": 1.5020112630732099,
      "grad_norm": 0.32610117000517014,
      "learning_rate": 5.8569710734375065e-06,
      "loss": 0.0235,
      "step": 5601
    },
    {
      "epoch": 1.5022794314829713,
      "grad_norm": 1.3333399260113405,
      "learning_rate": 5.855433931509808e-06,
      "loss": 0.0417,
      "step": 5602
    },
    {
      "epoch": 1.5025475998927327,
      "grad_norm": 0.3980826970751715,
      "learning_rate": 5.85389670629078e-06,
      "loss": 0.0229,
      "step": 5603
    },
    {
      "epoch": 1.502815768302494,
      "grad_norm": 0.20671233516502335,
      "learning_rate": 5.8523593979300965e-06,
      "loss": 0.0162,
      "step": 5604
    },
    {
      "epoch": 1.5030839367122553,
      "grad_norm": 0.23829169576732914,
      "learning_rate": 5.850822006577441e-06,
      "loss": 0.0213,
      "step": 5605
    },
    {
      "epoch": 1.5033521051220167,
      "grad_norm": 0.23039151914578046,
      "learning_rate": 5.849284532382507e-06,
      "loss": 0.018,
      "step": 5606
    },
    {
      "epoch": 1.5036202735317779,
      "grad_norm": 0.4364328325200387,
      "learning_rate": 5.84774697549499e-06,
      "loss": 0.0196,
      "step": 5607
    },
    {
      "epoch": 1.5038884419415393,
      "grad_norm": 0.3058393083448035,
      "learning_rate": 5.846209336064603e-06,
      "loss": 0.0214,
      "step": 5608
    },
    {
      "epoch": 1.5041566103513007,
      "grad_norm": 0.31919261816242034,
      "learning_rate": 5.844671614241059e-06,
      "loss": 0.0253,
      "step": 5609
    },
    {
      "epoch": 1.504424778761062,
      "grad_norm": 0.26337489246626966,
      "learning_rate": 5.843133810174082e-06,
      "loss": 0.0217,
      "step": 5610
    },
    {
      "epoch": 1.5046929471708232,
      "grad_norm": 0.19861726623475384,
      "learning_rate": 5.841595924013405e-06,
      "loss": 0.0142,
      "step": 5611
    },
    {
      "epoch": 1.5049611155805847,
      "grad_norm": 0.6083739116517318,
      "learning_rate": 5.840057955908767e-06,
      "loss": 0.0379,
      "step": 5612
    },
    {
      "epoch": 1.5052292839903458,
      "grad_norm": 0.2535710292704241,
      "learning_rate": 5.838519906009916e-06,
      "loss": 0.0351,
      "step": 5613
    },
    {
      "epoch": 1.5054974524001072,
      "grad_norm": 0.28833669027022435,
      "learning_rate": 5.836981774466609e-06,
      "loss": 0.0272,
      "step": 5614
    },
    {
      "epoch": 1.5057656208098686,
      "grad_norm": 0.2435681258709255,
      "learning_rate": 5.835443561428608e-06,
      "loss": 0.0283,
      "step": 5615
    },
    {
      "epoch": 1.50603378921963,
      "grad_norm": 0.2210469103699519,
      "learning_rate": 5.8339052670456855e-06,
      "loss": 0.0278,
      "step": 5616
    },
    {
      "epoch": 1.5063019576293912,
      "grad_norm": 0.2943264715787047,
      "learning_rate": 5.83236689146762e-06,
      "loss": 0.0297,
      "step": 5617
    },
    {
      "epoch": 1.5065701260391526,
      "grad_norm": 0.40096061083794754,
      "learning_rate": 5.830828434844203e-06,
      "loss": 0.0354,
      "step": 5618
    },
    {
      "epoch": 1.5068382944489138,
      "grad_norm": 0.21254842549853487,
      "learning_rate": 5.8292898973252246e-06,
      "loss": 0.0207,
      "step": 5619
    },
    {
      "epoch": 1.5071064628586752,
      "grad_norm": 0.22271247414305018,
      "learning_rate": 5.827751279060492e-06,
      "loss": 0.0222,
      "step": 5620
    },
    {
      "epoch": 1.5073746312684366,
      "grad_norm": 0.37835364344017663,
      "learning_rate": 5.826212580199814e-06,
      "loss": 0.0341,
      "step": 5621
    },
    {
      "epoch": 1.507642799678198,
      "grad_norm": 0.415501864027961,
      "learning_rate": 5.824673800893009e-06,
      "loss": 0.0192,
      "step": 5622
    },
    {
      "epoch": 1.5079109680879592,
      "grad_norm": 0.3331994448702936,
      "learning_rate": 5.823134941289908e-06,
      "loss": 0.0362,
      "step": 5623
    },
    {
      "epoch": 1.5081791364977206,
      "grad_norm": 0.22254154787461786,
      "learning_rate": 5.821596001540344e-06,
      "loss": 0.0276,
      "step": 5624
    },
    {
      "epoch": 1.5084473049074818,
      "grad_norm": 0.2868007354623883,
      "learning_rate": 5.820056981794156e-06,
      "loss": 0.025,
      "step": 5625
    },
    {
      "epoch": 1.5087154733172432,
      "grad_norm": 0.2787331439782924,
      "learning_rate": 5.818517882201199e-06,
      "loss": 0.023,
      "step": 5626
    },
    {
      "epoch": 1.5089836417270046,
      "grad_norm": 0.2369457390400773,
      "learning_rate": 5.8169787029113265e-06,
      "loss": 0.0195,
      "step": 5627
    },
    {
      "epoch": 1.509251810136766,
      "grad_norm": 0.2816889932192299,
      "learning_rate": 5.815439444074409e-06,
      "loss": 0.021,
      "step": 5628
    },
    {
      "epoch": 1.5095199785465272,
      "grad_norm": 0.3021655693505944,
      "learning_rate": 5.813900105840317e-06,
      "loss": 0.036,
      "step": 5629
    },
    {
      "epoch": 1.5097881469562886,
      "grad_norm": 0.21144757656709084,
      "learning_rate": 5.812360688358934e-06,
      "loss": 0.0143,
      "step": 5630
    },
    {
      "epoch": 1.5100563153660498,
      "grad_norm": 0.313374960497439,
      "learning_rate": 5.810821191780146e-06,
      "loss": 0.0241,
      "step": 5631
    },
    {
      "epoch": 1.5103244837758112,
      "grad_norm": 0.2551709234817503,
      "learning_rate": 5.8092816162538505e-06,
      "loss": 0.0253,
      "step": 5632
    },
    {
      "epoch": 1.5105926521855726,
      "grad_norm": 0.3686113233130328,
      "learning_rate": 5.807741961929954e-06,
      "loss": 0.0336,
      "step": 5633
    },
    {
      "epoch": 1.510860820595334,
      "grad_norm": 0.23682743778153306,
      "learning_rate": 5.8062022289583666e-06,
      "loss": 0.0202,
      "step": 5634
    },
    {
      "epoch": 1.5111289890050952,
      "grad_norm": 0.30746025824082746,
      "learning_rate": 5.804662417489009e-06,
      "loss": 0.0277,
      "step": 5635
    },
    {
      "epoch": 1.5113971574148566,
      "grad_norm": 0.4616300480713767,
      "learning_rate": 5.8031225276718086e-06,
      "loss": 0.0416,
      "step": 5636
    },
    {
      "epoch": 1.5116653258246178,
      "grad_norm": 0.25173439425613714,
      "learning_rate": 5.8015825596566975e-06,
      "loss": 0.0223,
      "step": 5637
    },
    {
      "epoch": 1.5119334942343792,
      "grad_norm": 0.3323393409565662,
      "learning_rate": 5.800042513593623e-06,
      "loss": 0.0335,
      "step": 5638
    },
    {
      "epoch": 1.5122016626441406,
      "grad_norm": 0.3322455504314431,
      "learning_rate": 5.798502389632532e-06,
      "loss": 0.0271,
      "step": 5639
    },
    {
      "epoch": 1.512469831053902,
      "grad_norm": 0.21080474798170887,
      "learning_rate": 5.7969621879233835e-06,
      "loss": 0.0179,
      "step": 5640
    },
    {
      "epoch": 1.5127379994636632,
      "grad_norm": 0.2691875061443105,
      "learning_rate": 5.7954219086161435e-06,
      "loss": 0.0262,
      "step": 5641
    },
    {
      "epoch": 1.5130061678734243,
      "grad_norm": 0.32874860876100154,
      "learning_rate": 5.793881551860785e-06,
      "loss": 0.034,
      "step": 5642
    },
    {
      "epoch": 1.5132743362831858,
      "grad_norm": 0.4384530856719144,
      "learning_rate": 5.792341117807284e-06,
      "loss": 0.0288,
      "step": 5643
    },
    {
      "epoch": 1.5135425046929472,
      "grad_norm": 0.23250615703674113,
      "learning_rate": 5.790800606605634e-06,
      "loss": 0.0194,
      "step": 5644
    },
    {
      "epoch": 1.5138106731027086,
      "grad_norm": 0.7509029240452141,
      "learning_rate": 5.789260018405829e-06,
      "loss": 0.0282,
      "step": 5645
    },
    {
      "epoch": 1.51407884151247,
      "grad_norm": 0.23355063075276608,
      "learning_rate": 5.7877193533578705e-06,
      "loss": 0.0237,
      "step": 5646
    },
    {
      "epoch": 1.5143470099222311,
      "grad_norm": 0.27121096490274155,
      "learning_rate": 5.786178611611771e-06,
      "loss": 0.0211,
      "step": 5647
    },
    {
      "epoch": 1.5146151783319923,
      "grad_norm": 0.24604209972002095,
      "learning_rate": 5.784637793317546e-06,
      "loss": 0.0262,
      "step": 5648
    },
    {
      "epoch": 1.5148833467417537,
      "grad_norm": 0.5360331439528703,
      "learning_rate": 5.783096898625223e-06,
      "loss": 0.0307,
      "step": 5649
    },
    {
      "epoch": 1.5151515151515151,
      "grad_norm": 0.2704959263174878,
      "learning_rate": 5.781555927684834e-06,
      "loss": 0.0253,
      "step": 5650
    },
    {
      "epoch": 1.5154196835612765,
      "grad_norm": 0.2884320982644645,
      "learning_rate": 5.7800148806464195e-06,
      "loss": 0.0246,
      "step": 5651
    },
    {
      "epoch": 1.515687851971038,
      "grad_norm": 0.921487710000508,
      "learning_rate": 5.778473757660027e-06,
      "loss": 0.0301,
      "step": 5652
    },
    {
      "epoch": 1.5159560203807991,
      "grad_norm": 0.25673545365398687,
      "learning_rate": 5.776932558875711e-06,
      "loss": 0.018,
      "step": 5653
    },
    {
      "epoch": 1.5162241887905603,
      "grad_norm": 0.24008968301534545,
      "learning_rate": 5.775391284443534e-06,
      "loss": 0.0177,
      "step": 5654
    },
    {
      "epoch": 1.5164923572003217,
      "grad_norm": 0.3521025599568721,
      "learning_rate": 5.773849934513568e-06,
      "loss": 0.0242,
      "step": 5655
    },
    {
      "epoch": 1.5167605256100831,
      "grad_norm": 0.27916171552616187,
      "learning_rate": 5.772308509235887e-06,
      "loss": 0.0189,
      "step": 5656
    },
    {
      "epoch": 1.5170286940198445,
      "grad_norm": 0.2960768244730077,
      "learning_rate": 5.770767008760577e-06,
      "loss": 0.0193,
      "step": 5657
    },
    {
      "epoch": 1.517296862429606,
      "grad_norm": 0.24117817444012207,
      "learning_rate": 5.769225433237731e-06,
      "loss": 0.0235,
      "step": 5658
    },
    {
      "epoch": 1.5175650308393671,
      "grad_norm": 0.2324261884603527,
      "learning_rate": 5.767683782817445e-06,
      "loss": 0.0221,
      "step": 5659
    },
    {
      "epoch": 1.5178331992491283,
      "grad_norm": 0.2751068012591845,
      "learning_rate": 5.7661420576498286e-06,
      "loss": 0.0239,
      "step": 5660
    },
    {
      "epoch": 1.5181013676588897,
      "grad_norm": 0.292053080152648,
      "learning_rate": 5.764600257884994e-06,
      "loss": 0.0289,
      "step": 5661
    },
    {
      "epoch": 1.518369536068651,
      "grad_norm": 0.28062267121928614,
      "learning_rate": 5.763058383673061e-06,
      "loss": 0.0202,
      "step": 5662
    },
    {
      "epoch": 1.5186377044784125,
      "grad_norm": 0.7502887509083108,
      "learning_rate": 5.7615164351641595e-06,
      "loss": 0.0267,
      "step": 5663
    },
    {
      "epoch": 1.518905872888174,
      "grad_norm": 0.30115760173570805,
      "learning_rate": 5.7599744125084254e-06,
      "loss": 0.0226,
      "step": 5664
    },
    {
      "epoch": 1.519174041297935,
      "grad_norm": 0.2997687071542531,
      "learning_rate": 5.758432315855998e-06,
      "loss": 0.0284,
      "step": 5665
    },
    {
      "epoch": 1.5194422097076963,
      "grad_norm": 0.2581725741934641,
      "learning_rate": 5.756890145357034e-06,
      "loss": 0.0217,
      "step": 5666
    },
    {
      "epoch": 1.5197103781174577,
      "grad_norm": 0.2123244143198249,
      "learning_rate": 5.755347901161683e-06,
      "loss": 0.0157,
      "step": 5667
    },
    {
      "epoch": 1.519978546527219,
      "grad_norm": 0.4653688411725856,
      "learning_rate": 5.7538055834201126e-06,
      "loss": 0.0399,
      "step": 5668
    },
    {
      "epoch": 1.5202467149369805,
      "grad_norm": 0.2281316463828835,
      "learning_rate": 5.752263192282494e-06,
      "loss": 0.0144,
      "step": 5669
    },
    {
      "epoch": 1.520514883346742,
      "grad_norm": 0.27445750167100974,
      "learning_rate": 5.750720727899005e-06,
      "loss": 0.0257,
      "step": 5670
    },
    {
      "epoch": 1.520783051756503,
      "grad_norm": 0.4169519438922934,
      "learning_rate": 5.7491781904198325e-06,
      "loss": 0.0208,
      "step": 5671
    },
    {
      "epoch": 1.5210512201662643,
      "grad_norm": 0.19831896585244732,
      "learning_rate": 5.74763557999517e-06,
      "loss": 0.0207,
      "step": 5672
    },
    {
      "epoch": 1.5213193885760257,
      "grad_norm": 0.3035046033768232,
      "learning_rate": 5.746092896775215e-06,
      "loss": 0.0263,
      "step": 5673
    },
    {
      "epoch": 1.521587556985787,
      "grad_norm": 0.2726092511200985,
      "learning_rate": 5.744550140910174e-06,
      "loss": 0.0278,
      "step": 5674
    },
    {
      "epoch": 1.5218557253955485,
      "grad_norm": 0.3546267777063928,
      "learning_rate": 5.743007312550262e-06,
      "loss": 0.0308,
      "step": 5675
    },
    {
      "epoch": 1.5221238938053099,
      "grad_norm": 0.2283742120629325,
      "learning_rate": 5.741464411845703e-06,
      "loss": 0.0212,
      "step": 5676
    },
    {
      "epoch": 1.522392062215071,
      "grad_norm": 0.21529046878764158,
      "learning_rate": 5.739921438946723e-06,
      "loss": 0.019,
      "step": 5677
    },
    {
      "epoch": 1.5226602306248322,
      "grad_norm": 0.23543639597915197,
      "learning_rate": 5.738378394003555e-06,
      "loss": 0.0233,
      "step": 5678
    },
    {
      "epoch": 1.5229283990345936,
      "grad_norm": 0.24227341971156127,
      "learning_rate": 5.736835277166446e-06,
      "loss": 0.0309,
      "step": 5679
    },
    {
      "epoch": 1.523196567444355,
      "grad_norm": 0.22815138996685275,
      "learning_rate": 5.735292088585639e-06,
      "loss": 0.0295,
      "step": 5680
    },
    {
      "epoch": 1.5234647358541165,
      "grad_norm": 0.27172423193096673,
      "learning_rate": 5.733748828411395e-06,
      "loss": 0.0207,
      "step": 5681
    },
    {
      "epoch": 1.5237329042638779,
      "grad_norm": 0.3760921854553939,
      "learning_rate": 5.732205496793975e-06,
      "loss": 0.0295,
      "step": 5682
    },
    {
      "epoch": 1.524001072673639,
      "grad_norm": 0.27021373041046626,
      "learning_rate": 5.730662093883651e-06,
      "loss": 0.025,
      "step": 5683
    },
    {
      "epoch": 1.5242692410834002,
      "grad_norm": 0.24832106154854647,
      "learning_rate": 5.729118619830698e-06,
      "loss": 0.0181,
      "step": 5684
    },
    {
      "epoch": 1.5245374094931616,
      "grad_norm": 0.2657935891027441,
      "learning_rate": 5.727575074785402e-06,
      "loss": 0.0245,
      "step": 5685
    },
    {
      "epoch": 1.524805577902923,
      "grad_norm": 0.2979300222517417,
      "learning_rate": 5.726031458898052e-06,
      "loss": 0.0208,
      "step": 5686
    },
    {
      "epoch": 1.5250737463126844,
      "grad_norm": 0.31210219315826204,
      "learning_rate": 5.724487772318945e-06,
      "loss": 0.0195,
      "step": 5687
    },
    {
      "epoch": 1.5253419147224458,
      "grad_norm": 0.2204749999575347,
      "learning_rate": 5.722944015198389e-06,
      "loss": 0.0214,
      "step": 5688
    },
    {
      "epoch": 1.525610083132207,
      "grad_norm": 0.30798072545184324,
      "learning_rate": 5.721400187686692e-06,
      "loss": 0.0193,
      "step": 5689
    },
    {
      "epoch": 1.5258782515419682,
      "grad_norm": 0.2463888221385075,
      "learning_rate": 5.719856289934175e-06,
      "loss": 0.0253,
      "step": 5690
    },
    {
      "epoch": 1.5261464199517296,
      "grad_norm": 0.2604091964240948,
      "learning_rate": 5.7183123220911615e-06,
      "loss": 0.0264,
      "step": 5691
    },
    {
      "epoch": 1.526414588361491,
      "grad_norm": 0.3154452037396957,
      "learning_rate": 5.716768284307984e-06,
      "loss": 0.0323,
      "step": 5692
    },
    {
      "epoch": 1.5266827567712524,
      "grad_norm": 0.25747267193062306,
      "learning_rate": 5.715224176734983e-06,
      "loss": 0.0278,
      "step": 5693
    },
    {
      "epoch": 1.5269509251810138,
      "grad_norm": 0.27490496500600864,
      "learning_rate": 5.7136799995225024e-06,
      "loss": 0.0221,
      "step": 5694
    },
    {
      "epoch": 1.527219093590775,
      "grad_norm": 0.38894697480927437,
      "learning_rate": 5.712135752820894e-06,
      "loss": 0.0254,
      "step": 5695
    },
    {
      "epoch": 1.5274872620005362,
      "grad_norm": 0.307632248326358,
      "learning_rate": 5.7105914367805195e-06,
      "loss": 0.0267,
      "step": 5696
    },
    {
      "epoch": 1.5277554304102976,
      "grad_norm": 0.23844044215938706,
      "learning_rate": 5.7090470515517424e-06,
      "loss": 0.0194,
      "step": 5697
    },
    {
      "epoch": 1.528023598820059,
      "grad_norm": 0.36420267485545255,
      "learning_rate": 5.707502597284936e-06,
      "loss": 0.0211,
      "step": 5698
    },
    {
      "epoch": 1.5282917672298204,
      "grad_norm": 0.44312382618731344,
      "learning_rate": 5.705958074130481e-06,
      "loss": 0.0364,
      "step": 5699
    },
    {
      "epoch": 1.5285599356395818,
      "grad_norm": 0.28332776399373194,
      "learning_rate": 5.704413482238763e-06,
      "loss": 0.0245,
      "step": 5700
    },
    {
      "epoch": 1.528828104049343,
      "grad_norm": 0.3052448684382652,
      "learning_rate": 5.7028688217601735e-06,
      "loss": 0.0358,
      "step": 5701
    },
    {
      "epoch": 1.5290962724591042,
      "grad_norm": 0.24420424355556966,
      "learning_rate": 5.7013240928451115e-06,
      "loss": 0.0184,
      "step": 5702
    },
    {
      "epoch": 1.5293644408688656,
      "grad_norm": 0.5802673732170867,
      "learning_rate": 5.699779295643988e-06,
      "loss": 0.0319,
      "step": 5703
    },
    {
      "epoch": 1.529632609278627,
      "grad_norm": 0.3134531794316103,
      "learning_rate": 5.69823443030721e-06,
      "loss": 0.0214,
      "step": 5704
    },
    {
      "epoch": 1.5299007776883884,
      "grad_norm": 0.2663682275767274,
      "learning_rate": 5.6966894969852e-06,
      "loss": 0.0223,
      "step": 5705
    },
    {
      "epoch": 1.5301689460981498,
      "grad_norm": 0.2835338997274253,
      "learning_rate": 5.695144495828384e-06,
      "loss": 0.0198,
      "step": 5706
    },
    {
      "epoch": 1.530437114507911,
      "grad_norm": 0.2458120205201112,
      "learning_rate": 5.6935994269871934e-06,
      "loss": 0.0235,
      "step": 5707
    },
    {
      "epoch": 1.5307052829176722,
      "grad_norm": 0.449432823421994,
      "learning_rate": 5.692054290612067e-06,
      "loss": 0.0234,
      "step": 5708
    },
    {
      "epoch": 1.5309734513274336,
      "grad_norm": 0.21637138911498027,
      "learning_rate": 5.690509086853453e-06,
      "loss": 0.0179,
      "step": 5709
    },
    {
      "epoch": 1.531241619737195,
      "grad_norm": 0.25692754519217453,
      "learning_rate": 5.6889638158618025e-06,
      "loss": 0.0201,
      "step": 5710
    },
    {
      "epoch": 1.5315097881469564,
      "grad_norm": 0.3676727015795023,
      "learning_rate": 5.687418477787574e-06,
      "loss": 0.0242,
      "step": 5711
    },
    {
      "epoch": 1.5317779565567178,
      "grad_norm": 0.2868699579114758,
      "learning_rate": 5.685873072781233e-06,
      "loss": 0.0314,
      "step": 5712
    },
    {
      "epoch": 1.532046124966479,
      "grad_norm": 0.31423446852688425,
      "learning_rate": 5.684327600993251e-06,
      "loss": 0.0265,
      "step": 5713
    },
    {
      "epoch": 1.5323142933762401,
      "grad_norm": 0.22972983571210082,
      "learning_rate": 5.682782062574109e-06,
      "loss": 0.0223,
      "step": 5714
    },
    {
      "epoch": 1.5325824617860015,
      "grad_norm": 0.3799780691609827,
      "learning_rate": 5.68123645767429e-06,
      "loss": 0.0249,
      "step": 5715
    },
    {
      "epoch": 1.532850630195763,
      "grad_norm": 0.2944892685817284,
      "learning_rate": 5.679690786444286e-06,
      "loss": 0.023,
      "step": 5716
    },
    {
      "epoch": 1.5331187986055244,
      "grad_norm": 0.2873578055352119,
      "learning_rate": 5.678145049034595e-06,
      "loss": 0.027,
      "step": 5717
    },
    {
      "epoch": 1.5333869670152858,
      "grad_norm": 0.5063012146416181,
      "learning_rate": 5.67659924559572e-06,
      "loss": 0.0336,
      "step": 5718
    },
    {
      "epoch": 1.533655135425047,
      "grad_norm": 0.2884435442890642,
      "learning_rate": 5.675053376278172e-06,
      "loss": 0.0228,
      "step": 5719
    },
    {
      "epoch": 1.5339233038348081,
      "grad_norm": 0.2529400235046711,
      "learning_rate": 5.673507441232471e-06,
      "loss": 0.0243,
      "step": 5720
    },
    {
      "epoch": 1.5341914722445695,
      "grad_norm": 0.32425283569712937,
      "learning_rate": 5.671961440609139e-06,
      "loss": 0.0304,
      "step": 5721
    },
    {
      "epoch": 1.534459640654331,
      "grad_norm": 0.17700257998097885,
      "learning_rate": 5.670415374558703e-06,
      "loss": 0.0167,
      "step": 5722
    },
    {
      "epoch": 1.5347278090640923,
      "grad_norm": 0.20182553022854352,
      "learning_rate": 5.668869243231703e-06,
      "loss": 0.0215,
      "step": 5723
    },
    {
      "epoch": 1.5349959774738537,
      "grad_norm": 0.25712961024746317,
      "learning_rate": 5.66732304677868e-06,
      "loss": 0.022,
      "step": 5724
    },
    {
      "epoch": 1.535264145883615,
      "grad_norm": 0.3313476328880328,
      "learning_rate": 5.665776785350185e-06,
      "loss": 0.0235,
      "step": 5725
    },
    {
      "epoch": 1.535532314293376,
      "grad_norm": 0.2842534625051747,
      "learning_rate": 5.664230459096771e-06,
      "loss": 0.0181,
      "step": 5726
    },
    {
      "epoch": 1.5358004827031375,
      "grad_norm": 0.30080164173995977,
      "learning_rate": 5.662684068169002e-06,
      "loss": 0.0264,
      "step": 5727
    },
    {
      "epoch": 1.536068651112899,
      "grad_norm": 0.3369639033568627,
      "learning_rate": 5.661137612717443e-06,
      "loss": 0.0309,
      "step": 5728
    },
    {
      "epoch": 1.5363368195226603,
      "grad_norm": 0.3455896171981691,
      "learning_rate": 5.659591092892671e-06,
      "loss": 0.0255,
      "step": 5729
    },
    {
      "epoch": 1.5366049879324215,
      "grad_norm": 0.25439846952598577,
      "learning_rate": 5.658044508845265e-06,
      "loss": 0.0266,
      "step": 5730
    },
    {
      "epoch": 1.536873156342183,
      "grad_norm": 0.2645720142203955,
      "learning_rate": 5.656497860725813e-06,
      "loss": 0.0202,
      "step": 5731
    },
    {
      "epoch": 1.537141324751944,
      "grad_norm": 0.33940589095765233,
      "learning_rate": 5.654951148684906e-06,
      "loss": 0.0273,
      "step": 5732
    },
    {
      "epoch": 1.5374094931617055,
      "grad_norm": 0.21455109519307344,
      "learning_rate": 5.653404372873147e-06,
      "loss": 0.0226,
      "step": 5733
    },
    {
      "epoch": 1.537677661571467,
      "grad_norm": 0.22834359695674483,
      "learning_rate": 5.651857533441135e-06,
      "loss": 0.0253,
      "step": 5734
    },
    {
      "epoch": 1.5379458299812283,
      "grad_norm": 0.2387426409412802,
      "learning_rate": 5.6503106305394886e-06,
      "loss": 0.02,
      "step": 5735
    },
    {
      "epoch": 1.5382139983909895,
      "grad_norm": 0.28209534713580514,
      "learning_rate": 5.648763664318822e-06,
      "loss": 0.0254,
      "step": 5736
    },
    {
      "epoch": 1.538482166800751,
      "grad_norm": 0.2779312710459575,
      "learning_rate": 5.64721663492976e-06,
      "loss": 0.0182,
      "step": 5737
    },
    {
      "epoch": 1.538750335210512,
      "grad_norm": 0.22084880881381772,
      "learning_rate": 5.645669542522932e-06,
      "loss": 0.0255,
      "step": 5738
    },
    {
      "epoch": 1.5390185036202735,
      "grad_norm": 0.2351995122580828,
      "learning_rate": 5.644122387248975e-06,
      "loss": 0.02,
      "step": 5739
    },
    {
      "epoch": 1.5392866720300349,
      "grad_norm": 0.26840350213046954,
      "learning_rate": 5.6425751692585306e-06,
      "loss": 0.0232,
      "step": 5740
    },
    {
      "epoch": 1.5395548404397963,
      "grad_norm": 0.22445596732401996,
      "learning_rate": 5.64102788870225e-06,
      "loss": 0.0197,
      "step": 5741
    },
    {
      "epoch": 1.5398230088495575,
      "grad_norm": 0.2872718551966616,
      "learning_rate": 5.6394805457307845e-06,
      "loss": 0.0222,
      "step": 5742
    },
    {
      "epoch": 1.5400911772593189,
      "grad_norm": 0.27616924281798305,
      "learning_rate": 5.637933140494795e-06,
      "loss": 0.0249,
      "step": 5743
    },
    {
      "epoch": 1.54035934566908,
      "grad_norm": 0.28644014535857876,
      "learning_rate": 5.636385673144951e-06,
      "loss": 0.0207,
      "step": 5744
    },
    {
      "epoch": 1.5406275140788415,
      "grad_norm": 0.23731642144448914,
      "learning_rate": 5.634838143831923e-06,
      "loss": 0.0183,
      "step": 5745
    },
    {
      "epoch": 1.5408956824886029,
      "grad_norm": 0.2661190331253653,
      "learning_rate": 5.6332905527063894e-06,
      "loss": 0.0246,
      "step": 5746
    },
    {
      "epoch": 1.5411638508983643,
      "grad_norm": 0.2175919733529343,
      "learning_rate": 5.631742899919038e-06,
      "loss": 0.02,
      "step": 5747
    },
    {
      "epoch": 1.5414320193081255,
      "grad_norm": 0.21254866486342017,
      "learning_rate": 5.630195185620556e-06,
      "loss": 0.0179,
      "step": 5748
    },
    {
      "epoch": 1.5417001877178869,
      "grad_norm": 0.2884630363663017,
      "learning_rate": 5.6286474099616426e-06,
      "loss": 0.0224,
      "step": 5749
    },
    {
      "epoch": 1.541968356127648,
      "grad_norm": 0.270084307882223,
      "learning_rate": 5.627099573092999e-06,
      "loss": 0.0236,
      "step": 5750
    },
    {
      "epoch": 1.5422365245374094,
      "grad_norm": 0.23498673770513662,
      "learning_rate": 5.6255516751653376e-06,
      "loss": 0.0294,
      "step": 5751
    },
    {
      "epoch": 1.5425046929471709,
      "grad_norm": 0.3956886363237127,
      "learning_rate": 5.624003716329368e-06,
      "loss": 0.0259,
      "step": 5752
    },
    {
      "epoch": 1.5427728613569323,
      "grad_norm": 0.27750588818098615,
      "learning_rate": 5.622455696735814e-06,
      "loss": 0.0334,
      "step": 5753
    },
    {
      "epoch": 1.5430410297666934,
      "grad_norm": 0.2332957865599169,
      "learning_rate": 5.620907616535401e-06,
      "loss": 0.0234,
      "step": 5754
    },
    {
      "epoch": 1.5433091981764548,
      "grad_norm": 0.2049568464963472,
      "learning_rate": 5.619359475878863e-06,
      "loss": 0.0173,
      "step": 5755
    },
    {
      "epoch": 1.543577366586216,
      "grad_norm": 0.21491316073123515,
      "learning_rate": 5.617811274916934e-06,
      "loss": 0.0227,
      "step": 5756
    },
    {
      "epoch": 1.5438455349959774,
      "grad_norm": 0.2252422932601072,
      "learning_rate": 5.616263013800365e-06,
      "loss": 0.029,
      "step": 5757
    },
    {
      "epoch": 1.5441137034057388,
      "grad_norm": 0.20270289684955514,
      "learning_rate": 5.614714692679902e-06,
      "loss": 0.0212,
      "step": 5758
    },
    {
      "epoch": 1.5443818718155002,
      "grad_norm": 0.2855415760518405,
      "learning_rate": 5.6131663117063e-06,
      "loss": 0.0246,
      "step": 5759
    },
    {
      "epoch": 1.5446500402252614,
      "grad_norm": 0.3296676413496741,
      "learning_rate": 5.611617871030323e-06,
      "loss": 0.0276,
      "step": 5760
    },
    {
      "epoch": 1.5449182086350228,
      "grad_norm": 0.23198940027234094,
      "learning_rate": 5.610069370802737e-06,
      "loss": 0.019,
      "step": 5761
    },
    {
      "epoch": 1.545186377044784,
      "grad_norm": 0.18675903922071008,
      "learning_rate": 5.608520811174316e-06,
      "loss": 0.0175,
      "step": 5762
    },
    {
      "epoch": 1.5454545454545454,
      "grad_norm": 0.5241161496963793,
      "learning_rate": 5.606972192295841e-06,
      "loss": 0.0286,
      "step": 5763
    },
    {
      "epoch": 1.5457227138643068,
      "grad_norm": 0.35228370447127927,
      "learning_rate": 5.605423514318093e-06,
      "loss": 0.0184,
      "step": 5764
    },
    {
      "epoch": 1.5459908822740682,
      "grad_norm": 0.26790111357365193,
      "learning_rate": 5.603874777391866e-06,
      "loss": 0.0293,
      "step": 5765
    },
    {
      "epoch": 1.5462590506838294,
      "grad_norm": 0.2450524478404383,
      "learning_rate": 5.602325981667954e-06,
      "loss": 0.0188,
      "step": 5766
    },
    {
      "epoch": 1.5465272190935908,
      "grad_norm": 0.2432962404596528,
      "learning_rate": 5.6007771272971616e-06,
      "loss": 0.0264,
      "step": 5767
    },
    {
      "epoch": 1.546795387503352,
      "grad_norm": 0.25385082226850986,
      "learning_rate": 5.599228214430294e-06,
      "loss": 0.0242,
      "step": 5768
    },
    {
      "epoch": 1.5470635559131134,
      "grad_norm": 0.26047603848496403,
      "learning_rate": 5.597679243218168e-06,
      "loss": 0.0242,
      "step": 5769
    },
    {
      "epoch": 1.5473317243228748,
      "grad_norm": 0.29797685700009646,
      "learning_rate": 5.596130213811601e-06,
      "loss": 0.0276,
      "step": 5770
    },
    {
      "epoch": 1.5475998927326362,
      "grad_norm": 0.3054756105588386,
      "learning_rate": 5.594581126361415e-06,
      "loss": 0.0289,
      "step": 5771
    },
    {
      "epoch": 1.5478680611423974,
      "grad_norm": 0.22135795426264612,
      "learning_rate": 5.593031981018445e-06,
      "loss": 0.0206,
      "step": 5772
    },
    {
      "epoch": 1.5481362295521588,
      "grad_norm": 0.1848947166486402,
      "learning_rate": 5.5914827779335254e-06,
      "loss": 0.0178,
      "step": 5773
    },
    {
      "epoch": 1.54840439796192,
      "grad_norm": 0.283183884330318,
      "learning_rate": 5.589933517257497e-06,
      "loss": 0.0272,
      "step": 5774
    },
    {
      "epoch": 1.5486725663716814,
      "grad_norm": 0.36648650992183907,
      "learning_rate": 5.588384199141211e-06,
      "loss": 0.0224,
      "step": 5775
    },
    {
      "epoch": 1.5489407347814428,
      "grad_norm": 0.2775681882105126,
      "learning_rate": 5.586834823735515e-06,
      "loss": 0.0303,
      "step": 5776
    },
    {
      "epoch": 1.5492089031912042,
      "grad_norm": 0.3798152430855655,
      "learning_rate": 5.585285391191272e-06,
      "loss": 0.0399,
      "step": 5777
    },
    {
      "epoch": 1.5494770716009654,
      "grad_norm": 0.44720095353450573,
      "learning_rate": 5.583735901659343e-06,
      "loss": 0.0275,
      "step": 5778
    },
    {
      "epoch": 1.5497452400107268,
      "grad_norm": 0.2443377044649307,
      "learning_rate": 5.5821863552905995e-06,
      "loss": 0.0211,
      "step": 5779
    },
    {
      "epoch": 1.550013408420488,
      "grad_norm": 0.43419135380242213,
      "learning_rate": 5.580636752235917e-06,
      "loss": 0.028,
      "step": 5780
    },
    {
      "epoch": 1.5502815768302494,
      "grad_norm": 0.3313408070125449,
      "learning_rate": 5.579087092646176e-06,
      "loss": 0.0267,
      "step": 5781
    },
    {
      "epoch": 1.5505497452400108,
      "grad_norm": 0.265268100258148,
      "learning_rate": 5.57753737667226e-06,
      "loss": 0.0221,
      "step": 5782
    },
    {
      "epoch": 1.5508179136497722,
      "grad_norm": 0.18036619365878084,
      "learning_rate": 5.575987604465064e-06,
      "loss": 0.0203,
      "step": 5783
    },
    {
      "epoch": 1.5510860820595334,
      "grad_norm": 0.31588799872982937,
      "learning_rate": 5.574437776175484e-06,
      "loss": 0.0145,
      "step": 5784
    },
    {
      "epoch": 1.5513542504692948,
      "grad_norm": 0.2515203888713182,
      "learning_rate": 5.572887891954424e-06,
      "loss": 0.0222,
      "step": 5785
    },
    {
      "epoch": 1.551622418879056,
      "grad_norm": 0.2888236038985623,
      "learning_rate": 5.57133795195279e-06,
      "loss": 0.0287,
      "step": 5786
    },
    {
      "epoch": 1.5518905872888173,
      "grad_norm": 0.5375321952717957,
      "learning_rate": 5.569787956321496e-06,
      "loss": 0.0386,
      "step": 5787
    },
    {
      "epoch": 1.5521587556985788,
      "grad_norm": 0.20647806258075907,
      "learning_rate": 5.568237905211462e-06,
      "loss": 0.0191,
      "step": 5788
    },
    {
      "epoch": 1.5524269241083402,
      "grad_norm": 0.2964841610412619,
      "learning_rate": 5.5666877987736125e-06,
      "loss": 0.0283,
      "step": 5789
    },
    {
      "epoch": 1.5526950925181013,
      "grad_norm": 0.23126085349292808,
      "learning_rate": 5.565137637158875e-06,
      "loss": 0.0189,
      "step": 5790
    },
    {
      "epoch": 1.5529632609278627,
      "grad_norm": 0.26927983158941593,
      "learning_rate": 5.563587420518187e-06,
      "loss": 0.0179,
      "step": 5791
    },
    {
      "epoch": 1.553231429337624,
      "grad_norm": 0.2558603066314946,
      "learning_rate": 5.562037149002488e-06,
      "loss": 0.0275,
      "step": 5792
    },
    {
      "epoch": 1.5534995977473853,
      "grad_norm": 0.30940498194298244,
      "learning_rate": 5.5604868227627254e-06,
      "loss": 0.0195,
      "step": 5793
    },
    {
      "epoch": 1.5537677661571467,
      "grad_norm": 0.2872279563516604,
      "learning_rate": 5.558936441949847e-06,
      "loss": 0.0181,
      "step": 5794
    },
    {
      "epoch": 1.5540359345669081,
      "grad_norm": 0.2611642610619393,
      "learning_rate": 5.557386006714813e-06,
      "loss": 0.0167,
      "step": 5795
    },
    {
      "epoch": 1.5543041029766693,
      "grad_norm": 0.20751634319980033,
      "learning_rate": 5.555835517208581e-06,
      "loss": 0.0159,
      "step": 5796
    },
    {
      "epoch": 1.5545722713864307,
      "grad_norm": 0.36702568854315043,
      "learning_rate": 5.5542849735821225e-06,
      "loss": 0.0379,
      "step": 5797
    },
    {
      "epoch": 1.554840439796192,
      "grad_norm": 0.24810260757580488,
      "learning_rate": 5.552734375986406e-06,
      "loss": 0.0228,
      "step": 5798
    },
    {
      "epoch": 1.5551086082059533,
      "grad_norm": 0.22295872824604504,
      "learning_rate": 5.551183724572411e-06,
      "loss": 0.024,
      "step": 5799
    },
    {
      "epoch": 1.5553767766157147,
      "grad_norm": 0.3411361468541338,
      "learning_rate": 5.549633019491122e-06,
      "loss": 0.0387,
      "step": 5800
    },
    {
      "epoch": 1.5556449450254761,
      "grad_norm": 0.2553148853813788,
      "learning_rate": 5.548082260893524e-06,
      "loss": 0.0249,
      "step": 5801
    },
    {
      "epoch": 1.5559131134352373,
      "grad_norm": 0.25435255963987563,
      "learning_rate": 5.546531448930611e-06,
      "loss": 0.024,
      "step": 5802
    },
    {
      "epoch": 1.5561812818449987,
      "grad_norm": 0.34857332406692054,
      "learning_rate": 5.544980583753382e-06,
      "loss": 0.0291,
      "step": 5803
    },
    {
      "epoch": 1.55644945025476,
      "grad_norm": 0.29514801438758986,
      "learning_rate": 5.543429665512841e-06,
      "loss": 0.0276,
      "step": 5804
    },
    {
      "epoch": 1.5567176186645213,
      "grad_norm": 0.2171926709747748,
      "learning_rate": 5.541878694359994e-06,
      "loss": 0.0154,
      "step": 5805
    },
    {
      "epoch": 1.5569857870742827,
      "grad_norm": 0.36382482094701857,
      "learning_rate": 5.540327670445861e-06,
      "loss": 0.0272,
      "step": 5806
    },
    {
      "epoch": 1.557253955484044,
      "grad_norm": 0.2789946163644132,
      "learning_rate": 5.538776593921455e-06,
      "loss": 0.025,
      "step": 5807
    },
    {
      "epoch": 1.5575221238938053,
      "grad_norm": 0.2648631296091051,
      "learning_rate": 5.537225464937803e-06,
      "loss": 0.0216,
      "step": 5808
    },
    {
      "epoch": 1.5577902923035667,
      "grad_norm": 0.2010730329586277,
      "learning_rate": 5.535674283645933e-06,
      "loss": 0.0147,
      "step": 5809
    },
    {
      "epoch": 1.5580584607133279,
      "grad_norm": 0.27928125347365207,
      "learning_rate": 5.53412305019688e-06,
      "loss": 0.0248,
      "step": 5810
    },
    {
      "epoch": 1.5583266291230893,
      "grad_norm": 0.26745939779367256,
      "learning_rate": 5.532571764741686e-06,
      "loss": 0.0236,
      "step": 5811
    },
    {
      "epoch": 1.5585947975328507,
      "grad_norm": 0.2544960304997697,
      "learning_rate": 5.531020427431392e-06,
      "loss": 0.0221,
      "step": 5812
    },
    {
      "epoch": 1.558862965942612,
      "grad_norm": 0.2561873102987595,
      "learning_rate": 5.5294690384170476e-06,
      "loss": 0.0232,
      "step": 5813
    },
    {
      "epoch": 1.5591311343523733,
      "grad_norm": 0.3007071286914221,
      "learning_rate": 5.527917597849709e-06,
      "loss": 0.0234,
      "step": 5814
    },
    {
      "epoch": 1.5593993027621347,
      "grad_norm": 0.3262051417004512,
      "learning_rate": 5.526366105880436e-06,
      "loss": 0.0297,
      "step": 5815
    },
    {
      "epoch": 1.5596674711718959,
      "grad_norm": 0.46757410464441185,
      "learning_rate": 5.524814562660292e-06,
      "loss": 0.0359,
      "step": 5816
    },
    {
      "epoch": 1.5599356395816573,
      "grad_norm": 0.2715996016485874,
      "learning_rate": 5.523262968340349e-06,
      "loss": 0.022,
      "step": 5817
    },
    {
      "epoch": 1.5602038079914187,
      "grad_norm": 0.2541366514131034,
      "learning_rate": 5.52171132307168e-06,
      "loss": 0.02,
      "step": 5818
    },
    {
      "epoch": 1.56047197640118,
      "grad_norm": 0.2666978913274109,
      "learning_rate": 5.520159627005363e-06,
      "loss": 0.0161,
      "step": 5819
    },
    {
      "epoch": 1.5607401448109413,
      "grad_norm": 0.31924283760853445,
      "learning_rate": 5.5186078802924845e-06,
      "loss": 0.0265,
      "step": 5820
    },
    {
      "epoch": 1.5610083132207024,
      "grad_norm": 0.2679156846724079,
      "learning_rate": 5.517056083084133e-06,
      "loss": 0.024,
      "step": 5821
    },
    {
      "epoch": 1.5612764816304638,
      "grad_norm": 0.3940345499201356,
      "learning_rate": 5.515504235531406e-06,
      "loss": 0.0291,
      "step": 5822
    },
    {
      "epoch": 1.5615446500402252,
      "grad_norm": 0.31645967781516077,
      "learning_rate": 5.513952337785398e-06,
      "loss": 0.0267,
      "step": 5823
    },
    {
      "epoch": 1.5618128184499867,
      "grad_norm": 0.401432240405321,
      "learning_rate": 5.512400389997216e-06,
      "loss": 0.0369,
      "step": 5824
    },
    {
      "epoch": 1.562080986859748,
      "grad_norm": 0.24132351316335632,
      "learning_rate": 5.5108483923179686e-06,
      "loss": 0.0219,
      "step": 5825
    },
    {
      "epoch": 1.5623491552695092,
      "grad_norm": 0.23293792904271826,
      "learning_rate": 5.509296344898769e-06,
      "loss": 0.0192,
      "step": 5826
    },
    {
      "epoch": 1.5626173236792704,
      "grad_norm": 0.2757454661697072,
      "learning_rate": 5.507744247890737e-06,
      "loss": 0.0202,
      "step": 5827
    },
    {
      "epoch": 1.5628854920890318,
      "grad_norm": 0.25516461802360024,
      "learning_rate": 5.506192101444996e-06,
      "loss": 0.0277,
      "step": 5828
    },
    {
      "epoch": 1.5631536604987932,
      "grad_norm": 0.15995263879387953,
      "learning_rate": 5.504639905712673e-06,
      "loss": 0.016,
      "step": 5829
    },
    {
      "epoch": 1.5634218289085546,
      "grad_norm": 0.24089787510113514,
      "learning_rate": 5.503087660844902e-06,
      "loss": 0.023,
      "step": 5830
    },
    {
      "epoch": 1.563689997318316,
      "grad_norm": 0.2034493903208532,
      "learning_rate": 5.501535366992822e-06,
      "loss": 0.0174,
      "step": 5831
    },
    {
      "epoch": 1.5639581657280772,
      "grad_norm": 0.2655021878270429,
      "learning_rate": 5.499983024307575e-06,
      "loss": 0.0225,
      "step": 5832
    },
    {
      "epoch": 1.5642263341378384,
      "grad_norm": 0.2549957199837677,
      "learning_rate": 5.498430632940308e-06,
      "loss": 0.0272,
      "step": 5833
    },
    {
      "epoch": 1.5644945025475998,
      "grad_norm": 0.3743131073677069,
      "learning_rate": 5.496878193042174e-06,
      "loss": 0.0218,
      "step": 5834
    },
    {
      "epoch": 1.5647626709573612,
      "grad_norm": 0.29935273657104855,
      "learning_rate": 5.4953257047643284e-06,
      "loss": 0.0258,
      "step": 5835
    },
    {
      "epoch": 1.5650308393671226,
      "grad_norm": 0.26077473137465346,
      "learning_rate": 5.493773168257934e-06,
      "loss": 0.0212,
      "step": 5836
    },
    {
      "epoch": 1.565299007776884,
      "grad_norm": 0.320467895905032,
      "learning_rate": 5.492220583674158e-06,
      "loss": 0.0236,
      "step": 5837
    },
    {
      "epoch": 1.5655671761866452,
      "grad_norm": 0.2527012922901406,
      "learning_rate": 5.49066795116417e-06,
      "loss": 0.021,
      "step": 5838
    },
    {
      "epoch": 1.5658353445964064,
      "grad_norm": 0.2645461569879721,
      "learning_rate": 5.4891152708791476e-06,
      "loss": 0.0212,
      "step": 5839
    },
    {
      "epoch": 1.5661035130061678,
      "grad_norm": 0.34149655158928116,
      "learning_rate": 5.487562542970268e-06,
      "loss": 0.0251,
      "step": 5840
    },
    {
      "epoch": 1.5663716814159292,
      "grad_norm": 0.27729191409247045,
      "learning_rate": 5.4860097675887184e-06,
      "loss": 0.0226,
      "step": 5841
    },
    {
      "epoch": 1.5666398498256906,
      "grad_norm": 0.27102400836669815,
      "learning_rate": 5.48445694488569e-06,
      "loss": 0.0202,
      "step": 5842
    },
    {
      "epoch": 1.566908018235452,
      "grad_norm": 0.21191969622596857,
      "learning_rate": 5.482904075012373e-06,
      "loss": 0.0155,
      "step": 5843
    },
    {
      "epoch": 1.5671761866452132,
      "grad_norm": 0.3332278755523665,
      "learning_rate": 5.481351158119969e-06,
      "loss": 0.0393,
      "step": 5844
    },
    {
      "epoch": 1.5674443550549744,
      "grad_norm": 0.2294410548586722,
      "learning_rate": 5.479798194359681e-06,
      "loss": 0.0169,
      "step": 5845
    },
    {
      "epoch": 1.5677125234647358,
      "grad_norm": 0.45667269274626,
      "learning_rate": 5.478245183882716e-06,
      "loss": 0.0281,
      "step": 5846
    },
    {
      "epoch": 1.5679806918744972,
      "grad_norm": 0.2274918709890738,
      "learning_rate": 5.476692126840287e-06,
      "loss": 0.0136,
      "step": 5847
    },
    {
      "epoch": 1.5682488602842586,
      "grad_norm": 0.2894774191830741,
      "learning_rate": 5.475139023383613e-06,
      "loss": 0.0248,
      "step": 5848
    },
    {
      "epoch": 1.56851702869402,
      "grad_norm": 0.4493864804380648,
      "learning_rate": 5.473585873663912e-06,
      "loss": 0.0207,
      "step": 5849
    },
    {
      "epoch": 1.5687851971037812,
      "grad_norm": 0.22222687959419377,
      "learning_rate": 5.472032677832413e-06,
      "loss": 0.018,
      "step": 5850
    },
    {
      "epoch": 1.5690533655135424,
      "grad_norm": 0.2726627808842648,
      "learning_rate": 5.470479436040345e-06,
      "loss": 0.0268,
      "step": 5851
    },
    {
      "epoch": 1.5693215339233038,
      "grad_norm": 0.23237193897466998,
      "learning_rate": 5.468926148438941e-06,
      "loss": 0.0174,
      "step": 5852
    },
    {
      "epoch": 1.5695897023330652,
      "grad_norm": 0.35849497184247436,
      "learning_rate": 5.4673728151794445e-06,
      "loss": 0.0412,
      "step": 5853
    },
    {
      "epoch": 1.5698578707428266,
      "grad_norm": 0.21967626194382844,
      "learning_rate": 5.4658194364131e-06,
      "loss": 0.0165,
      "step": 5854
    },
    {
      "epoch": 1.570126039152588,
      "grad_norm": 0.2336329812437427,
      "learning_rate": 5.464266012291152e-06,
      "loss": 0.0215,
      "step": 5855
    },
    {
      "epoch": 1.5703942075623492,
      "grad_norm": 0.2315289778832945,
      "learning_rate": 5.4627125429648554e-06,
      "loss": 0.0194,
      "step": 5856
    },
    {
      "epoch": 1.5706623759721103,
      "grad_norm": 0.2970776996570779,
      "learning_rate": 5.461159028585465e-06,
      "loss": 0.0269,
      "step": 5857
    },
    {
      "epoch": 1.5709305443818717,
      "grad_norm": 0.3559462998382894,
      "learning_rate": 5.459605469304246e-06,
      "loss": 0.0191,
      "step": 5858
    },
    {
      "epoch": 1.5711987127916331,
      "grad_norm": 0.2229350753484545,
      "learning_rate": 5.458051865272462e-06,
      "loss": 0.0221,
      "step": 5859
    },
    {
      "epoch": 1.5714668812013945,
      "grad_norm": 0.22543193089532465,
      "learning_rate": 5.456498216641386e-06,
      "loss": 0.0167,
      "step": 5860
    },
    {
      "epoch": 1.571735049611156,
      "grad_norm": 0.4031050755784418,
      "learning_rate": 5.45494452356229e-06,
      "loss": 0.0299,
      "step": 5861
    },
    {
      "epoch": 1.5720032180209171,
      "grad_norm": 0.24041060338818002,
      "learning_rate": 5.453390786186451e-06,
      "loss": 0.0173,
      "step": 5862
    },
    {
      "epoch": 1.5722713864306783,
      "grad_norm": 0.2667667689363464,
      "learning_rate": 5.451837004665157e-06,
      "loss": 0.0237,
      "step": 5863
    },
    {
      "epoch": 1.5725395548404397,
      "grad_norm": 0.4133156903539484,
      "learning_rate": 5.450283179149694e-06,
      "loss": 0.031,
      "step": 5864
    },
    {
      "epoch": 1.5728077232502011,
      "grad_norm": 0.4072169726708982,
      "learning_rate": 5.448729309791352e-06,
      "loss": 0.0303,
      "step": 5865
    },
    {
      "epoch": 1.5730758916599625,
      "grad_norm": 0.24392834757441723,
      "learning_rate": 5.44717539674143e-06,
      "loss": 0.0269,
      "step": 5866
    },
    {
      "epoch": 1.573344060069724,
      "grad_norm": 0.24925391065428767,
      "learning_rate": 5.445621440151224e-06,
      "loss": 0.0219,
      "step": 5867
    },
    {
      "epoch": 1.5736122284794851,
      "grad_norm": 0.23223701915434392,
      "learning_rate": 5.444067440172044e-06,
      "loss": 0.0204,
      "step": 5868
    },
    {
      "epoch": 1.5738803968892463,
      "grad_norm": 0.2391676909841526,
      "learning_rate": 5.4425133969551955e-06,
      "loss": 0.0206,
      "step": 5869
    },
    {
      "epoch": 1.5741485652990077,
      "grad_norm": 0.2414102179528656,
      "learning_rate": 5.440959310651992e-06,
      "loss": 0.0224,
      "step": 5870
    },
    {
      "epoch": 1.5744167337087691,
      "grad_norm": 0.3047875178474884,
      "learning_rate": 5.439405181413752e-06,
      "loss": 0.0309,
      "step": 5871
    },
    {
      "epoch": 1.5746849021185305,
      "grad_norm": 0.28948959480008596,
      "learning_rate": 5.437851009391796e-06,
      "loss": 0.0279,
      "step": 5872
    },
    {
      "epoch": 1.574953070528292,
      "grad_norm": 0.27862069444331405,
      "learning_rate": 5.436296794737449e-06,
      "loss": 0.0255,
      "step": 5873
    },
    {
      "epoch": 1.575221238938053,
      "grad_norm": 0.30435352952368366,
      "learning_rate": 5.4347425376020425e-06,
      "loss": 0.0392,
      "step": 5874
    },
    {
      "epoch": 1.5754894073478143,
      "grad_norm": 1.40333211171616,
      "learning_rate": 5.433188238136908e-06,
      "loss": 0.0336,
      "step": 5875
    },
    {
      "epoch": 1.5757575757575757,
      "grad_norm": 0.3040604626568992,
      "learning_rate": 5.431633896493386e-06,
      "loss": 0.0248,
      "step": 5876
    },
    {
      "epoch": 1.576025744167337,
      "grad_norm": 0.20880730600122188,
      "learning_rate": 5.430079512822816e-06,
      "loss": 0.016,
      "step": 5877
    },
    {
      "epoch": 1.5762939125770985,
      "grad_norm": 0.2617828668570294,
      "learning_rate": 5.428525087276547e-06,
      "loss": 0.0234,
      "step": 5878
    },
    {
      "epoch": 1.57656208098686,
      "grad_norm": 1.0351001448133075,
      "learning_rate": 5.426970620005928e-06,
      "loss": 0.0259,
      "step": 5879
    },
    {
      "epoch": 1.576830249396621,
      "grad_norm": 0.35742636956913065,
      "learning_rate": 5.425416111162313e-06,
      "loss": 0.0336,
      "step": 5880
    },
    {
      "epoch": 1.5770984178063823,
      "grad_norm": 0.39148249161812765,
      "learning_rate": 5.423861560897061e-06,
      "loss": 0.0272,
      "step": 5881
    },
    {
      "epoch": 1.5773665862161437,
      "grad_norm": 0.26025513724092664,
      "learning_rate": 5.422306969361534e-06,
      "loss": 0.0241,
      "step": 5882
    },
    {
      "epoch": 1.577634754625905,
      "grad_norm": 0.24563358992123274,
      "learning_rate": 5.420752336707098e-06,
      "loss": 0.0249,
      "step": 5883
    },
    {
      "epoch": 1.5779029230356665,
      "grad_norm": 0.26113091987661613,
      "learning_rate": 5.419197663085124e-06,
      "loss": 0.0189,
      "step": 5884
    },
    {
      "epoch": 1.5781710914454279,
      "grad_norm": 0.19755526317127137,
      "learning_rate": 5.4176429486469874e-06,
      "loss": 0.0205,
      "step": 5885
    },
    {
      "epoch": 1.578439259855189,
      "grad_norm": 0.27211830904507195,
      "learning_rate": 5.416088193544065e-06,
      "loss": 0.0237,
      "step": 5886
    },
    {
      "epoch": 1.5787074282649503,
      "grad_norm": 0.40750040005999355,
      "learning_rate": 5.41453339792774e-06,
      "loss": 0.0313,
      "step": 5887
    },
    {
      "epoch": 1.5789755966747117,
      "grad_norm": 0.30350027107092764,
      "learning_rate": 5.412978561949399e-06,
      "loss": 0.0206,
      "step": 5888
    },
    {
      "epoch": 1.579243765084473,
      "grad_norm": 0.2441571527278083,
      "learning_rate": 5.41142368576043e-06,
      "loss": 0.0239,
      "step": 5889
    },
    {
      "epoch": 1.5795119334942345,
      "grad_norm": 0.29404984651520594,
      "learning_rate": 5.409868769512232e-06,
      "loss": 0.0253,
      "step": 5890
    },
    {
      "epoch": 1.5797801019039959,
      "grad_norm": 0.39778103027842027,
      "learning_rate": 5.408313813356199e-06,
      "loss": 0.0276,
      "step": 5891
    },
    {
      "epoch": 1.580048270313757,
      "grad_norm": 0.271169906933484,
      "learning_rate": 5.406758817443734e-06,
      "loss": 0.0284,
      "step": 5892
    },
    {
      "epoch": 1.5803164387235182,
      "grad_norm": 0.23557018880994973,
      "learning_rate": 5.405203781926243e-06,
      "loss": 0.0243,
      "step": 5893
    },
    {
      "epoch": 1.5805846071332796,
      "grad_norm": 0.20517366119919953,
      "learning_rate": 5.403648706955134e-06,
      "loss": 0.0148,
      "step": 5894
    },
    {
      "epoch": 1.580852775543041,
      "grad_norm": 0.27899723205633437,
      "learning_rate": 5.402093592681823e-06,
      "loss": 0.0252,
      "step": 5895
    },
    {
      "epoch": 1.5811209439528024,
      "grad_norm": 0.3467633689642137,
      "learning_rate": 5.400538439257728e-06,
      "loss": 0.0239,
      "step": 5896
    },
    {
      "epoch": 1.5813891123625639,
      "grad_norm": 0.38836358272094407,
      "learning_rate": 5.398983246834266e-06,
      "loss": 0.0274,
      "step": 5897
    },
    {
      "epoch": 1.581657280772325,
      "grad_norm": 0.36891585054561965,
      "learning_rate": 5.397428015562867e-06,
      "loss": 0.0285,
      "step": 5898
    },
    {
      "epoch": 1.5819254491820862,
      "grad_norm": 0.2175745974120072,
      "learning_rate": 5.395872745594956e-06,
      "loss": 0.0198,
      "step": 5899
    },
    {
      "epoch": 1.5821936175918476,
      "grad_norm": 0.18597927012916518,
      "learning_rate": 5.3943174370819655e-06,
      "loss": 0.0156,
      "step": 5900
    },
    {
      "epoch": 1.582461786001609,
      "grad_norm": 0.20825942034196168,
      "learning_rate": 5.392762090175334e-06,
      "loss": 0.0281,
      "step": 5901
    },
    {
      "epoch": 1.5827299544113704,
      "grad_norm": 0.23259600966112637,
      "learning_rate": 5.391206705026503e-06,
      "loss": 0.0168,
      "step": 5902
    },
    {
      "epoch": 1.5829981228211318,
      "grad_norm": 0.401115360485174,
      "learning_rate": 5.3896512817869106e-06,
      "loss": 0.0292,
      "step": 5903
    },
    {
      "epoch": 1.583266291230893,
      "grad_norm": 0.2973006326095244,
      "learning_rate": 5.388095820608009e-06,
      "loss": 0.0286,
      "step": 5904
    },
    {
      "epoch": 1.5835344596406542,
      "grad_norm": 0.2709043594758099,
      "learning_rate": 5.386540321641246e-06,
      "loss": 0.0248,
      "step": 5905
    },
    {
      "epoch": 1.5838026280504156,
      "grad_norm": 0.2349072638507586,
      "learning_rate": 5.3849847850380795e-06,
      "loss": 0.0229,
      "step": 5906
    },
    {
      "epoch": 1.584070796460177,
      "grad_norm": 0.20478047669248844,
      "learning_rate": 5.383429210949967e-06,
      "loss": 0.0209,
      "step": 5907
    },
    {
      "epoch": 1.5843389648699384,
      "grad_norm": 0.4071892187392045,
      "learning_rate": 5.3818735995283696e-06,
      "loss": 0.0276,
      "step": 5908
    },
    {
      "epoch": 1.5846071332796996,
      "grad_norm": 0.37477060557175523,
      "learning_rate": 5.380317950924754e-06,
      "loss": 0.0287,
      "step": 5909
    },
    {
      "epoch": 1.584875301689461,
      "grad_norm": 0.1873676279645749,
      "learning_rate": 5.378762265290588e-06,
      "loss": 0.0166,
      "step": 5910
    },
    {
      "epoch": 1.5851434700992222,
      "grad_norm": 0.2326070774761628,
      "learning_rate": 5.377206542777347e-06,
      "loss": 0.0201,
      "step": 5911
    },
    {
      "epoch": 1.5854116385089836,
      "grad_norm": 0.41848050492964683,
      "learning_rate": 5.3756507835365055e-06,
      "loss": 0.0221,
      "step": 5912
    },
    {
      "epoch": 1.585679806918745,
      "grad_norm": 0.22848219656747337,
      "learning_rate": 5.374094987719546e-06,
      "loss": 0.0194,
      "step": 5913
    },
    {
      "epoch": 1.5859479753285064,
      "grad_norm": 0.27227213121114097,
      "learning_rate": 5.372539155477949e-06,
      "loss": 0.025,
      "step": 5914
    },
    {
      "epoch": 1.5862161437382676,
      "grad_norm": 0.2627599685784755,
      "learning_rate": 5.370983286963205e-06,
      "loss": 0.0197,
      "step": 5915
    },
    {
      "epoch": 1.586484312148029,
      "grad_norm": 0.3673819508367893,
      "learning_rate": 5.3694273823268025e-06,
      "loss": 0.0365,
      "step": 5916
    },
    {
      "epoch": 1.5867524805577902,
      "grad_norm": 0.24787909188307017,
      "learning_rate": 5.367871441720237e-06,
      "loss": 0.0221,
      "step": 5917
    },
    {
      "epoch": 1.5870206489675516,
      "grad_norm": 0.256494757800115,
      "learning_rate": 5.366315465295007e-06,
      "loss": 0.0273,
      "step": 5918
    },
    {
      "epoch": 1.587288817377313,
      "grad_norm": 0.22657997176062997,
      "learning_rate": 5.36475945320261e-06,
      "loss": 0.0203,
      "step": 5919
    },
    {
      "epoch": 1.5875569857870744,
      "grad_norm": 0.2571208789377835,
      "learning_rate": 5.363203405594556e-06,
      "loss": 0.0229,
      "step": 5920
    },
    {
      "epoch": 1.5878251541968356,
      "grad_norm": 0.24036200495959897,
      "learning_rate": 5.361647322622349e-06,
      "loss": 0.0199,
      "step": 5921
    },
    {
      "epoch": 1.588093322606597,
      "grad_norm": 0.29267885056346876,
      "learning_rate": 5.360091204437503e-06,
      "loss": 0.0302,
      "step": 5922
    },
    {
      "epoch": 1.5883614910163582,
      "grad_norm": 0.27554752034193536,
      "learning_rate": 5.358535051191533e-06,
      "loss": 0.0317,
      "step": 5923
    },
    {
      "epoch": 1.5886296594261196,
      "grad_norm": 0.36860529882388043,
      "learning_rate": 5.3569788630359555e-06,
      "loss": 0.0271,
      "step": 5924
    },
    {
      "epoch": 1.588897827835881,
      "grad_norm": 0.2001404589355475,
      "learning_rate": 5.355422640122295e-06,
      "loss": 0.0189,
      "step": 5925
    },
    {
      "epoch": 1.5891659962456424,
      "grad_norm": 0.30703510503116255,
      "learning_rate": 5.353866382602075e-06,
      "loss": 0.0296,
      "step": 5926
    },
    {
      "epoch": 1.5894341646554035,
      "grad_norm": 0.19684585323053286,
      "learning_rate": 5.352310090626825e-06,
      "loss": 0.0175,
      "step": 5927
    },
    {
      "epoch": 1.589702333065165,
      "grad_norm": 0.24399467623586565,
      "learning_rate": 5.350753764348078e-06,
      "loss": 0.0235,
      "step": 5928
    },
    {
      "epoch": 1.5899705014749261,
      "grad_norm": 0.22616408827129386,
      "learning_rate": 5.349197403917367e-06,
      "loss": 0.0167,
      "step": 5929
    },
    {
      "epoch": 1.5902386698846875,
      "grad_norm": 0.30793300379073,
      "learning_rate": 5.347641009486232e-06,
      "loss": 0.0359,
      "step": 5930
    },
    {
      "epoch": 1.590506838294449,
      "grad_norm": 0.22846964947760046,
      "learning_rate": 5.346084581206215e-06,
      "loss": 0.0203,
      "step": 5931
    },
    {
      "epoch": 1.5907750067042103,
      "grad_norm": 0.3042662760094639,
      "learning_rate": 5.34452811922886e-06,
      "loss": 0.0286,
      "step": 5932
    },
    {
      "epoch": 1.5910431751139715,
      "grad_norm": 0.3454143053370181,
      "learning_rate": 5.3429716237057195e-06,
      "loss": 0.0261,
      "step": 5933
    },
    {
      "epoch": 1.591311343523733,
      "grad_norm": 0.6230807884794777,
      "learning_rate": 5.3414150947883405e-06,
      "loss": 0.0348,
      "step": 5934
    },
    {
      "epoch": 1.5915795119334941,
      "grad_norm": 0.26295699529041155,
      "learning_rate": 5.33985853262828e-06,
      "loss": 0.0229,
      "step": 5935
    },
    {
      "epoch": 1.5918476803432555,
      "grad_norm": 0.28842054638345643,
      "learning_rate": 5.338301937377098e-06,
      "loss": 0.022,
      "step": 5936
    },
    {
      "epoch": 1.592115848753017,
      "grad_norm": 0.27344782957928476,
      "learning_rate": 5.336745309186352e-06,
      "loss": 0.0253,
      "step": 5937
    },
    {
      "epoch": 1.5923840171627783,
      "grad_norm": 0.27272408498609274,
      "learning_rate": 5.335188648207611e-06,
      "loss": 0.0217,
      "step": 5938
    },
    {
      "epoch": 1.5926521855725395,
      "grad_norm": 0.2515661389659537,
      "learning_rate": 5.333631954592443e-06,
      "loss": 0.0242,
      "step": 5939
    },
    {
      "epoch": 1.592920353982301,
      "grad_norm": 0.2772289401029188,
      "learning_rate": 5.332075228492415e-06,
      "loss": 0.0254,
      "step": 5940
    },
    {
      "epoch": 1.593188522392062,
      "grad_norm": 0.2964947650971952,
      "learning_rate": 5.330518470059104e-06,
      "loss": 0.0269,
      "step": 5941
    },
    {
      "epoch": 1.5934566908018235,
      "grad_norm": 0.25682096151794853,
      "learning_rate": 5.328961679444088e-06,
      "loss": 0.0181,
      "step": 5942
    },
    {
      "epoch": 1.593724859211585,
      "grad_norm": 0.2346649369431299,
      "learning_rate": 5.327404856798944e-06,
      "loss": 0.0237,
      "step": 5943
    },
    {
      "epoch": 1.5939930276213463,
      "grad_norm": 0.3113577938375682,
      "learning_rate": 5.3258480022752605e-06,
      "loss": 0.0358,
      "step": 5944
    },
    {
      "epoch": 1.5942611960311075,
      "grad_norm": 0.28413354263208634,
      "learning_rate": 5.324291116024624e-06,
      "loss": 0.0238,
      "step": 5945
    },
    {
      "epoch": 1.594529364440869,
      "grad_norm": 0.35919195492328093,
      "learning_rate": 5.3227341981986195e-06,
      "loss": 0.0274,
      "step": 5946
    },
    {
      "epoch": 1.59479753285063,
      "grad_norm": 0.29814673508237627,
      "learning_rate": 5.321177248948843e-06,
      "loss": 0.0269,
      "step": 5947
    },
    {
      "epoch": 1.5950657012603915,
      "grad_norm": 0.5370517814596732,
      "learning_rate": 5.31962026842689e-06,
      "loss": 0.0325,
      "step": 5948
    },
    {
      "epoch": 1.595333869670153,
      "grad_norm": 0.21727628818489103,
      "learning_rate": 5.318063256784362e-06,
      "loss": 0.0143,
      "step": 5949
    },
    {
      "epoch": 1.5956020380799143,
      "grad_norm": 0.31274065648392296,
      "learning_rate": 5.316506214172859e-06,
      "loss": 0.0256,
      "step": 5950
    },
    {
      "epoch": 1.5958702064896755,
      "grad_norm": 0.299232386328162,
      "learning_rate": 5.314949140743987e-06,
      "loss": 0.0393,
      "step": 5951
    },
    {
      "epoch": 1.5961383748994369,
      "grad_norm": 0.28199596439634944,
      "learning_rate": 5.313392036649352e-06,
      "loss": 0.0232,
      "step": 5952
    },
    {
      "epoch": 1.596406543309198,
      "grad_norm": 0.7740250048378449,
      "learning_rate": 5.311834902040565e-06,
      "loss": 0.0265,
      "step": 5953
    },
    {
      "epoch": 1.5966747117189595,
      "grad_norm": 0.2339104992962593,
      "learning_rate": 5.310277737069243e-06,
      "loss": 0.0199,
      "step": 5954
    },
    {
      "epoch": 1.5969428801287209,
      "grad_norm": 0.18196098600313304,
      "learning_rate": 5.3087205418870014e-06,
      "loss": 0.0183,
      "step": 5955
    },
    {
      "epoch": 1.5972110485384823,
      "grad_norm": 0.32928732196638727,
      "learning_rate": 5.30716331664546e-06,
      "loss": 0.0267,
      "step": 5956
    },
    {
      "epoch": 1.5974792169482435,
      "grad_norm": 0.21388404808914088,
      "learning_rate": 5.305606061496245e-06,
      "loss": 0.0206,
      "step": 5957
    },
    {
      "epoch": 1.5977473853580049,
      "grad_norm": 0.2699802934236477,
      "learning_rate": 5.304048776590974e-06,
      "loss": 0.0262,
      "step": 5958
    },
    {
      "epoch": 1.598015553767766,
      "grad_norm": 0.25632449599985513,
      "learning_rate": 5.302491462081284e-06,
      "loss": 0.0209,
      "step": 5959
    },
    {
      "epoch": 1.5982837221775275,
      "grad_norm": 0.18273788289269718,
      "learning_rate": 5.300934118118802e-06,
      "loss": 0.0168,
      "step": 5960
    },
    {
      "epoch": 1.5985518905872889,
      "grad_norm": 0.35557764025654975,
      "learning_rate": 5.299376744855164e-06,
      "loss": 0.0186,
      "step": 5961
    },
    {
      "epoch": 1.5988200589970503,
      "grad_norm": 0.2835202399023816,
      "learning_rate": 5.2978193424420085e-06,
      "loss": 0.0273,
      "step": 5962
    },
    {
      "epoch": 1.5990882274068114,
      "grad_norm": 0.23649122288407107,
      "learning_rate": 5.296261911030974e-06,
      "loss": 0.0194,
      "step": 5963
    },
    {
      "epoch": 1.5993563958165729,
      "grad_norm": 0.36240407725143836,
      "learning_rate": 5.294704450773703e-06,
      "loss": 0.0253,
      "step": 5964
    },
    {
      "epoch": 1.599624564226334,
      "grad_norm": 0.188571532012262,
      "learning_rate": 5.293146961821844e-06,
      "loss": 0.0171,
      "step": 5965
    },
    {
      "epoch": 1.5998927326360954,
      "grad_norm": 0.24442335426769185,
      "learning_rate": 5.291589444327041e-06,
      "loss": 0.0222,
      "step": 5966
    },
    {
      "epoch": 1.6001609010458568,
      "grad_norm": 0.41593302365873674,
      "learning_rate": 5.29003189844095e-06,
      "loss": 0.035,
      "step": 5967
    },
    {
      "epoch": 1.6004290694556182,
      "grad_norm": 0.32286049397605465,
      "learning_rate": 5.2884743243152215e-06,
      "loss": 0.0186,
      "step": 5968
    },
    {
      "epoch": 1.6006972378653794,
      "grad_norm": 0.32079716822399607,
      "learning_rate": 5.286916722101515e-06,
      "loss": 0.029,
      "step": 5969
    },
    {
      "epoch": 1.6009654062751408,
      "grad_norm": 0.3842356553383585,
      "learning_rate": 5.285359091951488e-06,
      "loss": 0.023,
      "step": 5970
    },
    {
      "epoch": 1.601233574684902,
      "grad_norm": 0.32706843211427317,
      "learning_rate": 5.283801434016804e-06,
      "loss": 0.0188,
      "step": 5971
    },
    {
      "epoch": 1.6015017430946634,
      "grad_norm": 0.22574854414977236,
      "learning_rate": 5.282243748449128e-06,
      "loss": 0.0235,
      "step": 5972
    },
    {
      "epoch": 1.6017699115044248,
      "grad_norm": 0.28283305260577324,
      "learning_rate": 5.280686035400129e-06,
      "loss": 0.0288,
      "step": 5973
    },
    {
      "epoch": 1.6020380799141862,
      "grad_norm": 0.22885691873132874,
      "learning_rate": 5.279128295021475e-06,
      "loss": 0.0174,
      "step": 5974
    },
    {
      "epoch": 1.6023062483239474,
      "grad_norm": 0.35956557902849146,
      "learning_rate": 5.277570527464839e-06,
      "loss": 0.0201,
      "step": 5975
    },
    {
      "epoch": 1.6025744167337088,
      "grad_norm": 0.2377556920890842,
      "learning_rate": 5.2760127328818985e-06,
      "loss": 0.0223,
      "step": 5976
    },
    {
      "epoch": 1.60284258514347,
      "grad_norm": 0.3101119214428328,
      "learning_rate": 5.274454911424332e-06,
      "loss": 0.0175,
      "step": 5977
    },
    {
      "epoch": 1.6031107535532314,
      "grad_norm": 0.3094101265218406,
      "learning_rate": 5.272897063243818e-06,
      "loss": 0.0265,
      "step": 5978
    },
    {
      "epoch": 1.6033789219629928,
      "grad_norm": 0.27841471904500864,
      "learning_rate": 5.2713391884920415e-06,
      "loss": 0.026,
      "step": 5979
    },
    {
      "epoch": 1.6036470903727542,
      "grad_norm": 0.21221480698318357,
      "learning_rate": 5.269781287320688e-06,
      "loss": 0.0204,
      "step": 5980
    },
    {
      "epoch": 1.6039152587825154,
      "grad_norm": 0.25175395440423437,
      "learning_rate": 5.268223359881449e-06,
      "loss": 0.0196,
      "step": 5981
    },
    {
      "epoch": 1.6041834271922768,
      "grad_norm": 0.18821356695325567,
      "learning_rate": 5.266665406326013e-06,
      "loss": 0.0157,
      "step": 5982
    },
    {
      "epoch": 1.604451595602038,
      "grad_norm": 0.39938492382497204,
      "learning_rate": 5.265107426806073e-06,
      "loss": 0.0345,
      "step": 5983
    },
    {
      "epoch": 1.6047197640117994,
      "grad_norm": 0.2870471474656807,
      "learning_rate": 5.263549421473327e-06,
      "loss": 0.024,
      "step": 5984
    },
    {
      "epoch": 1.6049879324215608,
      "grad_norm": 0.2572883142093176,
      "learning_rate": 5.261991390479473e-06,
      "loss": 0.0314,
      "step": 5985
    },
    {
      "epoch": 1.6052561008313222,
      "grad_norm": 0.2942134032355729,
      "learning_rate": 5.260433333976214e-06,
      "loss": 0.0214,
      "step": 5986
    },
    {
      "epoch": 1.6055242692410834,
      "grad_norm": 0.326256260329595,
      "learning_rate": 5.258875252115253e-06,
      "loss": 0.027,
      "step": 5987
    },
    {
      "epoch": 1.6057924376508448,
      "grad_norm": 0.26644813713302845,
      "learning_rate": 5.2573171450482964e-06,
      "loss": 0.0254,
      "step": 5988
    },
    {
      "epoch": 1.606060606060606,
      "grad_norm": 0.26920740438061475,
      "learning_rate": 5.255759012927051e-06,
      "loss": 0.0242,
      "step": 5989
    },
    {
      "epoch": 1.6063287744703674,
      "grad_norm": 0.2917397965364728,
      "learning_rate": 5.254200855903231e-06,
      "loss": 0.025,
      "step": 5990
    },
    {
      "epoch": 1.6065969428801288,
      "grad_norm": 0.22476125679757147,
      "learning_rate": 5.2526426741285465e-06,
      "loss": 0.0224,
      "step": 5991
    },
    {
      "epoch": 1.6068651112898902,
      "grad_norm": 0.24064636953706056,
      "learning_rate": 5.251084467754717e-06,
      "loss": 0.0265,
      "step": 5992
    },
    {
      "epoch": 1.6071332796996514,
      "grad_norm": 0.18621273956615175,
      "learning_rate": 5.24952623693346e-06,
      "loss": 0.0147,
      "step": 5993
    },
    {
      "epoch": 1.6074014481094125,
      "grad_norm": 0.2606553170068972,
      "learning_rate": 5.247967981816495e-06,
      "loss": 0.0234,
      "step": 5994
    },
    {
      "epoch": 1.607669616519174,
      "grad_norm": 0.27529611039384405,
      "learning_rate": 5.246409702555546e-06,
      "loss": 0.0321,
      "step": 5995
    },
    {
      "epoch": 1.6079377849289354,
      "grad_norm": 0.24850101839979494,
      "learning_rate": 5.244851399302337e-06,
      "loss": 0.0214,
      "step": 5996
    },
    {
      "epoch": 1.6082059533386968,
      "grad_norm": 0.280861568674063,
      "learning_rate": 5.243293072208599e-06,
      "loss": 0.0197,
      "step": 5997
    },
    {
      "epoch": 1.6084741217484582,
      "grad_norm": 0.26195068069906485,
      "learning_rate": 5.24173472142606e-06,
      "loss": 0.0263,
      "step": 5998
    },
    {
      "epoch": 1.6087422901582193,
      "grad_norm": 0.21410842407406058,
      "learning_rate": 5.240176347106455e-06,
      "loss": 0.0185,
      "step": 5999
    },
    {
      "epoch": 1.6090104585679805,
      "grad_norm": 0.2162034235587646,
      "learning_rate": 5.2386179494015155e-06,
      "loss": 0.0161,
      "step": 6000
    },
    {
      "epoch": 1.609278626977742,
      "grad_norm": 0.2556408572684341,
      "learning_rate": 5.237059528462978e-06,
      "loss": 0.0213,
      "step": 6001
    },
    {
      "epoch": 1.6095467953875033,
      "grad_norm": 0.2433063478818582,
      "learning_rate": 5.235501084442584e-06,
      "loss": 0.0206,
      "step": 6002
    },
    {
      "epoch": 1.6098149637972647,
      "grad_norm": 0.18704173480783193,
      "learning_rate": 5.233942617492077e-06,
      "loss": 0.0166,
      "step": 6003
    },
    {
      "epoch": 1.6100831322070261,
      "grad_norm": 0.3414662461042351,
      "learning_rate": 5.232384127763197e-06,
      "loss": 0.0286,
      "step": 6004
    },
    {
      "epoch": 1.6103513006167873,
      "grad_norm": 0.30623816719307223,
      "learning_rate": 5.230825615407692e-06,
      "loss": 0.0325,
      "step": 6005
    },
    {
      "epoch": 1.6106194690265485,
      "grad_norm": 0.30049527260513015,
      "learning_rate": 5.229267080577308e-06,
      "loss": 0.0245,
      "step": 6006
    },
    {
      "epoch": 1.61088763743631,
      "grad_norm": 0.3291948578205075,
      "learning_rate": 5.227708523423799e-06,
      "loss": 0.022,
      "step": 6007
    },
    {
      "epoch": 1.6111558058460713,
      "grad_norm": 0.24892906175819682,
      "learning_rate": 5.2261499440989145e-06,
      "loss": 0.018,
      "step": 6008
    },
    {
      "epoch": 1.6114239742558327,
      "grad_norm": 0.30053619352184857,
      "learning_rate": 5.22459134275441e-06,
      "loss": 0.0312,
      "step": 6009
    },
    {
      "epoch": 1.6116921426655941,
      "grad_norm": 0.2800944378429468,
      "learning_rate": 5.223032719542044e-06,
      "loss": 0.0251,
      "step": 6010
    },
    {
      "epoch": 1.6119603110753553,
      "grad_norm": 0.37663210803717057,
      "learning_rate": 5.221474074613574e-06,
      "loss": 0.0455,
      "step": 6011
    },
    {
      "epoch": 1.6122284794851165,
      "grad_norm": 0.29887326190554997,
      "learning_rate": 5.21991540812076e-06,
      "loss": 0.0202,
      "step": 6012
    },
    {
      "epoch": 1.612496647894878,
      "grad_norm": 0.18280999941730336,
      "learning_rate": 5.218356720215369e-06,
      "loss": 0.0187,
      "step": 6013
    },
    {
      "epoch": 1.6127648163046393,
      "grad_norm": 0.5243515358027353,
      "learning_rate": 5.216798011049162e-06,
      "loss": 0.037,
      "step": 6014
    },
    {
      "epoch": 1.6130329847144007,
      "grad_norm": 0.35393016554438766,
      "learning_rate": 5.215239280773908e-06,
      "loss": 0.0252,
      "step": 6015
    },
    {
      "epoch": 1.6133011531241621,
      "grad_norm": 0.2677728232561824,
      "learning_rate": 5.213680529541378e-06,
      "loss": 0.0149,
      "step": 6016
    },
    {
      "epoch": 1.6135693215339233,
      "grad_norm": 0.27862803882646314,
      "learning_rate": 5.212121757503341e-06,
      "loss": 0.0261,
      "step": 6017
    },
    {
      "epoch": 1.6138374899436845,
      "grad_norm": 0.27686407190652823,
      "learning_rate": 5.210562964811573e-06,
      "loss": 0.0211,
      "step": 6018
    },
    {
      "epoch": 1.6141056583534459,
      "grad_norm": 0.2561234659475586,
      "learning_rate": 5.209004151617847e-06,
      "loss": 0.0254,
      "step": 6019
    },
    {
      "epoch": 1.6143738267632073,
      "grad_norm": 0.4276827291126018,
      "learning_rate": 5.2074453180739415e-06,
      "loss": 0.0302,
      "step": 6020
    },
    {
      "epoch": 1.6146419951729687,
      "grad_norm": 0.3012334452437274,
      "learning_rate": 5.2058864643316376e-06,
      "loss": 0.0283,
      "step": 6021
    },
    {
      "epoch": 1.61491016358273,
      "grad_norm": 0.24279073294681833,
      "learning_rate": 5.204327590542714e-06,
      "loss": 0.0283,
      "step": 6022
    },
    {
      "epoch": 1.6151783319924913,
      "grad_norm": 0.7858266002805424,
      "learning_rate": 5.202768696858955e-06,
      "loss": 0.0272,
      "step": 6023
    },
    {
      "epoch": 1.6154465004022525,
      "grad_norm": 0.23956895246764823,
      "learning_rate": 5.201209783432148e-06,
      "loss": 0.0258,
      "step": 6024
    },
    {
      "epoch": 1.6157146688120139,
      "grad_norm": 0.27379602379957696,
      "learning_rate": 5.199650850414078e-06,
      "loss": 0.0217,
      "step": 6025
    },
    {
      "epoch": 1.6159828372217753,
      "grad_norm": 0.29975127098184795,
      "learning_rate": 5.198091897956533e-06,
      "loss": 0.0307,
      "step": 6026
    },
    {
      "epoch": 1.6162510056315367,
      "grad_norm": 0.29321911292415415,
      "learning_rate": 5.196532926211307e-06,
      "loss": 0.024,
      "step": 6027
    },
    {
      "epoch": 1.616519174041298,
      "grad_norm": 0.34472867113858013,
      "learning_rate": 5.194973935330192e-06,
      "loss": 0.0197,
      "step": 6028
    },
    {
      "epoch": 1.6167873424510593,
      "grad_norm": 0.29235380736397093,
      "learning_rate": 5.193414925464984e-06,
      "loss": 0.0217,
      "step": 6029
    },
    {
      "epoch": 1.6170555108608204,
      "grad_norm": 0.23448410003232165,
      "learning_rate": 5.191855896767476e-06,
      "loss": 0.0239,
      "step": 6030
    },
    {
      "epoch": 1.6173236792705818,
      "grad_norm": 0.27416450321143826,
      "learning_rate": 5.190296849389469e-06,
      "loss": 0.0215,
      "step": 6031
    },
    {
      "epoch": 1.6175918476803433,
      "grad_norm": 0.2892588302737628,
      "learning_rate": 5.188737783482763e-06,
      "loss": 0.0267,
      "step": 6032
    },
    {
      "epoch": 1.6178600160901047,
      "grad_norm": 0.2994178332088737,
      "learning_rate": 5.18717869919916e-06,
      "loss": 0.0297,
      "step": 6033
    },
    {
      "epoch": 1.618128184499866,
      "grad_norm": 0.3097190474461079,
      "learning_rate": 5.185619596690465e-06,
      "loss": 0.0278,
      "step": 6034
    },
    {
      "epoch": 1.6183963529096272,
      "grad_norm": 0.4767684147008618,
      "learning_rate": 5.184060476108482e-06,
      "loss": 0.0489,
      "step": 6035
    },
    {
      "epoch": 1.6186645213193884,
      "grad_norm": 0.23188519277538727,
      "learning_rate": 5.18250133760502e-06,
      "loss": 0.0232,
      "step": 6036
    },
    {
      "epoch": 1.6189326897291498,
      "grad_norm": 0.30037075235176075,
      "learning_rate": 5.180942181331887e-06,
      "loss": 0.0253,
      "step": 6037
    },
    {
      "epoch": 1.6192008581389112,
      "grad_norm": 0.266405387697413,
      "learning_rate": 5.179383007440895e-06,
      "loss": 0.0183,
      "step": 6038
    },
    {
      "epoch": 1.6194690265486726,
      "grad_norm": 0.2564585842780013,
      "learning_rate": 5.177823816083853e-06,
      "loss": 0.0306,
      "step": 6039
    },
    {
      "epoch": 1.619737194958434,
      "grad_norm": 0.3627470151918637,
      "learning_rate": 5.17626460741258e-06,
      "loss": 0.0257,
      "step": 6040
    },
    {
      "epoch": 1.6200053633681952,
      "grad_norm": 0.29836857151728524,
      "learning_rate": 5.174705381578889e-06,
      "loss": 0.0248,
      "step": 6041
    },
    {
      "epoch": 1.6202735317779564,
      "grad_norm": 0.2341975065759829,
      "learning_rate": 5.173146138734602e-06,
      "loss": 0.0225,
      "step": 6042
    },
    {
      "epoch": 1.6205417001877178,
      "grad_norm": 0.33041769120531833,
      "learning_rate": 5.171586879031533e-06,
      "loss": 0.0267,
      "step": 6043
    },
    {
      "epoch": 1.6208098685974792,
      "grad_norm": 0.2330067745350485,
      "learning_rate": 5.170027602621503e-06,
      "loss": 0.0193,
      "step": 6044
    },
    {
      "epoch": 1.6210780370072406,
      "grad_norm": 0.26148962792699865,
      "learning_rate": 5.1684683096563394e-06,
      "loss": 0.0289,
      "step": 6045
    },
    {
      "epoch": 1.621346205417002,
      "grad_norm": 0.20458666207828188,
      "learning_rate": 5.166909000287863e-06,
      "loss": 0.0204,
      "step": 6046
    },
    {
      "epoch": 1.6216143738267632,
      "grad_norm": 0.2363615157684955,
      "learning_rate": 5.1653496746679015e-06,
      "loss": 0.0249,
      "step": 6047
    },
    {
      "epoch": 1.6218825422365244,
      "grad_norm": 0.2530554438559414,
      "learning_rate": 5.163790332948281e-06,
      "loss": 0.0191,
      "step": 6048
    },
    {
      "epoch": 1.6221507106462858,
      "grad_norm": 0.2456789262000578,
      "learning_rate": 5.162230975280828e-06,
      "loss": 0.0216,
      "step": 6049
    },
    {
      "epoch": 1.6224188790560472,
      "grad_norm": 0.23074711947451249,
      "learning_rate": 5.160671601817378e-06,
      "loss": 0.0187,
      "step": 6050
    },
    {
      "epoch": 1.6226870474658086,
      "grad_norm": 0.23768296132007963,
      "learning_rate": 5.15911221270976e-06,
      "loss": 0.0207,
      "step": 6051
    },
    {
      "epoch": 1.62295521587557,
      "grad_norm": 0.25552137649982226,
      "learning_rate": 5.157552808109808e-06,
      "loss": 0.0245,
      "step": 6052
    },
    {
      "epoch": 1.6232233842853312,
      "grad_norm": 0.17883270148671665,
      "learning_rate": 5.155993388169358e-06,
      "loss": 0.015,
      "step": 6053
    },
    {
      "epoch": 1.6234915526950924,
      "grad_norm": 0.23532704641393742,
      "learning_rate": 5.154433953040247e-06,
      "loss": 0.0204,
      "step": 6054
    },
    {
      "epoch": 1.6237597211048538,
      "grad_norm": 0.25588525341142726,
      "learning_rate": 5.15287450287431e-06,
      "loss": 0.027,
      "step": 6055
    },
    {
      "epoch": 1.6240278895146152,
      "grad_norm": 0.4809724210214293,
      "learning_rate": 5.151315037823389e-06,
      "loss": 0.0234,
      "step": 6056
    },
    {
      "epoch": 1.6242960579243766,
      "grad_norm": 0.20666663597947887,
      "learning_rate": 5.149755558039325e-06,
      "loss": 0.0181,
      "step": 6057
    },
    {
      "epoch": 1.624564226334138,
      "grad_norm": 0.28275606776408146,
      "learning_rate": 5.14819606367396e-06,
      "loss": 0.0233,
      "step": 6058
    },
    {
      "epoch": 1.6248323947438992,
      "grad_norm": 0.3301984013941858,
      "learning_rate": 5.1466365548791374e-06,
      "loss": 0.0221,
      "step": 6059
    },
    {
      "epoch": 1.6251005631536604,
      "grad_norm": 0.2621536530004677,
      "learning_rate": 5.1450770318067035e-06,
      "loss": 0.0245,
      "step": 6060
    },
    {
      "epoch": 1.6253687315634218,
      "grad_norm": 0.24970437149897817,
      "learning_rate": 5.1435174946085035e-06,
      "loss": 0.0262,
      "step": 6061
    },
    {
      "epoch": 1.6256368999731832,
      "grad_norm": 0.2373632645904136,
      "learning_rate": 5.141957943436385e-06,
      "loss": 0.0181,
      "step": 6062
    },
    {
      "epoch": 1.6259050683829446,
      "grad_norm": 0.5757162828613352,
      "learning_rate": 5.140398378442201e-06,
      "loss": 0.0278,
      "step": 6063
    },
    {
      "epoch": 1.626173236792706,
      "grad_norm": 0.2624991113408737,
      "learning_rate": 5.138838799777798e-06,
      "loss": 0.026,
      "step": 6064
    },
    {
      "epoch": 1.6264414052024672,
      "grad_norm": 0.23832560544425088,
      "learning_rate": 5.137279207595029e-06,
      "loss": 0.0227,
      "step": 6065
    },
    {
      "epoch": 1.6267095736122283,
      "grad_norm": 0.24601496639083625,
      "learning_rate": 5.135719602045749e-06,
      "loss": 0.0255,
      "step": 6066
    },
    {
      "epoch": 1.6269777420219897,
      "grad_norm": 0.33635008573275843,
      "learning_rate": 5.134159983281812e-06,
      "loss": 0.0336,
      "step": 6067
    },
    {
      "epoch": 1.6272459104317512,
      "grad_norm": 0.3007709447918642,
      "learning_rate": 5.1326003514550734e-06,
      "loss": 0.0224,
      "step": 6068
    },
    {
      "epoch": 1.6275140788415126,
      "grad_norm": 0.3210673007629223,
      "learning_rate": 5.131040706717391e-06,
      "loss": 0.0287,
      "step": 6069
    },
    {
      "epoch": 1.627782247251274,
      "grad_norm": 0.29417267098866273,
      "learning_rate": 5.1294810492206215e-06,
      "loss": 0.021,
      "step": 6070
    },
    {
      "epoch": 1.6280504156610351,
      "grad_norm": 0.2301582234612842,
      "learning_rate": 5.127921379116626e-06,
      "loss": 0.018,
      "step": 6071
    },
    {
      "epoch": 1.6283185840707963,
      "grad_norm": 0.22663413201006224,
      "learning_rate": 5.126361696557267e-06,
      "loss": 0.0218,
      "step": 6072
    },
    {
      "epoch": 1.6285867524805577,
      "grad_norm": 0.18462084472413337,
      "learning_rate": 5.124802001694405e-06,
      "loss": 0.019,
      "step": 6073
    },
    {
      "epoch": 1.6288549208903191,
      "grad_norm": 0.33562067442140076,
      "learning_rate": 5.123242294679902e-06,
      "loss": 0.0193,
      "step": 6074
    },
    {
      "epoch": 1.6291230893000805,
      "grad_norm": 0.31308721531222267,
      "learning_rate": 5.121682575665625e-06,
      "loss": 0.0204,
      "step": 6075
    },
    {
      "epoch": 1.629391257709842,
      "grad_norm": 0.29623547599765365,
      "learning_rate": 5.120122844803438e-06,
      "loss": 0.0269,
      "step": 6076
    },
    {
      "epoch": 1.6296594261196031,
      "grad_norm": 0.2233984822872891,
      "learning_rate": 5.118563102245209e-06,
      "loss": 0.0191,
      "step": 6077
    },
    {
      "epoch": 1.6299275945293643,
      "grad_norm": 0.31272444422733586,
      "learning_rate": 5.117003348142806e-06,
      "loss": 0.0319,
      "step": 6078
    },
    {
      "epoch": 1.6301957629391257,
      "grad_norm": 0.17813320955010767,
      "learning_rate": 5.115443582648097e-06,
      "loss": 0.0152,
      "step": 6079
    },
    {
      "epoch": 1.6304639313488871,
      "grad_norm": 0.19825503189207636,
      "learning_rate": 5.113883805912954e-06,
      "loss": 0.0154,
      "step": 6080
    },
    {
      "epoch": 1.6307320997586485,
      "grad_norm": 0.26589363251148396,
      "learning_rate": 5.112324018089244e-06,
      "loss": 0.0206,
      "step": 6081
    },
    {
      "epoch": 1.6310002681684097,
      "grad_norm": 0.40941421157692737,
      "learning_rate": 5.1107642193288455e-06,
      "loss": 0.0221,
      "step": 6082
    },
    {
      "epoch": 1.631268436578171,
      "grad_norm": 0.3381186719260909,
      "learning_rate": 5.109204409783628e-06,
      "loss": 0.0248,
      "step": 6083
    },
    {
      "epoch": 1.6315366049879323,
      "grad_norm": 0.39045215737394795,
      "learning_rate": 5.107644589605468e-06,
      "loss": 0.028,
      "step": 6084
    },
    {
      "epoch": 1.6318047733976937,
      "grad_norm": 0.22200185977850678,
      "learning_rate": 5.106084758946239e-06,
      "loss": 0.0193,
      "step": 6085
    },
    {
      "epoch": 1.632072941807455,
      "grad_norm": 0.29156947592748633,
      "learning_rate": 5.1045249179578194e-06,
      "loss": 0.0192,
      "step": 6086
    },
    {
      "epoch": 1.6323411102172165,
      "grad_norm": 0.37698898476503057,
      "learning_rate": 5.102965066792085e-06,
      "loss": 0.0249,
      "step": 6087
    },
    {
      "epoch": 1.6326092786269777,
      "grad_norm": 0.1756982931067425,
      "learning_rate": 5.101405205600916e-06,
      "loss": 0.0175,
      "step": 6088
    },
    {
      "epoch": 1.632877447036739,
      "grad_norm": 0.394779058511191,
      "learning_rate": 5.099845334536192e-06,
      "loss": 0.0216,
      "step": 6089
    },
    {
      "epoch": 1.6331456154465003,
      "grad_norm": 0.22842017782334087,
      "learning_rate": 5.098285453749793e-06,
      "loss": 0.0173,
      "step": 6090
    },
    {
      "epoch": 1.6334137838562617,
      "grad_norm": 0.2359118837383206,
      "learning_rate": 5.0967255633936e-06,
      "loss": 0.0159,
      "step": 6091
    },
    {
      "epoch": 1.633681952266023,
      "grad_norm": 0.2970209266940588,
      "learning_rate": 5.095165663619493e-06,
      "loss": 0.0249,
      "step": 6092
    },
    {
      "epoch": 1.6339501206757845,
      "grad_norm": 0.22417536970352475,
      "learning_rate": 5.093605754579361e-06,
      "loss": 0.0238,
      "step": 6093
    },
    {
      "epoch": 1.6342182890855457,
      "grad_norm": 0.20927671610944049,
      "learning_rate": 5.092045836425085e-06,
      "loss": 0.0165,
      "step": 6094
    },
    {
      "epoch": 1.634486457495307,
      "grad_norm": 0.27103763245266027,
      "learning_rate": 5.09048590930855e-06,
      "loss": 0.0265,
      "step": 6095
    },
    {
      "epoch": 1.6347546259050683,
      "grad_norm": 0.24024158869159878,
      "learning_rate": 5.088925973381643e-06,
      "loss": 0.0202,
      "step": 6096
    },
    {
      "epoch": 1.6350227943148297,
      "grad_norm": 0.21876169914772395,
      "learning_rate": 5.087366028796248e-06,
      "loss": 0.0153,
      "step": 6097
    },
    {
      "epoch": 1.635290962724591,
      "grad_norm": 0.2845222612648831,
      "learning_rate": 5.085806075704256e-06,
      "loss": 0.0237,
      "step": 6098
    },
    {
      "epoch": 1.6355591311343525,
      "grad_norm": 0.19497967804508182,
      "learning_rate": 5.084246114257554e-06,
      "loss": 0.0173,
      "step": 6099
    },
    {
      "epoch": 1.6358272995441137,
      "grad_norm": 0.34828636169363086,
      "learning_rate": 5.082686144608031e-06,
      "loss": 0.0287,
      "step": 6100
    },
    {
      "epoch": 1.636095467953875,
      "grad_norm": 0.34117679103592735,
      "learning_rate": 5.081126166907579e-06,
      "loss": 0.0235,
      "step": 6101
    },
    {
      "epoch": 1.6363636363636362,
      "grad_norm": 0.3326581253294498,
      "learning_rate": 5.079566181308087e-06,
      "loss": 0.02,
      "step": 6102
    },
    {
      "epoch": 1.6366318047733976,
      "grad_norm": 0.42181084124198487,
      "learning_rate": 5.078006187961446e-06,
      "loss": 0.0312,
      "step": 6103
    },
    {
      "epoch": 1.636899973183159,
      "grad_norm": 0.316096892378578,
      "learning_rate": 5.076446187019551e-06,
      "loss": 0.0211,
      "step": 6104
    },
    {
      "epoch": 1.6371681415929205,
      "grad_norm": 0.2625254049210759,
      "learning_rate": 5.074886178634293e-06,
      "loss": 0.0241,
      "step": 6105
    },
    {
      "epoch": 1.6374363100026816,
      "grad_norm": 0.244530541594946,
      "learning_rate": 5.073326162957567e-06,
      "loss": 0.0275,
      "step": 6106
    },
    {
      "epoch": 1.637704478412443,
      "grad_norm": 0.23704026229732905,
      "learning_rate": 5.071766140141266e-06,
      "loss": 0.0242,
      "step": 6107
    },
    {
      "epoch": 1.6379726468222042,
      "grad_norm": 0.22107967236641637,
      "learning_rate": 5.070206110337289e-06,
      "loss": 0.0224,
      "step": 6108
    },
    {
      "epoch": 1.6382408152319656,
      "grad_norm": 0.25770741226244026,
      "learning_rate": 5.068646073697527e-06,
      "loss": 0.0218,
      "step": 6109
    },
    {
      "epoch": 1.638508983641727,
      "grad_norm": 0.24259167656555394,
      "learning_rate": 5.06708603037388e-06,
      "loss": 0.0185,
      "step": 6110
    },
    {
      "epoch": 1.6387771520514884,
      "grad_norm": 0.32348092987774324,
      "learning_rate": 5.065525980518244e-06,
      "loss": 0.0241,
      "step": 6111
    },
    {
      "epoch": 1.6390453204612496,
      "grad_norm": 0.3842292736350233,
      "learning_rate": 5.063965924282518e-06,
      "loss": 0.0211,
      "step": 6112
    },
    {
      "epoch": 1.639313488871011,
      "grad_norm": 0.2519085217158739,
      "learning_rate": 5.0624058618186e-06,
      "loss": 0.0227,
      "step": 6113
    },
    {
      "epoch": 1.6395816572807722,
      "grad_norm": 0.26504645280621525,
      "learning_rate": 5.060845793278389e-06,
      "loss": 0.0186,
      "step": 6114
    },
    {
      "epoch": 1.6398498256905336,
      "grad_norm": 0.2551925233916949,
      "learning_rate": 5.059285718813784e-06,
      "loss": 0.0275,
      "step": 6115
    },
    {
      "epoch": 1.640117994100295,
      "grad_norm": 0.3274824054181548,
      "learning_rate": 5.0577256385766845e-06,
      "loss": 0.019,
      "step": 6116
    },
    {
      "epoch": 1.6403861625100564,
      "grad_norm": 0.27776774778885893,
      "learning_rate": 5.056165552718995e-06,
      "loss": 0.0181,
      "step": 6117
    },
    {
      "epoch": 1.6406543309198176,
      "grad_norm": 0.24007962023790153,
      "learning_rate": 5.054605461392613e-06,
      "loss": 0.0193,
      "step": 6118
    },
    {
      "epoch": 1.640922499329579,
      "grad_norm": 0.2668968467847279,
      "learning_rate": 5.0530453647494424e-06,
      "loss": 0.0227,
      "step": 6119
    },
    {
      "epoch": 1.6411906677393402,
      "grad_norm": 0.35086458967795603,
      "learning_rate": 5.051485262941388e-06,
      "loss": 0.022,
      "step": 6120
    },
    {
      "epoch": 1.6414588361491016,
      "grad_norm": 0.27549829296499784,
      "learning_rate": 5.049925156120348e-06,
      "loss": 0.0239,
      "step": 6121
    },
    {
      "epoch": 1.641727004558863,
      "grad_norm": 0.2756624507691886,
      "learning_rate": 5.048365044438228e-06,
      "loss": 0.0238,
      "step": 6122
    },
    {
      "epoch": 1.6419951729686244,
      "grad_norm": 0.2528537300404635,
      "learning_rate": 5.046804928046933e-06,
      "loss": 0.0208,
      "step": 6123
    },
    {
      "epoch": 1.6422633413783856,
      "grad_norm": 0.2678735306392268,
      "learning_rate": 5.045244807098365e-06,
      "loss": 0.0228,
      "step": 6124
    },
    {
      "epoch": 1.642531509788147,
      "grad_norm": 0.25113900214577967,
      "learning_rate": 5.0436846817444305e-06,
      "loss": 0.0288,
      "step": 6125
    },
    {
      "epoch": 1.6427996781979082,
      "grad_norm": 0.23566121392342365,
      "learning_rate": 5.042124552137037e-06,
      "loss": 0.0186,
      "step": 6126
    },
    {
      "epoch": 1.6430678466076696,
      "grad_norm": 0.1996452594743113,
      "learning_rate": 5.040564418428087e-06,
      "loss": 0.0195,
      "step": 6127
    },
    {
      "epoch": 1.643336015017431,
      "grad_norm": 0.26773308708002486,
      "learning_rate": 5.039004280769485e-06,
      "loss": 0.0228,
      "step": 6128
    },
    {
      "epoch": 1.6436041834271924,
      "grad_norm": 0.2967392689698117,
      "learning_rate": 5.03744413931314e-06,
      "loss": 0.0338,
      "step": 6129
    },
    {
      "epoch": 1.6438723518369536,
      "grad_norm": 0.27226092342885233,
      "learning_rate": 5.035883994210959e-06,
      "loss": 0.0238,
      "step": 6130
    },
    {
      "epoch": 1.644140520246715,
      "grad_norm": 0.21322430627109468,
      "learning_rate": 5.0343238456148504e-06,
      "loss": 0.0215,
      "step": 6131
    },
    {
      "epoch": 1.6444086886564762,
      "grad_norm": 0.22294413621127496,
      "learning_rate": 5.03276369367672e-06,
      "loss": 0.0266,
      "step": 6132
    },
    {
      "epoch": 1.6446768570662376,
      "grad_norm": 0.30823491889454835,
      "learning_rate": 5.031203538548475e-06,
      "loss": 0.026,
      "step": 6133
    },
    {
      "epoch": 1.644945025475999,
      "grad_norm": 0.2518707478035237,
      "learning_rate": 5.029643380382023e-06,
      "loss": 0.0222,
      "step": 6134
    },
    {
      "epoch": 1.6452131938857604,
      "grad_norm": 0.2771416620034353,
      "learning_rate": 5.028083219329274e-06,
      "loss": 0.0277,
      "step": 6135
    },
    {
      "epoch": 1.6454813622955216,
      "grad_norm": 0.21601617624630448,
      "learning_rate": 5.026523055542137e-06,
      "loss": 0.0196,
      "step": 6136
    },
    {
      "epoch": 1.645749530705283,
      "grad_norm": 0.22492964582708713,
      "learning_rate": 5.024962889172521e-06,
      "loss": 0.0195,
      "step": 6137
    },
    {
      "epoch": 1.6460176991150441,
      "grad_norm": 0.21458198278153567,
      "learning_rate": 5.023402720372336e-06,
      "loss": 0.0166,
      "step": 6138
    },
    {
      "epoch": 1.6462858675248055,
      "grad_norm": 0.3791692395576329,
      "learning_rate": 5.0218425492934885e-06,
      "loss": 0.0257,
      "step": 6139
    },
    {
      "epoch": 1.646554035934567,
      "grad_norm": 0.1906638900915563,
      "learning_rate": 5.0202823760878895e-06,
      "loss": 0.021,
      "step": 6140
    },
    {
      "epoch": 1.6468222043443284,
      "grad_norm": 0.22461107377310655,
      "learning_rate": 5.01872220090745e-06,
      "loss": 0.0199,
      "step": 6141
    },
    {
      "epoch": 1.6470903727540895,
      "grad_norm": 0.2854405400539301,
      "learning_rate": 5.01716202390408e-06,
      "loss": 0.0223,
      "step": 6142
    },
    {
      "epoch": 1.647358541163851,
      "grad_norm": 0.24648452362140388,
      "learning_rate": 5.015601845229689e-06,
      "loss": 0.0214,
      "step": 6143
    },
    {
      "epoch": 1.6476267095736121,
      "grad_norm": 0.2719032134933166,
      "learning_rate": 5.014041665036188e-06,
      "loss": 0.0245,
      "step": 6144
    },
    {
      "epoch": 1.6478948779833735,
      "grad_norm": 0.25971498493430406,
      "learning_rate": 5.012481483475487e-06,
      "loss": 0.023,
      "step": 6145
    },
    {
      "epoch": 1.648163046393135,
      "grad_norm": 0.24443304645049677,
      "learning_rate": 5.010921300699496e-06,
      "loss": 0.0237,
      "step": 6146
    },
    {
      "epoch": 1.6484312148028963,
      "grad_norm": 1.4825258691671008,
      "learning_rate": 5.009361116860129e-06,
      "loss": 0.0298,
      "step": 6147
    },
    {
      "epoch": 1.6486993832126575,
      "grad_norm": 0.23378382024070332,
      "learning_rate": 5.007800932109294e-06,
      "loss": 0.0199,
      "step": 6148
    },
    {
      "epoch": 1.648967551622419,
      "grad_norm": 0.29686400432411814,
      "learning_rate": 5.006240746598903e-06,
      "loss": 0.0224,
      "step": 6149
    },
    {
      "epoch": 1.64923572003218,
      "grad_norm": 0.2778656312914394,
      "learning_rate": 5.004680560480867e-06,
      "loss": 0.0207,
      "step": 6150
    },
    {
      "epoch": 1.6495038884419415,
      "grad_norm": 0.24278804216190678,
      "learning_rate": 5.003120373907097e-06,
      "loss": 0.0238,
      "step": 6151
    },
    {
      "epoch": 1.649772056851703,
      "grad_norm": 0.25119924973138547,
      "learning_rate": 5.001560187029504e-06,
      "loss": 0.0211,
      "step": 6152
    },
    {
      "epoch": 1.6500402252614643,
      "grad_norm": 0.24211970386266002,
      "learning_rate": 5e-06,
      "loss": 0.0181,
      "step": 6153
    },
    {
      "epoch": 1.6503083936712255,
      "grad_norm": 0.5013893010478379,
      "learning_rate": 4.998439812970497e-06,
      "loss": 0.0249,
      "step": 6154
    },
    {
      "epoch": 1.650576562080987,
      "grad_norm": 0.2558768253063386,
      "learning_rate": 4.996879626092905e-06,
      "loss": 0.0199,
      "step": 6155
    },
    {
      "epoch": 1.650844730490748,
      "grad_norm": 0.2303959086260235,
      "learning_rate": 4.995319439519135e-06,
      "loss": 0.0197,
      "step": 6156
    },
    {
      "epoch": 1.6511128989005095,
      "grad_norm": 0.29805525445199327,
      "learning_rate": 4.993759253401098e-06,
      "loss": 0.0219,
      "step": 6157
    },
    {
      "epoch": 1.651381067310271,
      "grad_norm": 0.277233716207636,
      "learning_rate": 4.992199067890706e-06,
      "loss": 0.0294,
      "step": 6158
    },
    {
      "epoch": 1.6516492357200323,
      "grad_norm": 0.3356773507598051,
      "learning_rate": 4.990638883139872e-06,
      "loss": 0.0266,
      "step": 6159
    },
    {
      "epoch": 1.6519174041297935,
      "grad_norm": 0.24076851670671096,
      "learning_rate": 4.989078699300506e-06,
      "loss": 0.0202,
      "step": 6160
    },
    {
      "epoch": 1.652185572539555,
      "grad_norm": 0.7107944079502402,
      "learning_rate": 4.987518516524516e-06,
      "loss": 0.031,
      "step": 6161
    },
    {
      "epoch": 1.652453740949316,
      "grad_norm": 0.20879493738803492,
      "learning_rate": 4.985958334963814e-06,
      "loss": 0.0189,
      "step": 6162
    },
    {
      "epoch": 1.6527219093590775,
      "grad_norm": 1.0488750835186145,
      "learning_rate": 4.984398154770313e-06,
      "loss": 0.0287,
      "step": 6163
    },
    {
      "epoch": 1.6529900777688389,
      "grad_norm": 0.25427660525306484,
      "learning_rate": 4.982837976095922e-06,
      "loss": 0.0231,
      "step": 6164
    },
    {
      "epoch": 1.6532582461786003,
      "grad_norm": 0.2541122716445635,
      "learning_rate": 4.981277799092552e-06,
      "loss": 0.0219,
      "step": 6165
    },
    {
      "epoch": 1.6535264145883615,
      "grad_norm": 0.25981135939691236,
      "learning_rate": 4.979717623912112e-06,
      "loss": 0.0232,
      "step": 6166
    },
    {
      "epoch": 1.6537945829981229,
      "grad_norm": 0.34772785923116034,
      "learning_rate": 4.978157450706513e-06,
      "loss": 0.0338,
      "step": 6167
    },
    {
      "epoch": 1.654062751407884,
      "grad_norm": 0.23742788465566605,
      "learning_rate": 4.976597279627666e-06,
      "loss": 0.022,
      "step": 6168
    },
    {
      "epoch": 1.6543309198176455,
      "grad_norm": 0.3400946105740804,
      "learning_rate": 4.975037110827479e-06,
      "loss": 0.0351,
      "step": 6169
    },
    {
      "epoch": 1.6545990882274069,
      "grad_norm": 0.2267906729363588,
      "learning_rate": 4.973476944457863e-06,
      "loss": 0.0181,
      "step": 6170
    },
    {
      "epoch": 1.6548672566371683,
      "grad_norm": 0.39808383379433115,
      "learning_rate": 4.9719167806707265e-06,
      "loss": 0.0308,
      "step": 6171
    },
    {
      "epoch": 1.6551354250469295,
      "grad_norm": 0.3118057576019345,
      "learning_rate": 4.970356619617979e-06,
      "loss": 0.019,
      "step": 6172
    },
    {
      "epoch": 1.6554035934566906,
      "grad_norm": 0.28529835824501487,
      "learning_rate": 4.968796461451528e-06,
      "loss": 0.0186,
      "step": 6173
    },
    {
      "epoch": 1.655671761866452,
      "grad_norm": 0.25359484548992256,
      "learning_rate": 4.967236306323283e-06,
      "loss": 0.0214,
      "step": 6174
    },
    {
      "epoch": 1.6559399302762134,
      "grad_norm": 0.20736358203270808,
      "learning_rate": 4.965676154385151e-06,
      "loss": 0.0238,
      "step": 6175
    },
    {
      "epoch": 1.6562080986859748,
      "grad_norm": 0.19689528295429204,
      "learning_rate": 4.964116005789042e-06,
      "loss": 0.0192,
      "step": 6176
    },
    {
      "epoch": 1.6564762670957363,
      "grad_norm": 0.2600759461141454,
      "learning_rate": 4.962555860686861e-06,
      "loss": 0.0271,
      "step": 6177
    },
    {
      "epoch": 1.6567444355054974,
      "grad_norm": 0.3364115725642357,
      "learning_rate": 4.960995719230516e-06,
      "loss": 0.0388,
      "step": 6178
    },
    {
      "epoch": 1.6570126039152586,
      "grad_norm": 0.22916523317337809,
      "learning_rate": 4.9594355815719155e-06,
      "loss": 0.0216,
      "step": 6179
    },
    {
      "epoch": 1.65728077232502,
      "grad_norm": 0.22515940129017173,
      "learning_rate": 4.957875447862964e-06,
      "loss": 0.0224,
      "step": 6180
    },
    {
      "epoch": 1.6575489407347814,
      "grad_norm": 0.6924541070110336,
      "learning_rate": 4.956315318255569e-06,
      "loss": 0.0227,
      "step": 6181
    },
    {
      "epoch": 1.6578171091445428,
      "grad_norm": 0.25949980571586984,
      "learning_rate": 4.954755192901635e-06,
      "loss": 0.0216,
      "step": 6182
    },
    {
      "epoch": 1.6580852775543042,
      "grad_norm": 0.3409524754679679,
      "learning_rate": 4.953195071953069e-06,
      "loss": 0.0359,
      "step": 6183
    },
    {
      "epoch": 1.6583534459640654,
      "grad_norm": 0.32242139702372924,
      "learning_rate": 4.951634955561774e-06,
      "loss": 0.0206,
      "step": 6184
    },
    {
      "epoch": 1.6586216143738266,
      "grad_norm": 0.250284433218115,
      "learning_rate": 4.9500748438796545e-06,
      "loss": 0.0228,
      "step": 6185
    },
    {
      "epoch": 1.658889782783588,
      "grad_norm": 0.2947322972473216,
      "learning_rate": 4.948514737058615e-06,
      "loss": 0.0212,
      "step": 6186
    },
    {
      "epoch": 1.6591579511933494,
      "grad_norm": 0.6010174902729412,
      "learning_rate": 4.946954635250558e-06,
      "loss": 0.0245,
      "step": 6187
    },
    {
      "epoch": 1.6594261196031108,
      "grad_norm": 0.2703662506341328,
      "learning_rate": 4.945394538607388e-06,
      "loss": 0.0198,
      "step": 6188
    },
    {
      "epoch": 1.6596942880128722,
      "grad_norm": 0.21735997454713535,
      "learning_rate": 4.943834447281007e-06,
      "loss": 0.027,
      "step": 6189
    },
    {
      "epoch": 1.6599624564226334,
      "grad_norm": 0.24119598812394463,
      "learning_rate": 4.942274361423316e-06,
      "loss": 0.0264,
      "step": 6190
    },
    {
      "epoch": 1.6602306248323946,
      "grad_norm": 0.3858199920346831,
      "learning_rate": 4.940714281186219e-06,
      "loss": 0.0309,
      "step": 6191
    },
    {
      "epoch": 1.660498793242156,
      "grad_norm": 0.4388666048354723,
      "learning_rate": 4.939154206721614e-06,
      "loss": 0.0203,
      "step": 6192
    },
    {
      "epoch": 1.6607669616519174,
      "grad_norm": 0.31200839155819476,
      "learning_rate": 4.937594138181402e-06,
      "loss": 0.0273,
      "step": 6193
    },
    {
      "epoch": 1.6610351300616788,
      "grad_norm": 0.18783997027436997,
      "learning_rate": 4.936034075717482e-06,
      "loss": 0.0174,
      "step": 6194
    },
    {
      "epoch": 1.6613032984714402,
      "grad_norm": 0.2299260346666292,
      "learning_rate": 4.934474019481755e-06,
      "loss": 0.0158,
      "step": 6195
    },
    {
      "epoch": 1.6615714668812014,
      "grad_norm": 0.17236641844227232,
      "learning_rate": 4.932913969626122e-06,
      "loss": 0.0168,
      "step": 6196
    },
    {
      "epoch": 1.6618396352909626,
      "grad_norm": 0.30771126941958554,
      "learning_rate": 4.931353926302475e-06,
      "loss": 0.0331,
      "step": 6197
    },
    {
      "epoch": 1.662107803700724,
      "grad_norm": 1.2097126361748678,
      "learning_rate": 4.929793889662713e-06,
      "loss": 0.0228,
      "step": 6198
    },
    {
      "epoch": 1.6623759721104854,
      "grad_norm": 0.23614330205139014,
      "learning_rate": 4.9282338598587345e-06,
      "loss": 0.0229,
      "step": 6199
    },
    {
      "epoch": 1.6626441405202468,
      "grad_norm": 0.35045185449640587,
      "learning_rate": 4.926673837042434e-06,
      "loss": 0.0228,
      "step": 6200
    },
    {
      "epoch": 1.6629123089300082,
      "grad_norm": 0.27953354378675693,
      "learning_rate": 4.925113821365709e-06,
      "loss": 0.0242,
      "step": 6201
    },
    {
      "epoch": 1.6631804773397694,
      "grad_norm": 0.34948293206496317,
      "learning_rate": 4.92355381298045e-06,
      "loss": 0.0313,
      "step": 6202
    },
    {
      "epoch": 1.6634486457495306,
      "grad_norm": 0.30250952809969756,
      "learning_rate": 4.921993812038555e-06,
      "loss": 0.0262,
      "step": 6203
    },
    {
      "epoch": 1.663716814159292,
      "grad_norm": 0.29357577438494004,
      "learning_rate": 4.920433818691915e-06,
      "loss": 0.0216,
      "step": 6204
    },
    {
      "epoch": 1.6639849825690534,
      "grad_norm": 0.2280377785960035,
      "learning_rate": 4.918873833092421e-06,
      "loss": 0.0173,
      "step": 6205
    },
    {
      "epoch": 1.6642531509788148,
      "grad_norm": 0.3207970297352996,
      "learning_rate": 4.917313855391968e-06,
      "loss": 0.0259,
      "step": 6206
    },
    {
      "epoch": 1.6645213193885762,
      "grad_norm": 0.31177439601896406,
      "learning_rate": 4.915753885742446e-06,
      "loss": 0.0276,
      "step": 6207
    },
    {
      "epoch": 1.6647894877983374,
      "grad_norm": 0.3124497039447132,
      "learning_rate": 4.9141939242957465e-06,
      "loss": 0.0297,
      "step": 6208
    },
    {
      "epoch": 1.6650576562080985,
      "grad_norm": 0.2898099151039896,
      "learning_rate": 4.912633971203754e-06,
      "loss": 0.0209,
      "step": 6209
    },
    {
      "epoch": 1.66532582461786,
      "grad_norm": 0.34800484303163226,
      "learning_rate": 4.91107402661836e-06,
      "loss": 0.0216,
      "step": 6210
    },
    {
      "epoch": 1.6655939930276213,
      "grad_norm": 0.26100236297065904,
      "learning_rate": 4.909514090691452e-06,
      "loss": 0.0204,
      "step": 6211
    },
    {
      "epoch": 1.6658621614373827,
      "grad_norm": 0.23967511082758283,
      "learning_rate": 4.907954163574917e-06,
      "loss": 0.0149,
      "step": 6212
    },
    {
      "epoch": 1.6661303298471442,
      "grad_norm": 0.2766161703389561,
      "learning_rate": 4.90639424542064e-06,
      "loss": 0.0278,
      "step": 6213
    },
    {
      "epoch": 1.6663984982569053,
      "grad_norm": 0.40412787301831055,
      "learning_rate": 4.904834336380508e-06,
      "loss": 0.0285,
      "step": 6214
    },
    {
      "epoch": 1.6666666666666665,
      "grad_norm": 0.33513385336195617,
      "learning_rate": 4.903274436606402e-06,
      "loss": 0.0238,
      "step": 6215
    },
    {
      "epoch": 1.666934835076428,
      "grad_norm": 0.5253592480616325,
      "learning_rate": 4.9017145462502085e-06,
      "loss": 0.0271,
      "step": 6216
    },
    {
      "epoch": 1.6672030034861893,
      "grad_norm": 0.3573544908279431,
      "learning_rate": 4.9001546654638084e-06,
      "loss": 0.0357,
      "step": 6217
    },
    {
      "epoch": 1.6674711718959507,
      "grad_norm": 0.3010404230565133,
      "learning_rate": 4.898594794399084e-06,
      "loss": 0.0217,
      "step": 6218
    },
    {
      "epoch": 1.6677393403057121,
      "grad_norm": 0.47455039877559896,
      "learning_rate": 4.8970349332079155e-06,
      "loss": 0.0225,
      "step": 6219
    },
    {
      "epoch": 1.6680075087154733,
      "grad_norm": 0.3136234408321886,
      "learning_rate": 4.895475082042182e-06,
      "loss": 0.023,
      "step": 6220
    },
    {
      "epoch": 1.6682756771252345,
      "grad_norm": 0.32937213070687044,
      "learning_rate": 4.8939152410537624e-06,
      "loss": 0.027,
      "step": 6221
    },
    {
      "epoch": 1.668543845534996,
      "grad_norm": 0.22725253011225463,
      "learning_rate": 4.892355410394534e-06,
      "loss": 0.0178,
      "step": 6222
    },
    {
      "epoch": 1.6688120139447573,
      "grad_norm": 0.24566533284389136,
      "learning_rate": 4.890795590216373e-06,
      "loss": 0.0187,
      "step": 6223
    },
    {
      "epoch": 1.6690801823545187,
      "grad_norm": 0.2392447815990188,
      "learning_rate": 4.889235780671156e-06,
      "loss": 0.0204,
      "step": 6224
    },
    {
      "epoch": 1.6693483507642801,
      "grad_norm": 0.3777099239953422,
      "learning_rate": 4.887675981910758e-06,
      "loss": 0.0175,
      "step": 6225
    },
    {
      "epoch": 1.6696165191740413,
      "grad_norm": 0.6188552460311392,
      "learning_rate": 4.8861161940870485e-06,
      "loss": 0.028,
      "step": 6226
    },
    {
      "epoch": 1.6698846875838025,
      "grad_norm": 0.2882985003771321,
      "learning_rate": 4.884556417351904e-06,
      "loss": 0.0262,
      "step": 6227
    },
    {
      "epoch": 1.670152855993564,
      "grad_norm": 0.28665360160732434,
      "learning_rate": 4.882996651857195e-06,
      "loss": 0.0299,
      "step": 6228
    },
    {
      "epoch": 1.6704210244033253,
      "grad_norm": 0.23200212474144313,
      "learning_rate": 4.881436897754792e-06,
      "loss": 0.0239,
      "step": 6229
    },
    {
      "epoch": 1.6706891928130867,
      "grad_norm": 0.25939978062292507,
      "learning_rate": 4.879877155196563e-06,
      "loss": 0.0239,
      "step": 6230
    },
    {
      "epoch": 1.670957361222848,
      "grad_norm": 0.3825557905496175,
      "learning_rate": 4.8783174243343765e-06,
      "loss": 0.0389,
      "step": 6231
    },
    {
      "epoch": 1.6712255296326093,
      "grad_norm": 0.3196018341383587,
      "learning_rate": 4.8767577053200995e-06,
      "loss": 0.03,
      "step": 6232
    },
    {
      "epoch": 1.6714936980423705,
      "grad_norm": 0.2611207693747397,
      "learning_rate": 4.875197998305598e-06,
      "loss": 0.0235,
      "step": 6233
    },
    {
      "epoch": 1.6717618664521319,
      "grad_norm": 0.31331817672865775,
      "learning_rate": 4.873638303442736e-06,
      "loss": 0.0216,
      "step": 6234
    },
    {
      "epoch": 1.6720300348618933,
      "grad_norm": 0.34434648020018555,
      "learning_rate": 4.872078620883375e-06,
      "loss": 0.0238,
      "step": 6235
    },
    {
      "epoch": 1.6722982032716547,
      "grad_norm": 0.298754656036478,
      "learning_rate": 4.87051895077938e-06,
      "loss": 0.0336,
      "step": 6236
    },
    {
      "epoch": 1.672566371681416,
      "grad_norm": 0.41354456536356643,
      "learning_rate": 4.868959293282612e-06,
      "loss": 0.0345,
      "step": 6237
    },
    {
      "epoch": 1.6728345400911773,
      "grad_norm": 0.5907229727097764,
      "learning_rate": 4.867399648544928e-06,
      "loss": 0.0503,
      "step": 6238
    },
    {
      "epoch": 1.6731027085009385,
      "grad_norm": 0.33502611699016016,
      "learning_rate": 4.8658400167181895e-06,
      "loss": 0.0355,
      "step": 6239
    },
    {
      "epoch": 1.6733708769106999,
      "grad_norm": 0.22305208419511505,
      "learning_rate": 4.864280397954252e-06,
      "loss": 0.0183,
      "step": 6240
    },
    {
      "epoch": 1.6736390453204613,
      "grad_norm": 0.2571871582865664,
      "learning_rate": 4.862720792404972e-06,
      "loss": 0.024,
      "step": 6241
    },
    {
      "epoch": 1.6739072137302227,
      "grad_norm": 0.24332121626100825,
      "learning_rate": 4.861161200222202e-06,
      "loss": 0.0204,
      "step": 6242
    },
    {
      "epoch": 1.674175382139984,
      "grad_norm": 0.21537279266052337,
      "learning_rate": 4.8596016215578e-06,
      "loss": 0.0166,
      "step": 6243
    },
    {
      "epoch": 1.6744435505497453,
      "grad_norm": 0.21715795610663519,
      "learning_rate": 4.8580420565636166e-06,
      "loss": 0.0234,
      "step": 6244
    },
    {
      "epoch": 1.6747117189595064,
      "grad_norm": 0.31109998743971046,
      "learning_rate": 4.856482505391499e-06,
      "loss": 0.0299,
      "step": 6245
    },
    {
      "epoch": 1.6749798873692678,
      "grad_norm": 0.33321261418044074,
      "learning_rate": 4.854922968193298e-06,
      "loss": 0.0276,
      "step": 6246
    },
    {
      "epoch": 1.6752480557790292,
      "grad_norm": 0.3354433762144696,
      "learning_rate": 4.853363445120864e-06,
      "loss": 0.0346,
      "step": 6247
    },
    {
      "epoch": 1.6755162241887906,
      "grad_norm": 0.23717027549408892,
      "learning_rate": 4.851803936326041e-06,
      "loss": 0.0179,
      "step": 6248
    },
    {
      "epoch": 1.675784392598552,
      "grad_norm": 0.21892196581109968,
      "learning_rate": 4.850244441960676e-06,
      "loss": 0.0224,
      "step": 6249
    },
    {
      "epoch": 1.6760525610083132,
      "grad_norm": 0.28582952271418016,
      "learning_rate": 4.848684962176612e-06,
      "loss": 0.023,
      "step": 6250
    },
    {
      "epoch": 1.6763207294180744,
      "grad_norm": 0.30536437021563817,
      "learning_rate": 4.847125497125691e-06,
      "loss": 0.0219,
      "step": 6251
    },
    {
      "epoch": 1.6765888978278358,
      "grad_norm": 0.6540955027331902,
      "learning_rate": 4.845566046959756e-06,
      "loss": 0.0301,
      "step": 6252
    },
    {
      "epoch": 1.6768570662375972,
      "grad_norm": 0.19642702873777676,
      "learning_rate": 4.844006611830642e-06,
      "loss": 0.0187,
      "step": 6253
    },
    {
      "epoch": 1.6771252346473586,
      "grad_norm": 0.20567296040165778,
      "learning_rate": 4.842447191890192e-06,
      "loss": 0.021,
      "step": 6254
    },
    {
      "epoch": 1.67739340305712,
      "grad_norm": 0.2289053020925815,
      "learning_rate": 4.8408877872902404e-06,
      "loss": 0.0185,
      "step": 6255
    },
    {
      "epoch": 1.6776615714668812,
      "grad_norm": 0.8811787205644825,
      "learning_rate": 4.839328398182625e-06,
      "loss": 0.0303,
      "step": 6256
    },
    {
      "epoch": 1.6779297398766424,
      "grad_norm": 0.2869935013096887,
      "learning_rate": 4.837769024719173e-06,
      "loss": 0.0184,
      "step": 6257
    },
    {
      "epoch": 1.6781979082864038,
      "grad_norm": 0.2534741648671677,
      "learning_rate": 4.836209667051722e-06,
      "loss": 0.0289,
      "step": 6258
    },
    {
      "epoch": 1.6784660766961652,
      "grad_norm": 0.2522350860456307,
      "learning_rate": 4.834650325332101e-06,
      "loss": 0.0181,
      "step": 6259
    },
    {
      "epoch": 1.6787342451059266,
      "grad_norm": 0.36905823241810504,
      "learning_rate": 4.833090999712138e-06,
      "loss": 0.0209,
      "step": 6260
    },
    {
      "epoch": 1.6790024135156878,
      "grad_norm": 0.2143740879804254,
      "learning_rate": 4.831531690343662e-06,
      "loss": 0.0157,
      "step": 6261
    },
    {
      "epoch": 1.6792705819254492,
      "grad_norm": 0.3116936920628131,
      "learning_rate": 4.829972397378498e-06,
      "loss": 0.0242,
      "step": 6262
    },
    {
      "epoch": 1.6795387503352104,
      "grad_norm": 0.23565216066290012,
      "learning_rate": 4.828413120968469e-06,
      "loss": 0.0205,
      "step": 6263
    },
    {
      "epoch": 1.6798069187449718,
      "grad_norm": 0.2869115052076514,
      "learning_rate": 4.8268538612654005e-06,
      "loss": 0.0285,
      "step": 6264
    },
    {
      "epoch": 1.6800750871547332,
      "grad_norm": 0.34037826905564667,
      "learning_rate": 4.825294618421111e-06,
      "loss": 0.028,
      "step": 6265
    },
    {
      "epoch": 1.6803432555644946,
      "grad_norm": 0.38029819499718504,
      "learning_rate": 4.8237353925874205e-06,
      "loss": 0.0171,
      "step": 6266
    },
    {
      "epoch": 1.6806114239742558,
      "grad_norm": 0.26708579044907993,
      "learning_rate": 4.822176183916147e-06,
      "loss": 0.0244,
      "step": 6267
    },
    {
      "epoch": 1.6808795923840172,
      "grad_norm": 0.24329263095444462,
      "learning_rate": 4.8206169925591085e-06,
      "loss": 0.0252,
      "step": 6268
    },
    {
      "epoch": 1.6811477607937784,
      "grad_norm": 0.25872516581485283,
      "learning_rate": 4.819057818668115e-06,
      "loss": 0.0186,
      "step": 6269
    },
    {
      "epoch": 1.6814159292035398,
      "grad_norm": 0.23702091638054273,
      "learning_rate": 4.817498662394982e-06,
      "loss": 0.0167,
      "step": 6270
    },
    {
      "epoch": 1.6816840976133012,
      "grad_norm": 0.2387810229128028,
      "learning_rate": 4.815939523891519e-06,
      "loss": 0.0238,
      "step": 6271
    },
    {
      "epoch": 1.6819522660230626,
      "grad_norm": 0.2869172819536674,
      "learning_rate": 4.8143804033095365e-06,
      "loss": 0.0241,
      "step": 6272
    },
    {
      "epoch": 1.6822204344328238,
      "grad_norm": 0.2295049722018516,
      "learning_rate": 4.812821300800841e-06,
      "loss": 0.0151,
      "step": 6273
    },
    {
      "epoch": 1.6824886028425852,
      "grad_norm": 0.2249664855227666,
      "learning_rate": 4.8112622165172374e-06,
      "loss": 0.0218,
      "step": 6274
    },
    {
      "epoch": 1.6827567712523464,
      "grad_norm": 0.24728665860098115,
      "learning_rate": 4.809703150610532e-06,
      "loss": 0.0198,
      "step": 6275
    },
    {
      "epoch": 1.6830249396621078,
      "grad_norm": 0.3121812560598261,
      "learning_rate": 4.808144103232525e-06,
      "loss": 0.0199,
      "step": 6276
    },
    {
      "epoch": 1.6832931080718692,
      "grad_norm": 0.3468699691864694,
      "learning_rate": 4.806585074535018e-06,
      "loss": 0.0289,
      "step": 6277
    },
    {
      "epoch": 1.6835612764816306,
      "grad_norm": 0.2578331554775838,
      "learning_rate": 4.8050260646698085e-06,
      "loss": 0.0222,
      "step": 6278
    },
    {
      "epoch": 1.6838294448913917,
      "grad_norm": 0.35315272489694033,
      "learning_rate": 4.803467073788694e-06,
      "loss": 0.0239,
      "step": 6279
    },
    {
      "epoch": 1.6840976133011532,
      "grad_norm": 0.3836240289682844,
      "learning_rate": 4.801908102043467e-06,
      "loss": 0.0298,
      "step": 6280
    },
    {
      "epoch": 1.6843657817109143,
      "grad_norm": 0.295912208442318,
      "learning_rate": 4.800349149585926e-06,
      "loss": 0.0311,
      "step": 6281
    },
    {
      "epoch": 1.6846339501206757,
      "grad_norm": 0.2918125801933405,
      "learning_rate": 4.798790216567855e-06,
      "loss": 0.0218,
      "step": 6282
    },
    {
      "epoch": 1.6849021185304371,
      "grad_norm": 0.2276152979012104,
      "learning_rate": 4.797231303141046e-06,
      "loss": 0.0218,
      "step": 6283
    },
    {
      "epoch": 1.6851702869401985,
      "grad_norm": 0.3169084240024691,
      "learning_rate": 4.7956724094572875e-06,
      "loss": 0.0234,
      "step": 6284
    },
    {
      "epoch": 1.6854384553499597,
      "grad_norm": 0.2657307743035358,
      "learning_rate": 4.794113535668364e-06,
      "loss": 0.0277,
      "step": 6285
    },
    {
      "epoch": 1.6857066237597211,
      "grad_norm": 0.5833083127092006,
      "learning_rate": 4.792554681926059e-06,
      "loss": 0.0295,
      "step": 6286
    },
    {
      "epoch": 1.6859747921694823,
      "grad_norm": 0.2898574922016076,
      "learning_rate": 4.790995848382155e-06,
      "loss": 0.0353,
      "step": 6287
    },
    {
      "epoch": 1.6862429605792437,
      "grad_norm": 0.3071474411402569,
      "learning_rate": 4.789437035188429e-06,
      "loss": 0.0298,
      "step": 6288
    },
    {
      "epoch": 1.6865111289890051,
      "grad_norm": 0.22853000189798037,
      "learning_rate": 4.78787824249666e-06,
      "loss": 0.0213,
      "step": 6289
    },
    {
      "epoch": 1.6867792973987665,
      "grad_norm": 0.2888919265737888,
      "learning_rate": 4.786319470458623e-06,
      "loss": 0.0201,
      "step": 6290
    },
    {
      "epoch": 1.6870474658085277,
      "grad_norm": 0.5774403861127113,
      "learning_rate": 4.7847607192260916e-06,
      "loss": 0.0231,
      "step": 6291
    },
    {
      "epoch": 1.6873156342182891,
      "grad_norm": 0.24846526012438114,
      "learning_rate": 4.783201988950839e-06,
      "loss": 0.0206,
      "step": 6292
    },
    {
      "epoch": 1.6875838026280503,
      "grad_norm": 0.23008040308026992,
      "learning_rate": 4.781643279784634e-06,
      "loss": 0.0216,
      "step": 6293
    },
    {
      "epoch": 1.6878519710378117,
      "grad_norm": 0.23492649226686793,
      "learning_rate": 4.78008459187924e-06,
      "loss": 0.0221,
      "step": 6294
    },
    {
      "epoch": 1.688120139447573,
      "grad_norm": 0.2479433293250395,
      "learning_rate": 4.778525925386428e-06,
      "loss": 0.0214,
      "step": 6295
    },
    {
      "epoch": 1.6883883078573345,
      "grad_norm": 0.29170408003263393,
      "learning_rate": 4.776967280457958e-06,
      "loss": 0.0218,
      "step": 6296
    },
    {
      "epoch": 1.6886564762670957,
      "grad_norm": 0.35125627257341807,
      "learning_rate": 4.775408657245591e-06,
      "loss": 0.0245,
      "step": 6297
    },
    {
      "epoch": 1.688924644676857,
      "grad_norm": 0.31770460569189707,
      "learning_rate": 4.773850055901087e-06,
      "loss": 0.0275,
      "step": 6298
    },
    {
      "epoch": 1.6891928130866183,
      "grad_norm": 0.23659505127107036,
      "learning_rate": 4.772291476576203e-06,
      "loss": 0.0153,
      "step": 6299
    },
    {
      "epoch": 1.6894609814963797,
      "grad_norm": 0.21211417595374113,
      "learning_rate": 4.770732919422693e-06,
      "loss": 0.0192,
      "step": 6300
    },
    {
      "epoch": 1.689729149906141,
      "grad_norm": 0.27809539007935324,
      "learning_rate": 4.769174384592309e-06,
      "loss": 0.0195,
      "step": 6301
    },
    {
      "epoch": 1.6899973183159025,
      "grad_norm": 0.2790284931813096,
      "learning_rate": 4.767615872236804e-06,
      "loss": 0.0268,
      "step": 6302
    },
    {
      "epoch": 1.6902654867256637,
      "grad_norm": 0.34085502751266017,
      "learning_rate": 4.766057382507924e-06,
      "loss": 0.0253,
      "step": 6303
    },
    {
      "epoch": 1.690533655135425,
      "grad_norm": 0.5369143384605487,
      "learning_rate": 4.764498915557415e-06,
      "loss": 0.0318,
      "step": 6304
    },
    {
      "epoch": 1.6908018235451863,
      "grad_norm": 0.3432143055774044,
      "learning_rate": 4.762940471537023e-06,
      "loss": 0.0266,
      "step": 6305
    },
    {
      "epoch": 1.6910699919549477,
      "grad_norm": 0.27977175914961383,
      "learning_rate": 4.761382050598487e-06,
      "loss": 0.021,
      "step": 6306
    },
    {
      "epoch": 1.691338160364709,
      "grad_norm": 0.48627656837281513,
      "learning_rate": 4.759823652893547e-06,
      "loss": 0.0176,
      "step": 6307
    },
    {
      "epoch": 1.6916063287744705,
      "grad_norm": 0.28772215548885527,
      "learning_rate": 4.7582652785739405e-06,
      "loss": 0.0175,
      "step": 6308
    },
    {
      "epoch": 1.6918744971842317,
      "grad_norm": 0.688231324568534,
      "learning_rate": 4.7567069277914015e-06,
      "loss": 0.0235,
      "step": 6309
    },
    {
      "epoch": 1.692142665593993,
      "grad_norm": 0.19922517745761928,
      "learning_rate": 4.7551486006976635e-06,
      "loss": 0.0194,
      "step": 6310
    },
    {
      "epoch": 1.6924108340037542,
      "grad_norm": 0.24667781412463077,
      "learning_rate": 4.7535902974444555e-06,
      "loss": 0.0202,
      "step": 6311
    },
    {
      "epoch": 1.6926790024135157,
      "grad_norm": 0.25207793101525133,
      "learning_rate": 4.752032018183506e-06,
      "loss": 0.0199,
      "step": 6312
    },
    {
      "epoch": 1.692947170823277,
      "grad_norm": 0.2776264193532967,
      "learning_rate": 4.7504737630665415e-06,
      "loss": 0.0264,
      "step": 6313
    },
    {
      "epoch": 1.6932153392330385,
      "grad_norm": 0.32088960937767774,
      "learning_rate": 4.748915532245284e-06,
      "loss": 0.0257,
      "step": 6314
    },
    {
      "epoch": 1.6934835076427996,
      "grad_norm": 0.2432793248746285,
      "learning_rate": 4.747357325871454e-06,
      "loss": 0.0202,
      "step": 6315
    },
    {
      "epoch": 1.693751676052561,
      "grad_norm": 0.3049679032671894,
      "learning_rate": 4.745799144096771e-06,
      "loss": 0.0291,
      "step": 6316
    },
    {
      "epoch": 1.6940198444623222,
      "grad_norm": 0.3171388912523261,
      "learning_rate": 4.7442409870729515e-06,
      "loss": 0.0319,
      "step": 6317
    },
    {
      "epoch": 1.6942880128720836,
      "grad_norm": 0.2812546901113973,
      "learning_rate": 4.742682854951707e-06,
      "loss": 0.0282,
      "step": 6318
    },
    {
      "epoch": 1.694556181281845,
      "grad_norm": 0.19141470144703276,
      "learning_rate": 4.741124747884749e-06,
      "loss": 0.0163,
      "step": 6319
    },
    {
      "epoch": 1.6948243496916064,
      "grad_norm": 0.20716836513579462,
      "learning_rate": 4.7395666660237875e-06,
      "loss": 0.0163,
      "step": 6320
    },
    {
      "epoch": 1.6950925181013676,
      "grad_norm": 0.25896220227710987,
      "learning_rate": 4.738008609520528e-06,
      "loss": 0.0264,
      "step": 6321
    },
    {
      "epoch": 1.695360686511129,
      "grad_norm": 0.2965498330808392,
      "learning_rate": 4.7364505785266745e-06,
      "loss": 0.031,
      "step": 6322
    },
    {
      "epoch": 1.6956288549208902,
      "grad_norm": 0.27264077191870656,
      "learning_rate": 4.7348925731939284e-06,
      "loss": 0.0222,
      "step": 6323
    },
    {
      "epoch": 1.6958970233306516,
      "grad_norm": 0.8778555907344555,
      "learning_rate": 4.73333459367399e-06,
      "loss": 0.0435,
      "step": 6324
    },
    {
      "epoch": 1.696165191740413,
      "grad_norm": 0.3052699430561915,
      "learning_rate": 4.7317766401185526e-06,
      "loss": 0.0213,
      "step": 6325
    },
    {
      "epoch": 1.6964333601501744,
      "grad_norm": 0.4084887913366384,
      "learning_rate": 4.730218712679313e-06,
      "loss": 0.0302,
      "step": 6326
    },
    {
      "epoch": 1.6967015285599356,
      "grad_norm": 0.3794880910073861,
      "learning_rate": 4.72866081150796e-06,
      "loss": 0.0271,
      "step": 6327
    },
    {
      "epoch": 1.696969696969697,
      "grad_norm": 0.23805193813393927,
      "learning_rate": 4.727102936756182e-06,
      "loss": 0.0256,
      "step": 6328
    },
    {
      "epoch": 1.6972378653794582,
      "grad_norm": 0.27476507243719384,
      "learning_rate": 4.7255450885756715e-06,
      "loss": 0.0341,
      "step": 6329
    },
    {
      "epoch": 1.6975060337892196,
      "grad_norm": 0.2544707302982708,
      "learning_rate": 4.723987267118104e-06,
      "loss": 0.0265,
      "step": 6330
    },
    {
      "epoch": 1.697774202198981,
      "grad_norm": 0.28900254412485393,
      "learning_rate": 4.722429472535163e-06,
      "loss": 0.0226,
      "step": 6331
    },
    {
      "epoch": 1.6980423706087424,
      "grad_norm": 0.22219667525937103,
      "learning_rate": 4.720871704978527e-06,
      "loss": 0.0169,
      "step": 6332
    },
    {
      "epoch": 1.6983105390185036,
      "grad_norm": 0.6221915190987494,
      "learning_rate": 4.719313964599872e-06,
      "loss": 0.0193,
      "step": 6333
    },
    {
      "epoch": 1.698578707428265,
      "grad_norm": 0.22977728301444644,
      "learning_rate": 4.717756251550873e-06,
      "loss": 0.0197,
      "step": 6334
    },
    {
      "epoch": 1.6988468758380262,
      "grad_norm": 0.1999820524106122,
      "learning_rate": 4.7161985659831965e-06,
      "loss": 0.0197,
      "step": 6335
    },
    {
      "epoch": 1.6991150442477876,
      "grad_norm": 0.21939446591698422,
      "learning_rate": 4.714640908048513e-06,
      "loss": 0.0205,
      "step": 6336
    },
    {
      "epoch": 1.699383212657549,
      "grad_norm": 0.45029681448276454,
      "learning_rate": 4.713083277898487e-06,
      "loss": 0.0272,
      "step": 6337
    },
    {
      "epoch": 1.6996513810673104,
      "grad_norm": 0.24615787353965504,
      "learning_rate": 4.71152567568478e-06,
      "loss": 0.0184,
      "step": 6338
    },
    {
      "epoch": 1.6999195494770716,
      "grad_norm": 0.2642746308775313,
      "learning_rate": 4.709968101559051e-06,
      "loss": 0.0292,
      "step": 6339
    },
    {
      "epoch": 1.700187717886833,
      "grad_norm": 0.3734068638883106,
      "learning_rate": 4.708410555672959e-06,
      "loss": 0.0286,
      "step": 6340
    },
    {
      "epoch": 1.7004558862965942,
      "grad_norm": 0.23210075879332442,
      "learning_rate": 4.706853038178159e-06,
      "loss": 0.0175,
      "step": 6341
    },
    {
      "epoch": 1.7007240547063556,
      "grad_norm": 0.24502943848092656,
      "learning_rate": 4.705295549226298e-06,
      "loss": 0.0287,
      "step": 6342
    },
    {
      "epoch": 1.700992223116117,
      "grad_norm": 0.5181188086924858,
      "learning_rate": 4.703738088969027e-06,
      "loss": 0.0161,
      "step": 6343
    },
    {
      "epoch": 1.7012603915258784,
      "grad_norm": 0.2543993043994865,
      "learning_rate": 4.702180657557992e-06,
      "loss": 0.0248,
      "step": 6344
    },
    {
      "epoch": 1.7015285599356396,
      "grad_norm": 0.39582768305361116,
      "learning_rate": 4.700623255144836e-06,
      "loss": 0.0212,
      "step": 6345
    },
    {
      "epoch": 1.701796728345401,
      "grad_norm": 0.27506846570551235,
      "learning_rate": 4.699065881881199e-06,
      "loss": 0.0263,
      "step": 6346
    },
    {
      "epoch": 1.7020648967551621,
      "grad_norm": 0.20851567791630915,
      "learning_rate": 4.697508537918718e-06,
      "loss": 0.0207,
      "step": 6347
    },
    {
      "epoch": 1.7023330651649236,
      "grad_norm": 0.2284271920304712,
      "learning_rate": 4.695951223409027e-06,
      "loss": 0.0202,
      "step": 6348
    },
    {
      "epoch": 1.702601233574685,
      "grad_norm": 0.2884981162959601,
      "learning_rate": 4.694393938503758e-06,
      "loss": 0.0346,
      "step": 6349
    },
    {
      "epoch": 1.7028694019844464,
      "grad_norm": 0.2290169105839306,
      "learning_rate": 4.69283668335454e-06,
      "loss": 0.0231,
      "step": 6350
    },
    {
      "epoch": 1.7031375703942075,
      "grad_norm": 0.26134288237591163,
      "learning_rate": 4.6912794581129985e-06,
      "loss": 0.023,
      "step": 6351
    },
    {
      "epoch": 1.7034057388039687,
      "grad_norm": 0.3025765356927052,
      "learning_rate": 4.689722262930757e-06,
      "loss": 0.0337,
      "step": 6352
    },
    {
      "epoch": 1.7036739072137301,
      "grad_norm": 0.6042970536707212,
      "learning_rate": 4.688165097959436e-06,
      "loss": 0.0292,
      "step": 6353
    },
    {
      "epoch": 1.7039420756234915,
      "grad_norm": 0.351462532966335,
      "learning_rate": 4.686607963350651e-06,
      "loss": 0.0196,
      "step": 6354
    },
    {
      "epoch": 1.704210244033253,
      "grad_norm": 0.21502543418359332,
      "learning_rate": 4.685050859256016e-06,
      "loss": 0.0172,
      "step": 6355
    },
    {
      "epoch": 1.7044784124430143,
      "grad_norm": 0.6167146726303089,
      "learning_rate": 4.683493785827142e-06,
      "loss": 0.027,
      "step": 6356
    },
    {
      "epoch": 1.7047465808527755,
      "grad_norm": 0.30325661538503634,
      "learning_rate": 4.681936743215639e-06,
      "loss": 0.0223,
      "step": 6357
    },
    {
      "epoch": 1.7050147492625367,
      "grad_norm": 0.480340114941563,
      "learning_rate": 4.6803797315731105e-06,
      "loss": 0.0183,
      "step": 6358
    },
    {
      "epoch": 1.7052829176722981,
      "grad_norm": 0.3265955608107701,
      "learning_rate": 4.6788227510511574e-06,
      "loss": 0.0292,
      "step": 6359
    },
    {
      "epoch": 1.7055510860820595,
      "grad_norm": 0.3203826259528443,
      "learning_rate": 4.677265801801381e-06,
      "loss": 0.0209,
      "step": 6360
    },
    {
      "epoch": 1.705819254491821,
      "grad_norm": 0.29200179442589275,
      "learning_rate": 4.675708883975378e-06,
      "loss": 0.0244,
      "step": 6361
    },
    {
      "epoch": 1.7060874229015823,
      "grad_norm": 0.3419050571252299,
      "learning_rate": 4.6741519977247395e-06,
      "loss": 0.0239,
      "step": 6362
    },
    {
      "epoch": 1.7063555913113435,
      "grad_norm": 0.3888286137005385,
      "learning_rate": 4.672595143201056e-06,
      "loss": 0.0214,
      "step": 6363
    },
    {
      "epoch": 1.7066237597211047,
      "grad_norm": 0.35690950348030837,
      "learning_rate": 4.671038320555914e-06,
      "loss": 0.0209,
      "step": 6364
    },
    {
      "epoch": 1.706891928130866,
      "grad_norm": 0.25258727286638094,
      "learning_rate": 4.669481529940898e-06,
      "loss": 0.0199,
      "step": 6365
    },
    {
      "epoch": 1.7071600965406275,
      "grad_norm": 0.20964608752150893,
      "learning_rate": 4.667924771507588e-06,
      "loss": 0.0201,
      "step": 6366
    },
    {
      "epoch": 1.707428264950389,
      "grad_norm": 0.28034749108572227,
      "learning_rate": 4.66636804540756e-06,
      "loss": 0.0263,
      "step": 6367
    },
    {
      "epoch": 1.7076964333601503,
      "grad_norm": 0.28981534077986654,
      "learning_rate": 4.66481135179239e-06,
      "loss": 0.0269,
      "step": 6368
    },
    {
      "epoch": 1.7079646017699115,
      "grad_norm": 0.2918889591186924,
      "learning_rate": 4.6632546908136485e-06,
      "loss": 0.0276,
      "step": 6369
    },
    {
      "epoch": 1.7082327701796727,
      "grad_norm": 0.3392958239757785,
      "learning_rate": 4.6616980626229035e-06,
      "loss": 0.0231,
      "step": 6370
    },
    {
      "epoch": 1.708500938589434,
      "grad_norm": 0.21715828398995005,
      "learning_rate": 4.660141467371721e-06,
      "loss": 0.0203,
      "step": 6371
    },
    {
      "epoch": 1.7087691069991955,
      "grad_norm": 0.27390462159956414,
      "learning_rate": 4.65858490521166e-06,
      "loss": 0.0321,
      "step": 6372
    },
    {
      "epoch": 1.709037275408957,
      "grad_norm": 0.2144885215041506,
      "learning_rate": 4.657028376294282e-06,
      "loss": 0.0243,
      "step": 6373
    },
    {
      "epoch": 1.7093054438187183,
      "grad_norm": 0.2569661132113024,
      "learning_rate": 4.65547188077114e-06,
      "loss": 0.0201,
      "step": 6374
    },
    {
      "epoch": 1.7095736122284795,
      "grad_norm": 0.6061871034696534,
      "learning_rate": 4.653915418793786e-06,
      "loss": 0.0282,
      "step": 6375
    },
    {
      "epoch": 1.7098417806382407,
      "grad_norm": 0.2208418442614102,
      "learning_rate": 4.652358990513768e-06,
      "loss": 0.02,
      "step": 6376
    },
    {
      "epoch": 1.710109949048002,
      "grad_norm": 0.21293640868509356,
      "learning_rate": 4.650802596082635e-06,
      "loss": 0.0152,
      "step": 6377
    },
    {
      "epoch": 1.7103781174577635,
      "grad_norm": 0.36145123522996625,
      "learning_rate": 4.649246235651924e-06,
      "loss": 0.0237,
      "step": 6378
    },
    {
      "epoch": 1.7106462858675249,
      "grad_norm": 0.2004306247805123,
      "learning_rate": 4.647689909373177e-06,
      "loss": 0.0172,
      "step": 6379
    },
    {
      "epoch": 1.7109144542772863,
      "grad_norm": 0.25057747641026273,
      "learning_rate": 4.646133617397926e-06,
      "loss": 0.0182,
      "step": 6380
    },
    {
      "epoch": 1.7111826226870475,
      "grad_norm": 0.2404041274416568,
      "learning_rate": 4.644577359877707e-06,
      "loss": 0.0195,
      "step": 6381
    },
    {
      "epoch": 1.7114507910968086,
      "grad_norm": 0.24961038717265038,
      "learning_rate": 4.643021136964045e-06,
      "loss": 0.0163,
      "step": 6382
    },
    {
      "epoch": 1.71171895950657,
      "grad_norm": 0.6088510004065834,
      "learning_rate": 4.641464948808469e-06,
      "loss": 0.0276,
      "step": 6383
    },
    {
      "epoch": 1.7119871279163315,
      "grad_norm": 0.4054811489487194,
      "learning_rate": 4.6399087955624986e-06,
      "loss": 0.0342,
      "step": 6384
    },
    {
      "epoch": 1.7122552963260929,
      "grad_norm": 0.3511439766043749,
      "learning_rate": 4.638352677377652e-06,
      "loss": 0.0268,
      "step": 6385
    },
    {
      "epoch": 1.7125234647358543,
      "grad_norm": 0.47862065454059155,
      "learning_rate": 4.636796594405446e-06,
      "loss": 0.0234,
      "step": 6386
    },
    {
      "epoch": 1.7127916331456154,
      "grad_norm": 0.2562139056172085,
      "learning_rate": 4.63524054679739e-06,
      "loss": 0.0181,
      "step": 6387
    },
    {
      "epoch": 1.7130598015553766,
      "grad_norm": 0.23886231636990105,
      "learning_rate": 4.633684534704994e-06,
      "loss": 0.0232,
      "step": 6388
    },
    {
      "epoch": 1.713327969965138,
      "grad_norm": 0.27953331527246256,
      "learning_rate": 4.632128558279765e-06,
      "loss": 0.0265,
      "step": 6389
    },
    {
      "epoch": 1.7135961383748994,
      "grad_norm": 0.30283132133069074,
      "learning_rate": 4.630572617673198e-06,
      "loss": 0.0253,
      "step": 6390
    },
    {
      "epoch": 1.7138643067846608,
      "grad_norm": 0.3458181467931392,
      "learning_rate": 4.629016713036796e-06,
      "loss": 0.0233,
      "step": 6391
    },
    {
      "epoch": 1.7141324751944222,
      "grad_norm": 0.24099334684317164,
      "learning_rate": 4.627460844522052e-06,
      "loss": 0.0209,
      "step": 6392
    },
    {
      "epoch": 1.7144006436041834,
      "grad_norm": 0.297409239774494,
      "learning_rate": 4.625905012280455e-06,
      "loss": 0.0225,
      "step": 6393
    },
    {
      "epoch": 1.7146688120139446,
      "grad_norm": 0.36283418601111095,
      "learning_rate": 4.624349216463495e-06,
      "loss": 0.0219,
      "step": 6394
    },
    {
      "epoch": 1.714936980423706,
      "grad_norm": 0.2132015368426918,
      "learning_rate": 4.622793457222655e-06,
      "loss": 0.0192,
      "step": 6395
    },
    {
      "epoch": 1.7152051488334674,
      "grad_norm": 0.2185737026319438,
      "learning_rate": 4.621237734709414e-06,
      "loss": 0.0199,
      "step": 6396
    },
    {
      "epoch": 1.7154733172432288,
      "grad_norm": 0.21405495661617283,
      "learning_rate": 4.619682049075247e-06,
      "loss": 0.0201,
      "step": 6397
    },
    {
      "epoch": 1.7157414856529902,
      "grad_norm": 0.22595533867447962,
      "learning_rate": 4.618126400471631e-06,
      "loss": 0.0156,
      "step": 6398
    },
    {
      "epoch": 1.7160096540627514,
      "grad_norm": 0.24579561336061392,
      "learning_rate": 4.616570789050034e-06,
      "loss": 0.0244,
      "step": 6399
    },
    {
      "epoch": 1.7162778224725126,
      "grad_norm": 0.2006160706291946,
      "learning_rate": 4.615015214961921e-06,
      "loss": 0.0245,
      "step": 6400
    },
    {
      "epoch": 1.716545990882274,
      "grad_norm": 0.292840726954951,
      "learning_rate": 4.6134596783587545e-06,
      "loss": 0.0242,
      "step": 6401
    },
    {
      "epoch": 1.7168141592920354,
      "grad_norm": 0.2227734680946169,
      "learning_rate": 4.611904179391994e-06,
      "loss": 0.0224,
      "step": 6402
    },
    {
      "epoch": 1.7170823277017968,
      "grad_norm": 0.2782502528780826,
      "learning_rate": 4.610348718213091e-06,
      "loss": 0.0201,
      "step": 6403
    },
    {
      "epoch": 1.7173504961115582,
      "grad_norm": 0.26251425414886725,
      "learning_rate": 4.6087932949735e-06,
      "loss": 0.0248,
      "step": 6404
    },
    {
      "epoch": 1.7176186645213194,
      "grad_norm": 0.3159578346490633,
      "learning_rate": 4.607237909824667e-06,
      "loss": 0.0264,
      "step": 6405
    },
    {
      "epoch": 1.7178868329310806,
      "grad_norm": 0.22926726184711335,
      "learning_rate": 4.605682562918036e-06,
      "loss": 0.0234,
      "step": 6406
    },
    {
      "epoch": 1.718155001340842,
      "grad_norm": 0.4190480875532073,
      "learning_rate": 4.604127254405046e-06,
      "loss": 0.0209,
      "step": 6407
    },
    {
      "epoch": 1.7184231697506034,
      "grad_norm": 0.5877977492782358,
      "learning_rate": 4.602571984437135e-06,
      "loss": 0.0314,
      "step": 6408
    },
    {
      "epoch": 1.7186913381603648,
      "grad_norm": 0.31774377563547274,
      "learning_rate": 4.6010167531657345e-06,
      "loss": 0.0197,
      "step": 6409
    },
    {
      "epoch": 1.7189595065701262,
      "grad_norm": 0.29960669808198426,
      "learning_rate": 4.599461560742274e-06,
      "loss": 0.0302,
      "step": 6410
    },
    {
      "epoch": 1.7192276749798874,
      "grad_norm": 0.20843969555383424,
      "learning_rate": 4.5979064073181775e-06,
      "loss": 0.0217,
      "step": 6411
    },
    {
      "epoch": 1.7194958433896486,
      "grad_norm": 0.2516746683809809,
      "learning_rate": 4.5963512930448665e-06,
      "loss": 0.0236,
      "step": 6412
    },
    {
      "epoch": 1.71976401179941,
      "grad_norm": 0.18467585946167467,
      "learning_rate": 4.59479621807376e-06,
      "loss": 0.0147,
      "step": 6413
    },
    {
      "epoch": 1.7200321802091714,
      "grad_norm": 0.2976623812388055,
      "learning_rate": 4.5932411825562686e-06,
      "loss": 0.0265,
      "step": 6414
    },
    {
      "epoch": 1.7203003486189328,
      "grad_norm": 0.29087501312431563,
      "learning_rate": 4.591686186643804e-06,
      "loss": 0.0264,
      "step": 6415
    },
    {
      "epoch": 1.7205685170286942,
      "grad_norm": 0.1936673758862618,
      "learning_rate": 4.59013123048777e-06,
      "loss": 0.018,
      "step": 6416
    },
    {
      "epoch": 1.7208366854384554,
      "grad_norm": 0.23319293221166942,
      "learning_rate": 4.588576314239571e-06,
      "loss": 0.0172,
      "step": 6417
    },
    {
      "epoch": 1.7211048538482165,
      "grad_norm": 0.27132871628213684,
      "learning_rate": 4.587021438050602e-06,
      "loss": 0.0224,
      "step": 6418
    },
    {
      "epoch": 1.721373022257978,
      "grad_norm": 0.4189080153188926,
      "learning_rate": 4.585466602072261e-06,
      "loss": 0.0387,
      "step": 6419
    },
    {
      "epoch": 1.7216411906677394,
      "grad_norm": 0.2714348760447238,
      "learning_rate": 4.583911806455936e-06,
      "loss": 0.0215,
      "step": 6420
    },
    {
      "epoch": 1.7219093590775008,
      "grad_norm": 0.25303363834760845,
      "learning_rate": 4.582357051353014e-06,
      "loss": 0.026,
      "step": 6421
    },
    {
      "epoch": 1.7221775274872622,
      "grad_norm": 0.282215305120375,
      "learning_rate": 4.580802336914877e-06,
      "loss": 0.0172,
      "step": 6422
    },
    {
      "epoch": 1.7224456958970233,
      "grad_norm": 0.2904703912765796,
      "learning_rate": 4.579247663292903e-06,
      "loss": 0.0131,
      "step": 6423
    },
    {
      "epoch": 1.7227138643067845,
      "grad_norm": 0.8511263241464649,
      "learning_rate": 4.577693030638467e-06,
      "loss": 0.0318,
      "step": 6424
    },
    {
      "epoch": 1.722982032716546,
      "grad_norm": 0.1899432557181762,
      "learning_rate": 4.576138439102939e-06,
      "loss": 0.0209,
      "step": 6425
    },
    {
      "epoch": 1.7232502011263073,
      "grad_norm": 0.44431252992603865,
      "learning_rate": 4.5745838888376895e-06,
      "loss": 0.0236,
      "step": 6426
    },
    {
      "epoch": 1.7235183695360687,
      "grad_norm": 0.2693359998028363,
      "learning_rate": 4.5730293799940745e-06,
      "loss": 0.0219,
      "step": 6427
    },
    {
      "epoch": 1.7237865379458301,
      "grad_norm": 0.2011968930253362,
      "learning_rate": 4.571474912723454e-06,
      "loss": 0.015,
      "step": 6428
    },
    {
      "epoch": 1.7240547063555913,
      "grad_norm": 0.2170750760879635,
      "learning_rate": 4.5699204871771844e-06,
      "loss": 0.0207,
      "step": 6429
    },
    {
      "epoch": 1.7243228747653525,
      "grad_norm": 0.36953092681657856,
      "learning_rate": 4.568366103506616e-06,
      "loss": 0.0183,
      "step": 6430
    },
    {
      "epoch": 1.724591043175114,
      "grad_norm": 0.3232965925714277,
      "learning_rate": 4.566811761863093e-06,
      "loss": 0.0336,
      "step": 6431
    },
    {
      "epoch": 1.7248592115848753,
      "grad_norm": 0.212676947473235,
      "learning_rate": 4.565257462397959e-06,
      "loss": 0.0163,
      "step": 6432
    },
    {
      "epoch": 1.7251273799946367,
      "grad_norm": 0.3558702979747367,
      "learning_rate": 4.563703205262553e-06,
      "loss": 0.0261,
      "step": 6433
    },
    {
      "epoch": 1.7253955484043981,
      "grad_norm": 0.4435637556975014,
      "learning_rate": 4.562148990608205e-06,
      "loss": 0.0211,
      "step": 6434
    },
    {
      "epoch": 1.7256637168141593,
      "grad_norm": 0.3438820831116945,
      "learning_rate": 4.560594818586248e-06,
      "loss": 0.0227,
      "step": 6435
    },
    {
      "epoch": 1.7259318852239205,
      "grad_norm": 0.3603280861203436,
      "learning_rate": 4.559040689348007e-06,
      "loss": 0.0217,
      "step": 6436
    },
    {
      "epoch": 1.726200053633682,
      "grad_norm": 0.4490564042589232,
      "learning_rate": 4.557486603044805e-06,
      "loss": 0.0253,
      "step": 6437
    },
    {
      "epoch": 1.7264682220434433,
      "grad_norm": 0.24312458985997554,
      "learning_rate": 4.555932559827958e-06,
      "loss": 0.0225,
      "step": 6438
    },
    {
      "epoch": 1.7267363904532047,
      "grad_norm": 0.3166950316706425,
      "learning_rate": 4.554378559848777e-06,
      "loss": 0.0228,
      "step": 6439
    },
    {
      "epoch": 1.7270045588629659,
      "grad_norm": 0.3054955419968058,
      "learning_rate": 4.552824603258572e-06,
      "loss": 0.0319,
      "step": 6440
    },
    {
      "epoch": 1.7272727272727273,
      "grad_norm": 0.24554769717001518,
      "learning_rate": 4.55127069020865e-06,
      "loss": 0.0218,
      "step": 6441
    },
    {
      "epoch": 1.7275408956824885,
      "grad_norm": 0.2870202611499414,
      "learning_rate": 4.549716820850308e-06,
      "loss": 0.0269,
      "step": 6442
    },
    {
      "epoch": 1.7278090640922499,
      "grad_norm": 0.22628678079288225,
      "learning_rate": 4.5481629953348445e-06,
      "loss": 0.0179,
      "step": 6443
    },
    {
      "epoch": 1.7280772325020113,
      "grad_norm": 0.25315916559438045,
      "learning_rate": 4.54660921381355e-06,
      "loss": 0.0215,
      "step": 6444
    },
    {
      "epoch": 1.7283454009117727,
      "grad_norm": 0.334683159670059,
      "learning_rate": 4.545055476437712e-06,
      "loss": 0.0287,
      "step": 6445
    },
    {
      "epoch": 1.7286135693215339,
      "grad_norm": 0.31759386592107286,
      "learning_rate": 4.543501783358615e-06,
      "loss": 0.0216,
      "step": 6446
    },
    {
      "epoch": 1.7288817377312953,
      "grad_norm": 0.29661195932179135,
      "learning_rate": 4.541948134727538e-06,
      "loss": 0.0202,
      "step": 6447
    },
    {
      "epoch": 1.7291499061410565,
      "grad_norm": 0.2825993214530079,
      "learning_rate": 4.540394530695755e-06,
      "loss": 0.0263,
      "step": 6448
    },
    {
      "epoch": 1.7294180745508179,
      "grad_norm": 0.29270493692302146,
      "learning_rate": 4.538840971414535e-06,
      "loss": 0.0241,
      "step": 6449
    },
    {
      "epoch": 1.7296862429605793,
      "grad_norm": 0.26100436728179294,
      "learning_rate": 4.537287457035147e-06,
      "loss": 0.0216,
      "step": 6450
    },
    {
      "epoch": 1.7299544113703407,
      "grad_norm": 0.35005063879838255,
      "learning_rate": 4.53573398770885e-06,
      "loss": 0.0447,
      "step": 6451
    },
    {
      "epoch": 1.7302225797801019,
      "grad_norm": 0.2489727720738474,
      "learning_rate": 4.534180563586902e-06,
      "loss": 0.0252,
      "step": 6452
    },
    {
      "epoch": 1.7304907481898633,
      "grad_norm": 0.19111093427962636,
      "learning_rate": 4.532627184820556e-06,
      "loss": 0.0159,
      "step": 6453
    },
    {
      "epoch": 1.7307589165996244,
      "grad_norm": 0.2530464899545907,
      "learning_rate": 4.53107385156106e-06,
      "loss": 0.0209,
      "step": 6454
    },
    {
      "epoch": 1.7310270850093858,
      "grad_norm": 0.23644032975061133,
      "learning_rate": 4.5295205639596576e-06,
      "loss": 0.0199,
      "step": 6455
    },
    {
      "epoch": 1.7312952534191473,
      "grad_norm": 1.058208200639574,
      "learning_rate": 4.527967322167589e-06,
      "loss": 0.0294,
      "step": 6456
    },
    {
      "epoch": 1.7315634218289087,
      "grad_norm": 0.2596593704390698,
      "learning_rate": 4.526414126336089e-06,
      "loss": 0.0188,
      "step": 6457
    },
    {
      "epoch": 1.7318315902386698,
      "grad_norm": 0.25836639971373576,
      "learning_rate": 4.524860976616389e-06,
      "loss": 0.0253,
      "step": 6458
    },
    {
      "epoch": 1.7320997586484312,
      "grad_norm": 0.27490766985711773,
      "learning_rate": 4.523307873159714e-06,
      "loss": 0.025,
      "step": 6459
    },
    {
      "epoch": 1.7323679270581924,
      "grad_norm": 0.24715561741475772,
      "learning_rate": 4.5217548161172845e-06,
      "loss": 0.0223,
      "step": 6460
    },
    {
      "epoch": 1.7326360954679538,
      "grad_norm": 0.28262608764647656,
      "learning_rate": 4.5202018056403205e-06,
      "loss": 0.0232,
      "step": 6461
    },
    {
      "epoch": 1.7329042638777152,
      "grad_norm": 0.25568283633105937,
      "learning_rate": 4.518648841880033e-06,
      "loss": 0.0257,
      "step": 6462
    },
    {
      "epoch": 1.7331724322874766,
      "grad_norm": 0.9030726392649076,
      "learning_rate": 4.517095924987629e-06,
      "loss": 0.0247,
      "step": 6463
    },
    {
      "epoch": 1.7334406006972378,
      "grad_norm": 0.2569839859577526,
      "learning_rate": 4.515543055114313e-06,
      "loss": 0.0272,
      "step": 6464
    },
    {
      "epoch": 1.7337087691069992,
      "grad_norm": 0.24626860998587086,
      "learning_rate": 4.513990232411282e-06,
      "loss": 0.0196,
      "step": 6465
    },
    {
      "epoch": 1.7339769375167604,
      "grad_norm": 0.293138551042103,
      "learning_rate": 4.512437457029733e-06,
      "loss": 0.0244,
      "step": 6466
    },
    {
      "epoch": 1.7342451059265218,
      "grad_norm": 0.3069112491509141,
      "learning_rate": 4.510884729120854e-06,
      "loss": 0.0235,
      "step": 6467
    },
    {
      "epoch": 1.7345132743362832,
      "grad_norm": 0.2099436163462238,
      "learning_rate": 4.509332048835831e-06,
      "loss": 0.0166,
      "step": 6468
    },
    {
      "epoch": 1.7347814427460446,
      "grad_norm": 0.30581200350003007,
      "learning_rate": 4.5077794163258436e-06,
      "loss": 0.0222,
      "step": 6469
    },
    {
      "epoch": 1.7350496111558058,
      "grad_norm": 0.21905588396488787,
      "learning_rate": 4.506226831742067e-06,
      "loss": 0.0204,
      "step": 6470
    },
    {
      "epoch": 1.7353177795655672,
      "grad_norm": 0.22406103961009521,
      "learning_rate": 4.504674295235673e-06,
      "loss": 0.0216,
      "step": 6471
    },
    {
      "epoch": 1.7355859479753284,
      "grad_norm": 0.24113953060861756,
      "learning_rate": 4.503121806957827e-06,
      "loss": 0.0228,
      "step": 6472
    },
    {
      "epoch": 1.7358541163850898,
      "grad_norm": 0.2375838482667167,
      "learning_rate": 4.501569367059693e-06,
      "loss": 0.0206,
      "step": 6473
    },
    {
      "epoch": 1.7361222847948512,
      "grad_norm": 0.25021050566314373,
      "learning_rate": 4.500016975692428e-06,
      "loss": 0.0286,
      "step": 6474
    },
    {
      "epoch": 1.7363904532046126,
      "grad_norm": 0.2841938934805119,
      "learning_rate": 4.4984646330071804e-06,
      "loss": 0.0276,
      "step": 6475
    },
    {
      "epoch": 1.7366586216143738,
      "grad_norm": 0.27697248619501563,
      "learning_rate": 4.4969123391550986e-06,
      "loss": 0.0227,
      "step": 6476
    },
    {
      "epoch": 1.7369267900241352,
      "grad_norm": 0.8943681941849546,
      "learning_rate": 4.4953600942873285e-06,
      "loss": 0.0193,
      "step": 6477
    },
    {
      "epoch": 1.7371949584338964,
      "grad_norm": 0.3538927603089793,
      "learning_rate": 4.493807898555006e-06,
      "loss": 0.0211,
      "step": 6478
    },
    {
      "epoch": 1.7374631268436578,
      "grad_norm": 0.31957073225375754,
      "learning_rate": 4.4922557521092645e-06,
      "loss": 0.0236,
      "step": 6479
    },
    {
      "epoch": 1.7377312952534192,
      "grad_norm": 0.291148075935853,
      "learning_rate": 4.490703655101233e-06,
      "loss": 0.0239,
      "step": 6480
    },
    {
      "epoch": 1.7379994636631806,
      "grad_norm": 0.20784384933869116,
      "learning_rate": 4.489151607682033e-06,
      "loss": 0.0139,
      "step": 6481
    },
    {
      "epoch": 1.7382676320729418,
      "grad_norm": 0.358811818471631,
      "learning_rate": 4.487599610002785e-06,
      "loss": 0.0347,
      "step": 6482
    },
    {
      "epoch": 1.7385358004827032,
      "grad_norm": 0.2540495473011931,
      "learning_rate": 4.486047662214602e-06,
      "loss": 0.0193,
      "step": 6483
    },
    {
      "epoch": 1.7388039688924644,
      "grad_norm": 0.3714001253272932,
      "learning_rate": 4.484495764468595e-06,
      "loss": 0.0227,
      "step": 6484
    },
    {
      "epoch": 1.7390721373022258,
      "grad_norm": 0.3346176000637883,
      "learning_rate": 4.482943916915866e-06,
      "loss": 0.0267,
      "step": 6485
    },
    {
      "epoch": 1.7393403057119872,
      "grad_norm": 0.19609119227542604,
      "learning_rate": 4.481392119707516e-06,
      "loss": 0.0164,
      "step": 6486
    },
    {
      "epoch": 1.7396084741217486,
      "grad_norm": 0.25685979813932897,
      "learning_rate": 4.479840372994639e-06,
      "loss": 0.0272,
      "step": 6487
    },
    {
      "epoch": 1.7398766425315098,
      "grad_norm": 0.21554669626296594,
      "learning_rate": 4.478288676928323e-06,
      "loss": 0.0178,
      "step": 6488
    },
    {
      "epoch": 1.7401448109412712,
      "grad_norm": 0.29392994696640956,
      "learning_rate": 4.476737031659652e-06,
      "loss": 0.0243,
      "step": 6489
    },
    {
      "epoch": 1.7404129793510323,
      "grad_norm": 0.2652690122044581,
      "learning_rate": 4.475185437339709e-06,
      "loss": 0.019,
      "step": 6490
    },
    {
      "epoch": 1.7406811477607937,
      "grad_norm": 0.21638443378919595,
      "learning_rate": 4.473633894119566e-06,
      "loss": 0.0195,
      "step": 6491
    },
    {
      "epoch": 1.7409493161705552,
      "grad_norm": 0.2414075086187707,
      "learning_rate": 4.472082402150293e-06,
      "loss": 0.0165,
      "step": 6492
    },
    {
      "epoch": 1.7412174845803166,
      "grad_norm": 0.29728451527539523,
      "learning_rate": 4.470530961582953e-06,
      "loss": 0.0209,
      "step": 6493
    },
    {
      "epoch": 1.7414856529900777,
      "grad_norm": 0.29730796283734645,
      "learning_rate": 4.46897957256861e-06,
      "loss": 0.0275,
      "step": 6494
    },
    {
      "epoch": 1.7417538213998391,
      "grad_norm": 0.23686197439919526,
      "learning_rate": 4.467428235258315e-06,
      "loss": 0.0241,
      "step": 6495
    },
    {
      "epoch": 1.7420219898096003,
      "grad_norm": 0.2457365495072019,
      "learning_rate": 4.465876949803119e-06,
      "loss": 0.0183,
      "step": 6496
    },
    {
      "epoch": 1.7422901582193617,
      "grad_norm": 0.3807462059794738,
      "learning_rate": 4.464325716354068e-06,
      "loss": 0.0253,
      "step": 6497
    },
    {
      "epoch": 1.7425583266291231,
      "grad_norm": 0.30192227098828495,
      "learning_rate": 4.4627745350622e-06,
      "loss": 0.0196,
      "step": 6498
    },
    {
      "epoch": 1.7428264950388845,
      "grad_norm": 0.2820641896269207,
      "learning_rate": 4.461223406078547e-06,
      "loss": 0.0258,
      "step": 6499
    },
    {
      "epoch": 1.7430946634486457,
      "grad_norm": 0.22543569607692057,
      "learning_rate": 4.459672329554141e-06,
      "loss": 0.0169,
      "step": 6500
    },
    {
      "epoch": 1.7433628318584071,
      "grad_norm": 0.21931643050261612,
      "learning_rate": 4.4581213056400065e-06,
      "loss": 0.0197,
      "step": 6501
    },
    {
      "epoch": 1.7436310002681683,
      "grad_norm": 0.2256126373119399,
      "learning_rate": 4.456570334487162e-06,
      "loss": 0.0183,
      "step": 6502
    },
    {
      "epoch": 1.7438991686779297,
      "grad_norm": 0.21283835312360114,
      "learning_rate": 4.4550194162466186e-06,
      "loss": 0.0164,
      "step": 6503
    },
    {
      "epoch": 1.7441673370876911,
      "grad_norm": 0.29344583959686266,
      "learning_rate": 4.45346855106939e-06,
      "loss": 0.0273,
      "step": 6504
    },
    {
      "epoch": 1.7444355054974525,
      "grad_norm": 0.36467223590148035,
      "learning_rate": 4.451917739106477e-06,
      "loss": 0.0215,
      "step": 6505
    },
    {
      "epoch": 1.7447036739072137,
      "grad_norm": 0.2808819641689337,
      "learning_rate": 4.4503669805088795e-06,
      "loss": 0.0231,
      "step": 6506
    },
    {
      "epoch": 1.744971842316975,
      "grad_norm": 0.25427657464387626,
      "learning_rate": 4.448816275427589e-06,
      "loss": 0.018,
      "step": 6507
    },
    {
      "epoch": 1.7452400107267363,
      "grad_norm": 0.4088444785638285,
      "learning_rate": 4.4472656240135944e-06,
      "loss": 0.0279,
      "step": 6508
    },
    {
      "epoch": 1.7455081791364977,
      "grad_norm": 0.30612194341089866,
      "learning_rate": 4.445715026417879e-06,
      "loss": 0.0241,
      "step": 6509
    },
    {
      "epoch": 1.745776347546259,
      "grad_norm": 0.30212596918895857,
      "learning_rate": 4.444164482791421e-06,
      "loss": 0.0315,
      "step": 6510
    },
    {
      "epoch": 1.7460445159560205,
      "grad_norm": 0.25555227875874187,
      "learning_rate": 4.44261399328519e-06,
      "loss": 0.0215,
      "step": 6511
    },
    {
      "epoch": 1.7463126843657817,
      "grad_norm": 0.4050417566582068,
      "learning_rate": 4.441063558050155e-06,
      "loss": 0.0378,
      "step": 6512
    },
    {
      "epoch": 1.746580852775543,
      "grad_norm": 0.25937021127220256,
      "learning_rate": 4.439513177237278e-06,
      "loss": 0.0208,
      "step": 6513
    },
    {
      "epoch": 1.7468490211853043,
      "grad_norm": 0.2984237796435717,
      "learning_rate": 4.437962850997513e-06,
      "loss": 0.0233,
      "step": 6514
    },
    {
      "epoch": 1.7471171895950657,
      "grad_norm": 0.27073554464521254,
      "learning_rate": 4.436412579481814e-06,
      "loss": 0.0244,
      "step": 6515
    },
    {
      "epoch": 1.747385358004827,
      "grad_norm": 0.21261532557665283,
      "learning_rate": 4.434862362841127e-06,
      "loss": 0.0182,
      "step": 6516
    },
    {
      "epoch": 1.7476535264145885,
      "grad_norm": 0.25845891614663813,
      "learning_rate": 4.43331220122639e-06,
      "loss": 0.02,
      "step": 6517
    },
    {
      "epoch": 1.7479216948243497,
      "grad_norm": 0.408710817489067,
      "learning_rate": 4.431762094788539e-06,
      "loss": 0.0194,
      "step": 6518
    },
    {
      "epoch": 1.748189863234111,
      "grad_norm": 0.360335422346512,
      "learning_rate": 4.430212043678506e-06,
      "loss": 0.0234,
      "step": 6519
    },
    {
      "epoch": 1.7484580316438723,
      "grad_norm": 0.24712552031865853,
      "learning_rate": 4.428662048047211e-06,
      "loss": 0.0169,
      "step": 6520
    },
    {
      "epoch": 1.7487262000536337,
      "grad_norm": 0.25596790137806275,
      "learning_rate": 4.427112108045577e-06,
      "loss": 0.0242,
      "step": 6521
    },
    {
      "epoch": 1.748994368463395,
      "grad_norm": 0.2696038836553065,
      "learning_rate": 4.425562223824516e-06,
      "loss": 0.0279,
      "step": 6522
    },
    {
      "epoch": 1.7492625368731565,
      "grad_norm": 0.3816278598487673,
      "learning_rate": 4.424012395534938e-06,
      "loss": 0.0304,
      "step": 6523
    },
    {
      "epoch": 1.7495307052829177,
      "grad_norm": 0.25264010398938386,
      "learning_rate": 4.422462623327741e-06,
      "loss": 0.0188,
      "step": 6524
    },
    {
      "epoch": 1.749798873692679,
      "grad_norm": 0.38698846448960067,
      "learning_rate": 4.420912907353827e-06,
      "loss": 0.0216,
      "step": 6525
    },
    {
      "epoch": 1.7500670421024402,
      "grad_norm": 0.2397148260481921,
      "learning_rate": 4.4193632477640845e-06,
      "loss": 0.02,
      "step": 6526
    },
    {
      "epoch": 1.7503352105122016,
      "grad_norm": 0.2354028156650949,
      "learning_rate": 4.417813644709401e-06,
      "loss": 0.0153,
      "step": 6527
    },
    {
      "epoch": 1.750603378921963,
      "grad_norm": 0.31743057499584115,
      "learning_rate": 4.4162640983406576e-06,
      "loss": 0.0289,
      "step": 6528
    },
    {
      "epoch": 1.7508715473317245,
      "grad_norm": 0.22589095778137178,
      "learning_rate": 4.41471460880873e-06,
      "loss": 0.0182,
      "step": 6529
    },
    {
      "epoch": 1.7511397157414856,
      "grad_norm": 0.24397849366700827,
      "learning_rate": 4.413165176264486e-06,
      "loss": 0.0146,
      "step": 6530
    },
    {
      "epoch": 1.7514078841512468,
      "grad_norm": 0.3887415229182343,
      "learning_rate": 4.41161580085879e-06,
      "loss": 0.0227,
      "step": 6531
    },
    {
      "epoch": 1.7516760525610082,
      "grad_norm": 0.22014004856332256,
      "learning_rate": 4.410066482742502e-06,
      "loss": 0.0173,
      "step": 6532
    },
    {
      "epoch": 1.7519442209707696,
      "grad_norm": 0.21221940604558215,
      "learning_rate": 4.408517222066475e-06,
      "loss": 0.0226,
      "step": 6533
    },
    {
      "epoch": 1.752212389380531,
      "grad_norm": 0.23280535262751043,
      "learning_rate": 4.4069680189815555e-06,
      "loss": 0.0242,
      "step": 6534
    },
    {
      "epoch": 1.7524805577902924,
      "grad_norm": 0.28071235175100595,
      "learning_rate": 4.405418873638587e-06,
      "loss": 0.0248,
      "step": 6535
    },
    {
      "epoch": 1.7527487262000536,
      "grad_norm": 0.3019807266059885,
      "learning_rate": 4.403869786188402e-06,
      "loss": 0.0227,
      "step": 6536
    },
    {
      "epoch": 1.7530168946098148,
      "grad_norm": 0.449080921453452,
      "learning_rate": 4.402320756781834e-06,
      "loss": 0.0237,
      "step": 6537
    },
    {
      "epoch": 1.7532850630195762,
      "grad_norm": 0.40570327690455504,
      "learning_rate": 4.400771785569707e-06,
      "loss": 0.0225,
      "step": 6538
    },
    {
      "epoch": 1.7535532314293376,
      "grad_norm": 0.22825364680278976,
      "learning_rate": 4.39922287270284e-06,
      "loss": 0.0193,
      "step": 6539
    },
    {
      "epoch": 1.753821399839099,
      "grad_norm": 0.40499817569131374,
      "learning_rate": 4.397674018332048e-06,
      "loss": 0.0323,
      "step": 6540
    },
    {
      "epoch": 1.7540895682488604,
      "grad_norm": 0.22502654489241705,
      "learning_rate": 4.396125222608135e-06,
      "loss": 0.0181,
      "step": 6541
    },
    {
      "epoch": 1.7543577366586216,
      "grad_norm": 0.32987327685191936,
      "learning_rate": 4.394576485681908e-06,
      "loss": 0.027,
      "step": 6542
    },
    {
      "epoch": 1.7546259050683828,
      "grad_norm": 0.2901487910018218,
      "learning_rate": 4.39302780770416e-06,
      "loss": 0.0256,
      "step": 6543
    },
    {
      "epoch": 1.7548940734781442,
      "grad_norm": 0.3162872961713637,
      "learning_rate": 4.391479188825685e-06,
      "loss": 0.015,
      "step": 6544
    },
    {
      "epoch": 1.7551622418879056,
      "grad_norm": 0.23497007261101452,
      "learning_rate": 4.389930629197264e-06,
      "loss": 0.0216,
      "step": 6545
    },
    {
      "epoch": 1.755430410297667,
      "grad_norm": 0.3313479747763374,
      "learning_rate": 4.388382128969678e-06,
      "loss": 0.0238,
      "step": 6546
    },
    {
      "epoch": 1.7556985787074284,
      "grad_norm": 0.2122541864861825,
      "learning_rate": 4.386833688293702e-06,
      "loss": 0.0162,
      "step": 6547
    },
    {
      "epoch": 1.7559667471171896,
      "grad_norm": 0.2730854040466608,
      "learning_rate": 4.385285307320101e-06,
      "loss": 0.0203,
      "step": 6548
    },
    {
      "epoch": 1.7562349155269508,
      "grad_norm": 0.3223586988120385,
      "learning_rate": 4.383736986199637e-06,
      "loss": 0.0307,
      "step": 6549
    },
    {
      "epoch": 1.7565030839367122,
      "grad_norm": 0.22277545145014743,
      "learning_rate": 4.382188725083067e-06,
      "loss": 0.0242,
      "step": 6550
    },
    {
      "epoch": 1.7567712523464736,
      "grad_norm": 0.24326848272924081,
      "learning_rate": 4.380640524121139e-06,
      "loss": 0.0235,
      "step": 6551
    },
    {
      "epoch": 1.757039420756235,
      "grad_norm": 0.2276587514413136,
      "learning_rate": 4.3790923834646e-06,
      "loss": 0.0263,
      "step": 6552
    },
    {
      "epoch": 1.7573075891659964,
      "grad_norm": 0.4418116757750436,
      "learning_rate": 4.377544303264187e-06,
      "loss": 0.0212,
      "step": 6553
    },
    {
      "epoch": 1.7575757575757576,
      "grad_norm": 0.2728853888795205,
      "learning_rate": 4.3759962836706335e-06,
      "loss": 0.0215,
      "step": 6554
    },
    {
      "epoch": 1.7578439259855188,
      "grad_norm": 0.35453506518115274,
      "learning_rate": 4.374448324834664e-06,
      "loss": 0.0187,
      "step": 6555
    },
    {
      "epoch": 1.7581120943952802,
      "grad_norm": 0.2388154121816742,
      "learning_rate": 4.372900426907001e-06,
      "loss": 0.0199,
      "step": 6556
    },
    {
      "epoch": 1.7583802628050416,
      "grad_norm": 0.2304444919425137,
      "learning_rate": 4.371352590038358e-06,
      "loss": 0.0194,
      "step": 6557
    },
    {
      "epoch": 1.758648431214803,
      "grad_norm": 0.3046340493735745,
      "learning_rate": 4.369804814379444e-06,
      "loss": 0.0236,
      "step": 6558
    },
    {
      "epoch": 1.7589165996245644,
      "grad_norm": 0.4868312479756422,
      "learning_rate": 4.3682571000809645e-06,
      "loss": 0.0277,
      "step": 6559
    },
    {
      "epoch": 1.7591847680343256,
      "grad_norm": 0.24376103363638615,
      "learning_rate": 4.366709447293612e-06,
      "loss": 0.0251,
      "step": 6560
    },
    {
      "epoch": 1.7594529364440867,
      "grad_norm": 0.2939820559270936,
      "learning_rate": 4.36516185616808e-06,
      "loss": 0.0326,
      "step": 6561
    },
    {
      "epoch": 1.7597211048538481,
      "grad_norm": 0.42947584252657406,
      "learning_rate": 4.363614326855051e-06,
      "loss": 0.0215,
      "step": 6562
    },
    {
      "epoch": 1.7599892732636095,
      "grad_norm": 0.22593132605632218,
      "learning_rate": 4.362066859505206e-06,
      "loss": 0.0157,
      "step": 6563
    },
    {
      "epoch": 1.760257441673371,
      "grad_norm": 0.28119489178972057,
      "learning_rate": 4.360519454269217e-06,
      "loss": 0.0199,
      "step": 6564
    },
    {
      "epoch": 1.7605256100831324,
      "grad_norm": 0.4091881239249858,
      "learning_rate": 4.358972111297753e-06,
      "loss": 0.0186,
      "step": 6565
    },
    {
      "epoch": 1.7607937784928935,
      "grad_norm": 0.20849350309215653,
      "learning_rate": 4.35742483074147e-06,
      "loss": 0.0204,
      "step": 6566
    },
    {
      "epoch": 1.7610619469026547,
      "grad_norm": 0.21730521025640218,
      "learning_rate": 4.355877612751027e-06,
      "loss": 0.0206,
      "step": 6567
    },
    {
      "epoch": 1.7613301153124161,
      "grad_norm": 0.26825021426511597,
      "learning_rate": 4.354330457477068e-06,
      "loss": 0.0245,
      "step": 6568
    },
    {
      "epoch": 1.7615982837221775,
      "grad_norm": 0.2727637142938813,
      "learning_rate": 4.352783365070241e-06,
      "loss": 0.0274,
      "step": 6569
    },
    {
      "epoch": 1.761866452131939,
      "grad_norm": 0.2769239225359918,
      "learning_rate": 4.351236335681179e-06,
      "loss": 0.0164,
      "step": 6570
    },
    {
      "epoch": 1.7621346205417003,
      "grad_norm": 0.22370329671912081,
      "learning_rate": 4.349689369460514e-06,
      "loss": 0.0198,
      "step": 6571
    },
    {
      "epoch": 1.7624027889514615,
      "grad_norm": 0.2637611985272846,
      "learning_rate": 4.348142466558866e-06,
      "loss": 0.0196,
      "step": 6572
    },
    {
      "epoch": 1.7626709573612227,
      "grad_norm": 0.3104679890912792,
      "learning_rate": 4.346595627126856e-06,
      "loss": 0.0164,
      "step": 6573
    },
    {
      "epoch": 1.762939125770984,
      "grad_norm": 0.2658216491005964,
      "learning_rate": 4.345048851315095e-06,
      "loss": 0.0205,
      "step": 6574
    },
    {
      "epoch": 1.7632072941807455,
      "grad_norm": 0.20642752954177654,
      "learning_rate": 4.343502139274189e-06,
      "loss": 0.0183,
      "step": 6575
    },
    {
      "epoch": 1.763475462590507,
      "grad_norm": 0.22056467714912686,
      "learning_rate": 4.3419554911547365e-06,
      "loss": 0.0225,
      "step": 6576
    },
    {
      "epoch": 1.7637436310002683,
      "grad_norm": 0.2085381637499523,
      "learning_rate": 4.340408907107331e-06,
      "loss": 0.0206,
      "step": 6577
    },
    {
      "epoch": 1.7640117994100295,
      "grad_norm": 0.30350482707148163,
      "learning_rate": 4.3388623872825585e-06,
      "loss": 0.0233,
      "step": 6578
    },
    {
      "epoch": 1.7642799678197907,
      "grad_norm": 0.2849709172782828,
      "learning_rate": 4.337315931830999e-06,
      "loss": 0.0227,
      "step": 6579
    },
    {
      "epoch": 1.764548136229552,
      "grad_norm": 0.2542651365145202,
      "learning_rate": 4.3357695409032295e-06,
      "loss": 0.0216,
      "step": 6580
    },
    {
      "epoch": 1.7648163046393135,
      "grad_norm": 0.2826327438094655,
      "learning_rate": 4.334223214649815e-06,
      "loss": 0.0256,
      "step": 6581
    },
    {
      "epoch": 1.765084473049075,
      "grad_norm": 0.2689120885388809,
      "learning_rate": 4.33267695322132e-06,
      "loss": 0.0289,
      "step": 6582
    },
    {
      "epoch": 1.7653526414588363,
      "grad_norm": 0.23496053556283258,
      "learning_rate": 4.331130756768299e-06,
      "loss": 0.0169,
      "step": 6583
    },
    {
      "epoch": 1.7656208098685975,
      "grad_norm": 0.301231505837828,
      "learning_rate": 4.329584625441299e-06,
      "loss": 0.0346,
      "step": 6584
    },
    {
      "epoch": 1.7658889782783587,
      "grad_norm": 0.24854590129953075,
      "learning_rate": 4.328038559390864e-06,
      "loss": 0.0216,
      "step": 6585
    },
    {
      "epoch": 1.76615714668812,
      "grad_norm": 0.22083538140857897,
      "learning_rate": 4.326492558767531e-06,
      "loss": 0.0202,
      "step": 6586
    },
    {
      "epoch": 1.7664253150978815,
      "grad_norm": 0.2897861630179568,
      "learning_rate": 4.324946623721829e-06,
      "loss": 0.0229,
      "step": 6587
    },
    {
      "epoch": 1.7666934835076429,
      "grad_norm": 0.2857102913831269,
      "learning_rate": 4.323400754404282e-06,
      "loss": 0.025,
      "step": 6588
    },
    {
      "epoch": 1.7669616519174043,
      "grad_norm": 0.25771446990462094,
      "learning_rate": 4.321854950965407e-06,
      "loss": 0.0197,
      "step": 6589
    },
    {
      "epoch": 1.7672298203271655,
      "grad_norm": 0.26539367910330064,
      "learning_rate": 4.320309213555715e-06,
      "loss": 0.025,
      "step": 6590
    },
    {
      "epoch": 1.7674979887369267,
      "grad_norm": 0.2264616827536108,
      "learning_rate": 4.318763542325711e-06,
      "loss": 0.0394,
      "step": 6591
    },
    {
      "epoch": 1.767766157146688,
      "grad_norm": 0.31945189718486294,
      "learning_rate": 4.317217937425892e-06,
      "loss": 0.0204,
      "step": 6592
    },
    {
      "epoch": 1.7680343255564495,
      "grad_norm": 0.2142969217486577,
      "learning_rate": 4.315672399006749e-06,
      "loss": 0.0166,
      "step": 6593
    },
    {
      "epoch": 1.7683024939662109,
      "grad_norm": 0.23465187644585472,
      "learning_rate": 4.314126927218768e-06,
      "loss": 0.0227,
      "step": 6594
    },
    {
      "epoch": 1.7685706623759723,
      "grad_norm": 0.28450310807171997,
      "learning_rate": 4.312581522212429e-06,
      "loss": 0.029,
      "step": 6595
    },
    {
      "epoch": 1.7688388307857335,
      "grad_norm": 0.25405642049352506,
      "learning_rate": 4.3110361841382e-06,
      "loss": 0.0201,
      "step": 6596
    },
    {
      "epoch": 1.7691069991954946,
      "grad_norm": 0.23905117818461016,
      "learning_rate": 4.30949091314655e-06,
      "loss": 0.0223,
      "step": 6597
    },
    {
      "epoch": 1.769375167605256,
      "grad_norm": 0.2567288657088796,
      "learning_rate": 4.307945709387935e-06,
      "loss": 0.0224,
      "step": 6598
    },
    {
      "epoch": 1.7696433360150174,
      "grad_norm": 0.2422462014906323,
      "learning_rate": 4.306400573012808e-06,
      "loss": 0.019,
      "step": 6599
    },
    {
      "epoch": 1.7699115044247788,
      "grad_norm": 0.24886571120796347,
      "learning_rate": 4.304855504171618e-06,
      "loss": 0.0267,
      "step": 6600
    },
    {
      "epoch": 1.7701796728345403,
      "grad_norm": 0.3151559384829323,
      "learning_rate": 4.303310503014801e-06,
      "loss": 0.0326,
      "step": 6601
    },
    {
      "epoch": 1.7704478412443014,
      "grad_norm": 0.31400754102830536,
      "learning_rate": 4.301765569692791e-06,
      "loss": 0.037,
      "step": 6602
    },
    {
      "epoch": 1.7707160096540626,
      "grad_norm": 0.3000391082581648,
      "learning_rate": 4.3002207043560135e-06,
      "loss": 0.0171,
      "step": 6603
    },
    {
      "epoch": 1.770984178063824,
      "grad_norm": 0.3805458777459763,
      "learning_rate": 4.298675907154888e-06,
      "loss": 0.0224,
      "step": 6604
    },
    {
      "epoch": 1.7712523464735854,
      "grad_norm": 0.2891592412137003,
      "learning_rate": 4.297131178239828e-06,
      "loss": 0.0215,
      "step": 6605
    },
    {
      "epoch": 1.7715205148833468,
      "grad_norm": 0.33864301742074604,
      "learning_rate": 4.295586517761238e-06,
      "loss": 0.0271,
      "step": 6606
    },
    {
      "epoch": 1.7717886832931082,
      "grad_norm": 0.2190997532747725,
      "learning_rate": 4.294041925869521e-06,
      "loss": 0.0167,
      "step": 6607
    },
    {
      "epoch": 1.7720568517028694,
      "grad_norm": 0.28211744425414365,
      "learning_rate": 4.292497402715066e-06,
      "loss": 0.0183,
      "step": 6608
    },
    {
      "epoch": 1.7723250201126306,
      "grad_norm": 0.2837826267334727,
      "learning_rate": 4.29095294844826e-06,
      "loss": 0.0221,
      "step": 6609
    },
    {
      "epoch": 1.772593188522392,
      "grad_norm": 0.29304468394839417,
      "learning_rate": 4.289408563219482e-06,
      "loss": 0.0299,
      "step": 6610
    },
    {
      "epoch": 1.7728613569321534,
      "grad_norm": 0.2835078102077901,
      "learning_rate": 4.287864247179107e-06,
      "loss": 0.026,
      "step": 6611
    },
    {
      "epoch": 1.7731295253419148,
      "grad_norm": 0.3458443714615249,
      "learning_rate": 4.286320000477499e-06,
      "loss": 0.0351,
      "step": 6612
    },
    {
      "epoch": 1.7733976937516762,
      "grad_norm": 0.7772405112241315,
      "learning_rate": 4.284775823265018e-06,
      "loss": 0.039,
      "step": 6613
    },
    {
      "epoch": 1.7736658621614374,
      "grad_norm": 0.285768535386129,
      "learning_rate": 4.2832317156920165e-06,
      "loss": 0.0213,
      "step": 6614
    },
    {
      "epoch": 1.7739340305711986,
      "grad_norm": 0.2853992890874127,
      "learning_rate": 4.28168767790884e-06,
      "loss": 0.0248,
      "step": 6615
    },
    {
      "epoch": 1.77420219898096,
      "grad_norm": 0.2700787323201239,
      "learning_rate": 4.280143710065826e-06,
      "loss": 0.0202,
      "step": 6616
    },
    {
      "epoch": 1.7744703673907214,
      "grad_norm": 0.3455657918272816,
      "learning_rate": 4.278599812313309e-06,
      "loss": 0.0261,
      "step": 6617
    },
    {
      "epoch": 1.7747385358004828,
      "grad_norm": 0.2698580209550594,
      "learning_rate": 4.277055984801612e-06,
      "loss": 0.0282,
      "step": 6618
    },
    {
      "epoch": 1.775006704210244,
      "grad_norm": 0.30556582883085825,
      "learning_rate": 4.275512227681058e-06,
      "loss": 0.026,
      "step": 6619
    },
    {
      "epoch": 1.7752748726200054,
      "grad_norm": 0.2438454443865886,
      "learning_rate": 4.273968541101951e-06,
      "loss": 0.0239,
      "step": 6620
    },
    {
      "epoch": 1.7755430410297666,
      "grad_norm": 0.2694373769326546,
      "learning_rate": 4.2724249252146e-06,
      "loss": 0.0239,
      "step": 6621
    },
    {
      "epoch": 1.775811209439528,
      "grad_norm": 0.20522869808468355,
      "learning_rate": 4.270881380169303e-06,
      "loss": 0.0182,
      "step": 6622
    },
    {
      "epoch": 1.7760793778492894,
      "grad_norm": 0.2054025402508133,
      "learning_rate": 4.26933790611635e-06,
      "loss": 0.0182,
      "step": 6623
    },
    {
      "epoch": 1.7763475462590508,
      "grad_norm": 0.23689554537916407,
      "learning_rate": 4.267794503206026e-06,
      "loss": 0.0185,
      "step": 6624
    },
    {
      "epoch": 1.776615714668812,
      "grad_norm": 0.27403239441667726,
      "learning_rate": 4.266251171588606e-06,
      "loss": 0.0284,
      "step": 6625
    },
    {
      "epoch": 1.7768838830785734,
      "grad_norm": 0.40455277045562754,
      "learning_rate": 4.264707911414362e-06,
      "loss": 0.0243,
      "step": 6626
    },
    {
      "epoch": 1.7771520514883345,
      "grad_norm": 0.319652170605408,
      "learning_rate": 4.263164722833556e-06,
      "loss": 0.0242,
      "step": 6627
    },
    {
      "epoch": 1.777420219898096,
      "grad_norm": 0.43047627434277724,
      "learning_rate": 4.261621605996444e-06,
      "loss": 0.0271,
      "step": 6628
    },
    {
      "epoch": 1.7776883883078574,
      "grad_norm": 0.2223414149876421,
      "learning_rate": 4.260078561053278e-06,
      "loss": 0.0227,
      "step": 6629
    },
    {
      "epoch": 1.7779565567176188,
      "grad_norm": 0.24430550881236943,
      "learning_rate": 4.258535588154296e-06,
      "loss": 0.0216,
      "step": 6630
    },
    {
      "epoch": 1.77822472512738,
      "grad_norm": 0.22457773850604412,
      "learning_rate": 4.256992687449738e-06,
      "loss": 0.0173,
      "step": 6631
    },
    {
      "epoch": 1.7784928935371414,
      "grad_norm": 0.2982168046736609,
      "learning_rate": 4.255449859089828e-06,
      "loss": 0.0238,
      "step": 6632
    },
    {
      "epoch": 1.7787610619469025,
      "grad_norm": 0.3127617701661512,
      "learning_rate": 4.253907103224788e-06,
      "loss": 0.0249,
      "step": 6633
    },
    {
      "epoch": 1.779029230356664,
      "grad_norm": 0.3291656367646268,
      "learning_rate": 4.252364420004833e-06,
      "loss": 0.0251,
      "step": 6634
    },
    {
      "epoch": 1.7792973987664253,
      "grad_norm": 0.3225332072314181,
      "learning_rate": 4.250821809580169e-06,
      "loss": 0.0352,
      "step": 6635
    },
    {
      "epoch": 1.7795655671761867,
      "grad_norm": 0.24731290273540263,
      "learning_rate": 4.249279272100997e-06,
      "loss": 0.0227,
      "step": 6636
    },
    {
      "epoch": 1.779833735585948,
      "grad_norm": 0.183353280575414,
      "learning_rate": 4.2477368077175074e-06,
      "loss": 0.0166,
      "step": 6637
    },
    {
      "epoch": 1.7801019039957093,
      "grad_norm": 0.2156822787505547,
      "learning_rate": 4.246194416579888e-06,
      "loss": 0.0213,
      "step": 6638
    },
    {
      "epoch": 1.7803700724054705,
      "grad_norm": 0.19218463468083016,
      "learning_rate": 4.2446520988383185e-06,
      "loss": 0.0258,
      "step": 6639
    },
    {
      "epoch": 1.780638240815232,
      "grad_norm": 0.23637738317656248,
      "learning_rate": 4.243109854642968e-06,
      "loss": 0.0196,
      "step": 6640
    },
    {
      "epoch": 1.7809064092249933,
      "grad_norm": 0.27330496861106013,
      "learning_rate": 4.241567684144002e-06,
      "loss": 0.024,
      "step": 6641
    },
    {
      "epoch": 1.7811745776347547,
      "grad_norm": 0.23886036500790972,
      "learning_rate": 4.240025587491576e-06,
      "loss": 0.0175,
      "step": 6642
    },
    {
      "epoch": 1.781442746044516,
      "grad_norm": 0.27525386060529294,
      "learning_rate": 4.238483564835841e-06,
      "loss": 0.0257,
      "step": 6643
    },
    {
      "epoch": 1.7817109144542773,
      "grad_norm": 0.2882858717851429,
      "learning_rate": 4.236941616326941e-06,
      "loss": 0.0272,
      "step": 6644
    },
    {
      "epoch": 1.7819790828640385,
      "grad_norm": 0.2347400865077592,
      "learning_rate": 4.2353997421150095e-06,
      "loss": 0.0249,
      "step": 6645
    },
    {
      "epoch": 1.7822472512738,
      "grad_norm": 0.21453441943170148,
      "learning_rate": 4.233857942350175e-06,
      "loss": 0.0147,
      "step": 6646
    },
    {
      "epoch": 1.7825154196835613,
      "grad_norm": 0.2371919650673341,
      "learning_rate": 4.232316217182556e-06,
      "loss": 0.0163,
      "step": 6647
    },
    {
      "epoch": 1.7827835880933227,
      "grad_norm": 0.2607805258384071,
      "learning_rate": 4.230774566762271e-06,
      "loss": 0.0228,
      "step": 6648
    },
    {
      "epoch": 1.783051756503084,
      "grad_norm": 0.2734213446793632,
      "learning_rate": 4.229232991239424e-06,
      "loss": 0.0215,
      "step": 6649
    },
    {
      "epoch": 1.7833199249128453,
      "grad_norm": 0.355790570257276,
      "learning_rate": 4.227691490764114e-06,
      "loss": 0.0188,
      "step": 6650
    },
    {
      "epoch": 1.7835880933226065,
      "grad_norm": 0.1677779038204809,
      "learning_rate": 4.226150065486434e-06,
      "loss": 0.015,
      "step": 6651
    },
    {
      "epoch": 1.7838562617323679,
      "grad_norm": 0.2695579857489374,
      "learning_rate": 4.224608715556467e-06,
      "loss": 0.0313,
      "step": 6652
    },
    {
      "epoch": 1.7841244301421293,
      "grad_norm": 0.252829111384665,
      "learning_rate": 4.2230674411242904e-06,
      "loss": 0.0194,
      "step": 6653
    },
    {
      "epoch": 1.7843925985518907,
      "grad_norm": 0.4937866565790345,
      "learning_rate": 4.221526242339974e-06,
      "loss": 0.0206,
      "step": 6654
    },
    {
      "epoch": 1.7846607669616519,
      "grad_norm": 0.21933337991268634,
      "learning_rate": 4.219985119353581e-06,
      "loss": 0.0218,
      "step": 6655
    },
    {
      "epoch": 1.7849289353714133,
      "grad_norm": 0.3400303813794031,
      "learning_rate": 4.218444072315169e-06,
      "loss": 0.0254,
      "step": 6656
    },
    {
      "epoch": 1.7851971037811745,
      "grad_norm": 0.28531163506413326,
      "learning_rate": 4.216903101374779e-06,
      "loss": 0.0205,
      "step": 6657
    },
    {
      "epoch": 1.7854652721909359,
      "grad_norm": 0.21210337302631257,
      "learning_rate": 4.215362206682455e-06,
      "loss": 0.0215,
      "step": 6658
    },
    {
      "epoch": 1.7857334406006973,
      "grad_norm": 0.2690984888102947,
      "learning_rate": 4.213821388388231e-06,
      "loss": 0.0254,
      "step": 6659
    },
    {
      "epoch": 1.7860016090104587,
      "grad_norm": 0.2554775458014713,
      "learning_rate": 4.21228064664213e-06,
      "loss": 0.0267,
      "step": 6660
    },
    {
      "epoch": 1.7862697774202199,
      "grad_norm": 0.44127575643964523,
      "learning_rate": 4.210739981594173e-06,
      "loss": 0.0242,
      "step": 6661
    },
    {
      "epoch": 1.7865379458299813,
      "grad_norm": 0.4805196866925903,
      "learning_rate": 4.209199393394367e-06,
      "loss": 0.034,
      "step": 6662
    },
    {
      "epoch": 1.7868061142397424,
      "grad_norm": 0.18466641479117393,
      "learning_rate": 4.207658882192717e-06,
      "loss": 0.0145,
      "step": 6663
    },
    {
      "epoch": 1.7870742826495039,
      "grad_norm": 0.24062821369102835,
      "learning_rate": 4.206118448139217e-06,
      "loss": 0.0165,
      "step": 6664
    },
    {
      "epoch": 1.7873424510592653,
      "grad_norm": 0.4057592044932434,
      "learning_rate": 4.2045780913838565e-06,
      "loss": 0.0323,
      "step": 6665
    },
    {
      "epoch": 1.7876106194690267,
      "grad_norm": 0.24449401586288344,
      "learning_rate": 4.2030378120766164e-06,
      "loss": 0.0331,
      "step": 6666
    },
    {
      "epoch": 1.7878787878787878,
      "grad_norm": 0.29611562893005733,
      "learning_rate": 4.201497610367468e-06,
      "loss": 0.0303,
      "step": 6667
    },
    {
      "epoch": 1.7881469562885492,
      "grad_norm": 0.3123916389928919,
      "learning_rate": 4.199957486406378e-06,
      "loss": 0.0263,
      "step": 6668
    },
    {
      "epoch": 1.7884151246983104,
      "grad_norm": 0.23309353967094745,
      "learning_rate": 4.198417440343303e-06,
      "loss": 0.0182,
      "step": 6669
    },
    {
      "epoch": 1.7886832931080718,
      "grad_norm": 0.49380835765925324,
      "learning_rate": 4.196877472328195e-06,
      "loss": 0.0213,
      "step": 6670
    },
    {
      "epoch": 1.7889514615178332,
      "grad_norm": 0.3222763372938126,
      "learning_rate": 4.1953375825109934e-06,
      "loss": 0.0283,
      "step": 6671
    },
    {
      "epoch": 1.7892196299275946,
      "grad_norm": 0.22420052426870496,
      "learning_rate": 4.193797771041636e-06,
      "loss": 0.0192,
      "step": 6672
    },
    {
      "epoch": 1.7894877983373558,
      "grad_norm": 0.25966430477193336,
      "learning_rate": 4.192258038070048e-06,
      "loss": 0.018,
      "step": 6673
    },
    {
      "epoch": 1.7897559667471172,
      "grad_norm": 0.24389508586166533,
      "learning_rate": 4.190718383746151e-06,
      "loss": 0.0245,
      "step": 6674
    },
    {
      "epoch": 1.7900241351568784,
      "grad_norm": 0.2627313193531553,
      "learning_rate": 4.189178808219856e-06,
      "loss": 0.0259,
      "step": 6675
    },
    {
      "epoch": 1.7902923035666398,
      "grad_norm": 0.2022195252926253,
      "learning_rate": 4.187639311641067e-06,
      "loss": 0.0125,
      "step": 6676
    },
    {
      "epoch": 1.7905604719764012,
      "grad_norm": 0.24133090625837397,
      "learning_rate": 4.186099894159683e-06,
      "loss": 0.0201,
      "step": 6677
    },
    {
      "epoch": 1.7908286403861626,
      "grad_norm": 0.28689997135300455,
      "learning_rate": 4.184560555925592e-06,
      "loss": 0.0329,
      "step": 6678
    },
    {
      "epoch": 1.7910968087959238,
      "grad_norm": 0.3400058649605173,
      "learning_rate": 4.1830212970886726e-06,
      "loss": 0.0243,
      "step": 6679
    },
    {
      "epoch": 1.7913649772056852,
      "grad_norm": 0.34549925543862436,
      "learning_rate": 4.181482117798804e-06,
      "loss": 0.0255,
      "step": 6680
    },
    {
      "epoch": 1.7916331456154464,
      "grad_norm": 0.5544224351823214,
      "learning_rate": 4.1799430182058455e-06,
      "loss": 0.0241,
      "step": 6681
    },
    {
      "epoch": 1.7919013140252078,
      "grad_norm": 0.20483333931320563,
      "learning_rate": 4.178403998459659e-06,
      "loss": 0.0189,
      "step": 6682
    },
    {
      "epoch": 1.7921694824349692,
      "grad_norm": 0.24882463791480655,
      "learning_rate": 4.176865058710093e-06,
      "loss": 0.0301,
      "step": 6683
    },
    {
      "epoch": 1.7924376508447306,
      "grad_norm": 0.24334359663094082,
      "learning_rate": 4.1753261991069925e-06,
      "loss": 0.0168,
      "step": 6684
    },
    {
      "epoch": 1.7927058192544918,
      "grad_norm": 0.29999370509245216,
      "learning_rate": 4.173787419800188e-06,
      "loss": 0.019,
      "step": 6685
    },
    {
      "epoch": 1.7929739876642532,
      "grad_norm": 0.27181358265214045,
      "learning_rate": 4.1722487209395105e-06,
      "loss": 0.0281,
      "step": 6686
    },
    {
      "epoch": 1.7932421560740144,
      "grad_norm": 0.25289061369119714,
      "learning_rate": 4.170710102674777e-06,
      "loss": 0.0233,
      "step": 6687
    },
    {
      "epoch": 1.7935103244837758,
      "grad_norm": 0.33237193067746496,
      "learning_rate": 4.1691715651558e-06,
      "loss": 0.0257,
      "step": 6688
    },
    {
      "epoch": 1.7937784928935372,
      "grad_norm": 0.19949116673577705,
      "learning_rate": 4.1676331085323805e-06,
      "loss": 0.0162,
      "step": 6689
    },
    {
      "epoch": 1.7940466613032986,
      "grad_norm": 0.2598115223396874,
      "learning_rate": 4.166094732954316e-06,
      "loss": 0.0166,
      "step": 6690
    },
    {
      "epoch": 1.7943148297130598,
      "grad_norm": 0.2468456822813418,
      "learning_rate": 4.164556438571394e-06,
      "loss": 0.0267,
      "step": 6691
    },
    {
      "epoch": 1.7945829981228212,
      "grad_norm": 0.2723044786984433,
      "learning_rate": 4.163018225533394e-06,
      "loss": 0.0231,
      "step": 6692
    },
    {
      "epoch": 1.7948511665325824,
      "grad_norm": 0.2048621594059903,
      "learning_rate": 4.161480093990085e-06,
      "loss": 0.0263,
      "step": 6693
    },
    {
      "epoch": 1.7951193349423438,
      "grad_norm": 0.27400691103231534,
      "learning_rate": 4.159942044091235e-06,
      "loss": 0.027,
      "step": 6694
    },
    {
      "epoch": 1.7953875033521052,
      "grad_norm": 0.21666605568021075,
      "learning_rate": 4.158404075986596e-06,
      "loss": 0.0194,
      "step": 6695
    },
    {
      "epoch": 1.7956556717618666,
      "grad_norm": 0.23443305710684675,
      "learning_rate": 4.156866189825919e-06,
      "loss": 0.0223,
      "step": 6696
    },
    {
      "epoch": 1.7959238401716278,
      "grad_norm": 0.20601941173937266,
      "learning_rate": 4.155328385758943e-06,
      "loss": 0.0196,
      "step": 6697
    },
    {
      "epoch": 1.7961920085813892,
      "grad_norm": 0.24045230340890567,
      "learning_rate": 4.1537906639353984e-06,
      "loss": 0.0235,
      "step": 6698
    },
    {
      "epoch": 1.7964601769911503,
      "grad_norm": 0.18517583947282584,
      "learning_rate": 4.152253024505011e-06,
      "loss": 0.0139,
      "step": 6699
    },
    {
      "epoch": 1.7967283454009118,
      "grad_norm": 1.227824663229762,
      "learning_rate": 4.150715467617496e-06,
      "loss": 0.0469,
      "step": 6700
    },
    {
      "epoch": 1.7969965138106732,
      "grad_norm": 0.20497978050103396,
      "learning_rate": 4.14917799342256e-06,
      "loss": 0.018,
      "step": 6701
    },
    {
      "epoch": 1.7972646822204346,
      "grad_norm": 0.2305021146981779,
      "learning_rate": 4.1476406020699035e-06,
      "loss": 0.0172,
      "step": 6702
    },
    {
      "epoch": 1.7975328506301957,
      "grad_norm": 0.28063696584716963,
      "learning_rate": 4.14610329370922e-06,
      "loss": 0.018,
      "step": 6703
    },
    {
      "epoch": 1.797801019039957,
      "grad_norm": 0.28669868868680687,
      "learning_rate": 4.144566068490194e-06,
      "loss": 0.0244,
      "step": 6704
    },
    {
      "epoch": 1.7980691874497183,
      "grad_norm": 0.1998920811528924,
      "learning_rate": 4.143028926562496e-06,
      "loss": 0.0206,
      "step": 6705
    },
    {
      "epoch": 1.7983373558594797,
      "grad_norm": 0.6375603109420804,
      "learning_rate": 4.141491868075796e-06,
      "loss": 0.0308,
      "step": 6706
    },
    {
      "epoch": 1.7986055242692411,
      "grad_norm": 0.2803581789714871,
      "learning_rate": 4.139954893179753e-06,
      "loss": 0.0246,
      "step": 6707
    },
    {
      "epoch": 1.7988736926790025,
      "grad_norm": 0.2553709709015671,
      "learning_rate": 4.13841800202402e-06,
      "loss": 0.0257,
      "step": 6708
    },
    {
      "epoch": 1.7991418610887637,
      "grad_norm": 0.30195671463843493,
      "learning_rate": 4.136881194758239e-06,
      "loss": 0.0332,
      "step": 6709
    },
    {
      "epoch": 1.799410029498525,
      "grad_norm": 0.2709945461966601,
      "learning_rate": 4.1353444715320435e-06,
      "loss": 0.027,
      "step": 6710
    },
    {
      "epoch": 1.7996781979082863,
      "grad_norm": 0.3727611576541873,
      "learning_rate": 4.133807832495062e-06,
      "loss": 0.0239,
      "step": 6711
    },
    {
      "epoch": 1.7999463663180477,
      "grad_norm": 0.25567542171079866,
      "learning_rate": 4.13227127779691e-06,
      "loss": 0.0271,
      "step": 6712
    },
    {
      "epoch": 1.8002145347278091,
      "grad_norm": 0.204581720476797,
      "learning_rate": 4.130734807587202e-06,
      "loss": 0.0236,
      "step": 6713
    },
    {
      "epoch": 1.8004827031375705,
      "grad_norm": 0.4480581152810066,
      "learning_rate": 4.129198422015538e-06,
      "loss": 0.0259,
      "step": 6714
    },
    {
      "epoch": 1.8007508715473317,
      "grad_norm": 0.7823187443641172,
      "learning_rate": 4.127662121231511e-06,
      "loss": 0.0265,
      "step": 6715
    },
    {
      "epoch": 1.801019039957093,
      "grad_norm": 0.23277483801871576,
      "learning_rate": 4.126125905384708e-06,
      "loss": 0.0194,
      "step": 6716
    },
    {
      "epoch": 1.8012872083668543,
      "grad_norm": 0.2596039986991663,
      "learning_rate": 4.124589774624705e-06,
      "loss": 0.0157,
      "step": 6717
    },
    {
      "epoch": 1.8015553767766157,
      "grad_norm": 0.2944941310907993,
      "learning_rate": 4.123053729101072e-06,
      "loss": 0.0162,
      "step": 6718
    },
    {
      "epoch": 1.801823545186377,
      "grad_norm": 0.2154380049989296,
      "learning_rate": 4.121517768963368e-06,
      "loss": 0.0193,
      "step": 6719
    },
    {
      "epoch": 1.8020917135961385,
      "grad_norm": 0.23528207213000266,
      "learning_rate": 4.119981894361148e-06,
      "loss": 0.0152,
      "step": 6720
    },
    {
      "epoch": 1.8023598820058997,
      "grad_norm": 0.2279863941059893,
      "learning_rate": 4.118446105443954e-06,
      "loss": 0.021,
      "step": 6721
    },
    {
      "epoch": 1.8026280504156609,
      "grad_norm": 0.3406190347016827,
      "learning_rate": 4.116910402361321e-06,
      "loss": 0.0247,
      "step": 6722
    },
    {
      "epoch": 1.8028962188254223,
      "grad_norm": 0.21943475353075056,
      "learning_rate": 4.1153747852627775e-06,
      "loss": 0.0222,
      "step": 6723
    },
    {
      "epoch": 1.8031643872351837,
      "grad_norm": 0.21148308557683881,
      "learning_rate": 4.113839254297844e-06,
      "loss": 0.023,
      "step": 6724
    },
    {
      "epoch": 1.803432555644945,
      "grad_norm": 0.2800476033652728,
      "learning_rate": 4.112303809616029e-06,
      "loss": 0.0203,
      "step": 6725
    },
    {
      "epoch": 1.8037007240547065,
      "grad_norm": 0.22080487358032752,
      "learning_rate": 4.110768451366835e-06,
      "loss": 0.0208,
      "step": 6726
    },
    {
      "epoch": 1.8039688924644677,
      "grad_norm": 0.232841307215043,
      "learning_rate": 4.1092331796997565e-06,
      "loss": 0.0166,
      "step": 6727
    },
    {
      "epoch": 1.8042370608742289,
      "grad_norm": 0.5051797158104181,
      "learning_rate": 4.107697994764279e-06,
      "loss": 0.0209,
      "step": 6728
    },
    {
      "epoch": 1.8045052292839903,
      "grad_norm": 0.275117312626732,
      "learning_rate": 4.106162896709879e-06,
      "loss": 0.022,
      "step": 6729
    },
    {
      "epoch": 1.8047733976937517,
      "grad_norm": 0.24154150452115744,
      "learning_rate": 4.104627885686024e-06,
      "loss": 0.0229,
      "step": 6730
    },
    {
      "epoch": 1.805041566103513,
      "grad_norm": 0.217497691804683,
      "learning_rate": 4.1030929618421734e-06,
      "loss": 0.0193,
      "step": 6731
    },
    {
      "epoch": 1.8053097345132745,
      "grad_norm": 0.2199335744863198,
      "learning_rate": 4.101558125327781e-06,
      "loss": 0.0215,
      "step": 6732
    },
    {
      "epoch": 1.8055779029230357,
      "grad_norm": 0.20939191927980935,
      "learning_rate": 4.100023376292287e-06,
      "loss": 0.0181,
      "step": 6733
    },
    {
      "epoch": 1.8058460713327968,
      "grad_norm": 0.2195496023332162,
      "learning_rate": 4.098488714885128e-06,
      "loss": 0.0257,
      "step": 6734
    },
    {
      "epoch": 1.8061142397425582,
      "grad_norm": 0.27899400976937916,
      "learning_rate": 4.096954141255731e-06,
      "loss": 0.0291,
      "step": 6735
    },
    {
      "epoch": 1.8063824081523197,
      "grad_norm": 0.2230556743039783,
      "learning_rate": 4.095419655553511e-06,
      "loss": 0.0183,
      "step": 6736
    },
    {
      "epoch": 1.806650576562081,
      "grad_norm": 0.2873994415925082,
      "learning_rate": 4.093885257927877e-06,
      "loss": 0.026,
      "step": 6737
    },
    {
      "epoch": 1.8069187449718425,
      "grad_norm": 0.28089788550910544,
      "learning_rate": 4.092350948528231e-06,
      "loss": 0.0253,
      "step": 6738
    },
    {
      "epoch": 1.8071869133816036,
      "grad_norm": 0.2418715010373118,
      "learning_rate": 4.090816727503963e-06,
      "loss": 0.0177,
      "step": 6739
    },
    {
      "epoch": 1.8074550817913648,
      "grad_norm": 0.2971758679036104,
      "learning_rate": 4.089282595004456e-06,
      "loss": 0.0296,
      "step": 6740
    },
    {
      "epoch": 1.8077232502011262,
      "grad_norm": 0.39536350742392024,
      "learning_rate": 4.087748551179085e-06,
      "loss": 0.028,
      "step": 6741
    },
    {
      "epoch": 1.8079914186108876,
      "grad_norm": 0.32338214288373573,
      "learning_rate": 4.086214596177215e-06,
      "loss": 0.0325,
      "step": 6742
    },
    {
      "epoch": 1.808259587020649,
      "grad_norm": 0.23977566774657652,
      "learning_rate": 4.084680730148203e-06,
      "loss": 0.022,
      "step": 6743
    },
    {
      "epoch": 1.8085277554304104,
      "grad_norm": 0.3316673780638422,
      "learning_rate": 4.0831469532414e-06,
      "loss": 0.0312,
      "step": 6744
    },
    {
      "epoch": 1.8087959238401716,
      "grad_norm": 0.2626437346661217,
      "learning_rate": 4.081613265606143e-06,
      "loss": 0.0246,
      "step": 6745
    },
    {
      "epoch": 1.8090640922499328,
      "grad_norm": 0.19738911239542084,
      "learning_rate": 4.080079667391764e-06,
      "loss": 0.0238,
      "step": 6746
    },
    {
      "epoch": 1.8093322606596942,
      "grad_norm": 0.18392613338760166,
      "learning_rate": 4.078546158747586e-06,
      "loss": 0.0157,
      "step": 6747
    },
    {
      "epoch": 1.8096004290694556,
      "grad_norm": 0.3463968639828081,
      "learning_rate": 4.077012739822922e-06,
      "loss": 0.029,
      "step": 6748
    },
    {
      "epoch": 1.809868597479217,
      "grad_norm": 0.299426391573603,
      "learning_rate": 4.075479410767077e-06,
      "loss": 0.0206,
      "step": 6749
    },
    {
      "epoch": 1.8101367658889784,
      "grad_norm": 0.2548237033694923,
      "learning_rate": 4.073946171729345e-06,
      "loss": 0.017,
      "step": 6750
    },
    {
      "epoch": 1.8104049342987396,
      "grad_norm": 0.36604643494492584,
      "learning_rate": 4.072413022859019e-06,
      "loss": 0.0313,
      "step": 6751
    },
    {
      "epoch": 1.8106731027085008,
      "grad_norm": 0.22311092784413067,
      "learning_rate": 4.070879964305375e-06,
      "loss": 0.0156,
      "step": 6752
    },
    {
      "epoch": 1.8109412711182622,
      "grad_norm": 0.2864596293380473,
      "learning_rate": 4.06934699621768e-06,
      "loss": 0.0242,
      "step": 6753
    },
    {
      "epoch": 1.8112094395280236,
      "grad_norm": 0.3757063814748383,
      "learning_rate": 4.067814118745196e-06,
      "loss": 0.0258,
      "step": 6754
    },
    {
      "epoch": 1.811477607937785,
      "grad_norm": 0.34204519093042285,
      "learning_rate": 4.066281332037177e-06,
      "loss": 0.0268,
      "step": 6755
    },
    {
      "epoch": 1.8117457763475464,
      "grad_norm": 0.2621929833168547,
      "learning_rate": 4.064748636242866e-06,
      "loss": 0.0195,
      "step": 6756
    },
    {
      "epoch": 1.8120139447573076,
      "grad_norm": 0.19095296990533725,
      "learning_rate": 4.063216031511498e-06,
      "loss": 0.0158,
      "step": 6757
    },
    {
      "epoch": 1.8122821131670688,
      "grad_norm": 0.3425119346371788,
      "learning_rate": 4.061683517992298e-06,
      "loss": 0.0393,
      "step": 6758
    },
    {
      "epoch": 1.8125502815768302,
      "grad_norm": 0.21724192638083836,
      "learning_rate": 4.060151095834482e-06,
      "loss": 0.0269,
      "step": 6759
    },
    {
      "epoch": 1.8128184499865916,
      "grad_norm": 0.28678584327931267,
      "learning_rate": 4.0586187651872576e-06,
      "loss": 0.0189,
      "step": 6760
    },
    {
      "epoch": 1.813086618396353,
      "grad_norm": 0.22954393230681405,
      "learning_rate": 4.057086526199826e-06,
      "loss": 0.023,
      "step": 6761
    },
    {
      "epoch": 1.8133547868061144,
      "grad_norm": 0.2629698468416312,
      "learning_rate": 4.055554379021376e-06,
      "loss": 0.017,
      "step": 6762
    },
    {
      "epoch": 1.8136229552158756,
      "grad_norm": 0.29882126979966056,
      "learning_rate": 4.054022323801088e-06,
      "loss": 0.0293,
      "step": 6763
    },
    {
      "epoch": 1.8138911236256368,
      "grad_norm": 0.24581747260462916,
      "learning_rate": 4.052490360688136e-06,
      "loss": 0.0168,
      "step": 6764
    },
    {
      "epoch": 1.8141592920353982,
      "grad_norm": 0.3093360289318342,
      "learning_rate": 4.050958489831682e-06,
      "loss": 0.0259,
      "step": 6765
    },
    {
      "epoch": 1.8144274604451596,
      "grad_norm": 0.2657436843714907,
      "learning_rate": 4.049426711380879e-06,
      "loss": 0.0169,
      "step": 6766
    },
    {
      "epoch": 1.814695628854921,
      "grad_norm": 0.2238164247076444,
      "learning_rate": 4.047895025484874e-06,
      "loss": 0.0189,
      "step": 6767
    },
    {
      "epoch": 1.8149637972646824,
      "grad_norm": 0.22517291718914573,
      "learning_rate": 4.046363432292803e-06,
      "loss": 0.0165,
      "step": 6768
    },
    {
      "epoch": 1.8152319656744436,
      "grad_norm": 0.1938959493010603,
      "learning_rate": 4.044831931953792e-06,
      "loss": 0.0168,
      "step": 6769
    },
    {
      "epoch": 1.8155001340842047,
      "grad_norm": 0.2539273665245295,
      "learning_rate": 4.043300524616959e-06,
      "loss": 0.0197,
      "step": 6770
    },
    {
      "epoch": 1.8157683024939661,
      "grad_norm": 0.23172378738021313,
      "learning_rate": 4.041769210431414e-06,
      "loss": 0.0177,
      "step": 6771
    },
    {
      "epoch": 1.8160364709037276,
      "grad_norm": 0.2717901896887012,
      "learning_rate": 4.040237989546258e-06,
      "loss": 0.0195,
      "step": 6772
    },
    {
      "epoch": 1.816304639313489,
      "grad_norm": 0.2998323827895008,
      "learning_rate": 4.038706862110581e-06,
      "loss": 0.0258,
      "step": 6773
    },
    {
      "epoch": 1.8165728077232504,
      "grad_norm": 0.2037346963684397,
      "learning_rate": 4.037175828273464e-06,
      "loss": 0.0188,
      "step": 6774
    },
    {
      "epoch": 1.8168409761330115,
      "grad_norm": 0.2507450837111786,
      "learning_rate": 4.035644888183981e-06,
      "loss": 0.0195,
      "step": 6775
    },
    {
      "epoch": 1.8171091445427727,
      "grad_norm": 0.25337597882070734,
      "learning_rate": 4.034114041991194e-06,
      "loss": 0.0244,
      "step": 6776
    },
    {
      "epoch": 1.8173773129525341,
      "grad_norm": 0.2504672897396629,
      "learning_rate": 4.032583289844161e-06,
      "loss": 0.0191,
      "step": 6777
    },
    {
      "epoch": 1.8176454813622955,
      "grad_norm": 0.2747984583872415,
      "learning_rate": 4.0310526318919215e-06,
      "loss": 0.0309,
      "step": 6778
    },
    {
      "epoch": 1.817913649772057,
      "grad_norm": 0.2980035015276697,
      "learning_rate": 4.0295220682835165e-06,
      "loss": 0.0349,
      "step": 6779
    },
    {
      "epoch": 1.8181818181818183,
      "grad_norm": 0.24208563459654803,
      "learning_rate": 4.02799159916797e-06,
      "loss": 0.0144,
      "step": 6780
    },
    {
      "epoch": 1.8184499865915795,
      "grad_norm": 0.203754230039691,
      "learning_rate": 4.0264612246943005e-06,
      "loss": 0.0178,
      "step": 6781
    },
    {
      "epoch": 1.8187181550013407,
      "grad_norm": 0.24106893433842919,
      "learning_rate": 4.024930945011517e-06,
      "loss": 0.0214,
      "step": 6782
    },
    {
      "epoch": 1.8189863234111021,
      "grad_norm": 0.1775754813791592,
      "learning_rate": 4.02340076026862e-06,
      "loss": 0.0139,
      "step": 6783
    },
    {
      "epoch": 1.8192544918208635,
      "grad_norm": 0.37858519432593574,
      "learning_rate": 4.021870670614598e-06,
      "loss": 0.0344,
      "step": 6784
    },
    {
      "epoch": 1.819522660230625,
      "grad_norm": 0.18031010900343794,
      "learning_rate": 4.020340676198431e-06,
      "loss": 0.0173,
      "step": 6785
    },
    {
      "epoch": 1.8197908286403863,
      "grad_norm": 0.2840417034112351,
      "learning_rate": 4.0188107771690915e-06,
      "loss": 0.0243,
      "step": 6786
    },
    {
      "epoch": 1.8200589970501475,
      "grad_norm": 0.2757190835114367,
      "learning_rate": 4.017280973675542e-06,
      "loss": 0.019,
      "step": 6787
    },
    {
      "epoch": 1.8203271654599087,
      "grad_norm": 0.28824303360994463,
      "learning_rate": 4.015751265866734e-06,
      "loss": 0.0237,
      "step": 6788
    },
    {
      "epoch": 1.82059533386967,
      "grad_norm": 0.29892945555152645,
      "learning_rate": 4.014221653891615e-06,
      "loss": 0.0281,
      "step": 6789
    },
    {
      "epoch": 1.8208635022794315,
      "grad_norm": 0.3756929818448466,
      "learning_rate": 4.012692137899115e-06,
      "loss": 0.0256,
      "step": 6790
    },
    {
      "epoch": 1.821131670689193,
      "grad_norm": 0.2056648016688155,
      "learning_rate": 4.011162718038158e-06,
      "loss": 0.0163,
      "step": 6791
    },
    {
      "epoch": 1.8213998390989543,
      "grad_norm": 0.29864134874987647,
      "learning_rate": 4.009633394457664e-06,
      "loss": 0.0168,
      "step": 6792
    },
    {
      "epoch": 1.8216680075087155,
      "grad_norm": 0.3009766078845872,
      "learning_rate": 4.008104167306536e-06,
      "loss": 0.0225,
      "step": 6793
    },
    {
      "epoch": 1.8219361759184767,
      "grad_norm": 0.29875943916046327,
      "learning_rate": 4.006575036733672e-06,
      "loss": 0.0363,
      "step": 6794
    },
    {
      "epoch": 1.822204344328238,
      "grad_norm": 0.2742650673904684,
      "learning_rate": 4.00504600288796e-06,
      "loss": 0.0237,
      "step": 6795
    },
    {
      "epoch": 1.8224725127379995,
      "grad_norm": 0.20830622921579933,
      "learning_rate": 4.003517065918276e-06,
      "loss": 0.0183,
      "step": 6796
    },
    {
      "epoch": 1.8227406811477609,
      "grad_norm": 0.3679081977175839,
      "learning_rate": 4.001988225973491e-06,
      "loss": 0.0203,
      "step": 6797
    },
    {
      "epoch": 1.823008849557522,
      "grad_norm": 0.22142711152058328,
      "learning_rate": 4.000459483202461e-06,
      "loss": 0.0182,
      "step": 6798
    },
    {
      "epoch": 1.8232770179672835,
      "grad_norm": 0.22516563897539243,
      "learning_rate": 3.998930837754039e-06,
      "loss": 0.0168,
      "step": 6799
    },
    {
      "epoch": 1.8235451863770447,
      "grad_norm": 0.35351982823343914,
      "learning_rate": 3.997402289777062e-06,
      "loss": 0.0323,
      "step": 6800
    },
    {
      "epoch": 1.823813354786806,
      "grad_norm": 0.20877106570325116,
      "learning_rate": 3.995873839420366e-06,
      "loss": 0.0197,
      "step": 6801
    },
    {
      "epoch": 1.8240815231965675,
      "grad_norm": 0.21577659199885574,
      "learning_rate": 3.994345486832765e-06,
      "loss": 0.0191,
      "step": 6802
    },
    {
      "epoch": 1.8243496916063289,
      "grad_norm": 0.20263713304583783,
      "learning_rate": 3.992817232163074e-06,
      "loss": 0.0198,
      "step": 6803
    },
    {
      "epoch": 1.82461786001609,
      "grad_norm": 0.2824159740656318,
      "learning_rate": 3.991289075560096e-06,
      "loss": 0.0245,
      "step": 6804
    },
    {
      "epoch": 1.8248860284258515,
      "grad_norm": 0.216934186634219,
      "learning_rate": 3.989761017172622e-06,
      "loss": 0.0214,
      "step": 6805
    },
    {
      "epoch": 1.8251541968356126,
      "grad_norm": 0.43027848493971427,
      "learning_rate": 3.988233057149436e-06,
      "loss": 0.0284,
      "step": 6806
    },
    {
      "epoch": 1.825422365245374,
      "grad_norm": 0.23051301465692267,
      "learning_rate": 3.9867051956393114e-06,
      "loss": 0.0196,
      "step": 6807
    },
    {
      "epoch": 1.8256905336551355,
      "grad_norm": 0.1988892980410211,
      "learning_rate": 3.98517743279101e-06,
      "loss": 0.0158,
      "step": 6808
    },
    {
      "epoch": 1.8259587020648969,
      "grad_norm": 0.35709483893954613,
      "learning_rate": 3.983649768753289e-06,
      "loss": 0.0243,
      "step": 6809
    },
    {
      "epoch": 1.826226870474658,
      "grad_norm": 0.2125081941853152,
      "learning_rate": 3.982122203674891e-06,
      "loss": 0.0195,
      "step": 6810
    },
    {
      "epoch": 1.8264950388844194,
      "grad_norm": 0.2688770451637419,
      "learning_rate": 3.980594737704552e-06,
      "loss": 0.0234,
      "step": 6811
    },
    {
      "epoch": 1.8267632072941806,
      "grad_norm": 0.24043044830035892,
      "learning_rate": 3.979067370990998e-06,
      "loss": 0.0215,
      "step": 6812
    },
    {
      "epoch": 1.827031375703942,
      "grad_norm": 0.2598625661582212,
      "learning_rate": 3.977540103682943e-06,
      "loss": 0.018,
      "step": 6813
    },
    {
      "epoch": 1.8272995441137034,
      "grad_norm": 0.23956972153666029,
      "learning_rate": 3.976012935929093e-06,
      "loss": 0.0247,
      "step": 6814
    },
    {
      "epoch": 1.8275677125234648,
      "grad_norm": 0.23936322454548165,
      "learning_rate": 3.974485867878145e-06,
      "loss": 0.0195,
      "step": 6815
    },
    {
      "epoch": 1.827835880933226,
      "grad_norm": 0.22774332461626628,
      "learning_rate": 3.972958899678786e-06,
      "loss": 0.0161,
      "step": 6816
    },
    {
      "epoch": 1.8281040493429874,
      "grad_norm": 0.4217813254027379,
      "learning_rate": 3.971432031479691e-06,
      "loss": 0.0194,
      "step": 6817
    },
    {
      "epoch": 1.8283722177527486,
      "grad_norm": 0.23866481546410268,
      "learning_rate": 3.969905263429529e-06,
      "loss": 0.0261,
      "step": 6818
    },
    {
      "epoch": 1.82864038616251,
      "grad_norm": 0.17992612296186683,
      "learning_rate": 3.968378595676956e-06,
      "loss": 0.017,
      "step": 6819
    },
    {
      "epoch": 1.8289085545722714,
      "grad_norm": 0.19829157259462094,
      "learning_rate": 3.96685202837062e-06,
      "loss": 0.0177,
      "step": 6820
    },
    {
      "epoch": 1.8291767229820328,
      "grad_norm": 0.22727604953837693,
      "learning_rate": 3.96532556165916e-06,
      "loss": 0.0159,
      "step": 6821
    },
    {
      "epoch": 1.829444891391794,
      "grad_norm": 0.3051590683026175,
      "learning_rate": 3.963799195691202e-06,
      "loss": 0.0189,
      "step": 6822
    },
    {
      "epoch": 1.8297130598015554,
      "grad_norm": 0.20057192286069714,
      "learning_rate": 3.962272930615367e-06,
      "loss": 0.0185,
      "step": 6823
    },
    {
      "epoch": 1.8299812282113166,
      "grad_norm": 0.186962097277256,
      "learning_rate": 3.96074676658026e-06,
      "loss": 0.0126,
      "step": 6824
    },
    {
      "epoch": 1.830249396621078,
      "grad_norm": 0.25601154824880307,
      "learning_rate": 3.959220703734483e-06,
      "loss": 0.0219,
      "step": 6825
    },
    {
      "epoch": 1.8305175650308394,
      "grad_norm": 0.3674955839651876,
      "learning_rate": 3.957694742226622e-06,
      "loss": 0.0191,
      "step": 6826
    },
    {
      "epoch": 1.8307857334406008,
      "grad_norm": 0.26116638812738735,
      "learning_rate": 3.956168882205257e-06,
      "loss": 0.0236,
      "step": 6827
    },
    {
      "epoch": 1.831053901850362,
      "grad_norm": 0.21275240977905405,
      "learning_rate": 3.9546431238189556e-06,
      "loss": 0.017,
      "step": 6828
    },
    {
      "epoch": 1.8313220702601234,
      "grad_norm": 0.3226555855535195,
      "learning_rate": 3.953117467216277e-06,
      "loss": 0.0232,
      "step": 6829
    },
    {
      "epoch": 1.8315902386698846,
      "grad_norm": 0.3442536500584343,
      "learning_rate": 3.951591912545773e-06,
      "loss": 0.0317,
      "step": 6830
    },
    {
      "epoch": 1.831858407079646,
      "grad_norm": 0.19991822717616334,
      "learning_rate": 3.95006645995598e-06,
      "loss": 0.0157,
      "step": 6831
    },
    {
      "epoch": 1.8321265754894074,
      "grad_norm": 0.5828679670967211,
      "learning_rate": 3.948541109595429e-06,
      "loss": 0.0224,
      "step": 6832
    },
    {
      "epoch": 1.8323947438991688,
      "grad_norm": 0.31329316749037833,
      "learning_rate": 3.947015861612639e-06,
      "loss": 0.0243,
      "step": 6833
    },
    {
      "epoch": 1.83266291230893,
      "grad_norm": 0.3191192436104779,
      "learning_rate": 3.945490716156117e-06,
      "loss": 0.0216,
      "step": 6834
    },
    {
      "epoch": 1.8329310807186914,
      "grad_norm": 0.22617607198159442,
      "learning_rate": 3.943965673374367e-06,
      "loss": 0.0131,
      "step": 6835
    },
    {
      "epoch": 1.8331992491284526,
      "grad_norm": 0.24348389254197622,
      "learning_rate": 3.942440733415873e-06,
      "loss": 0.0219,
      "step": 6836
    },
    {
      "epoch": 1.833467417538214,
      "grad_norm": 0.21359357406744642,
      "learning_rate": 3.94091589642912e-06,
      "loss": 0.0139,
      "step": 6837
    },
    {
      "epoch": 1.8337355859479754,
      "grad_norm": 0.33468017246055964,
      "learning_rate": 3.939391162562573e-06,
      "loss": 0.0224,
      "step": 6838
    },
    {
      "epoch": 1.8340037543577368,
      "grad_norm": 0.25589175427869687,
      "learning_rate": 3.937866531964691e-06,
      "loss": 0.0242,
      "step": 6839
    },
    {
      "epoch": 1.834271922767498,
      "grad_norm": 0.24865337370624824,
      "learning_rate": 3.936342004783926e-06,
      "loss": 0.0195,
      "step": 6840
    },
    {
      "epoch": 1.8345400911772594,
      "grad_norm": 0.27366062312724504,
      "learning_rate": 3.934817581168716e-06,
      "loss": 0.0135,
      "step": 6841
    },
    {
      "epoch": 1.8348082595870205,
      "grad_norm": 0.40039656771597726,
      "learning_rate": 3.93329326126749e-06,
      "loss": 0.0238,
      "step": 6842
    },
    {
      "epoch": 1.835076427996782,
      "grad_norm": 0.22879054148184938,
      "learning_rate": 3.931769045228668e-06,
      "loss": 0.0179,
      "step": 6843
    },
    {
      "epoch": 1.8353445964065433,
      "grad_norm": 0.25529385671504007,
      "learning_rate": 3.930244933200657e-06,
      "loss": 0.0245,
      "step": 6844
    },
    {
      "epoch": 1.8356127648163048,
      "grad_norm": 0.28914893650218615,
      "learning_rate": 3.928720925331857e-06,
      "loss": 0.0194,
      "step": 6845
    },
    {
      "epoch": 1.835880933226066,
      "grad_norm": 0.32301510600306277,
      "learning_rate": 3.9271970217706555e-06,
      "loss": 0.023,
      "step": 6846
    },
    {
      "epoch": 1.8361491016358273,
      "grad_norm": 0.21494747235990944,
      "learning_rate": 3.9256732226654325e-06,
      "loss": 0.0202,
      "step": 6847
    },
    {
      "epoch": 1.8364172700455885,
      "grad_norm": 0.23488292999177632,
      "learning_rate": 3.924149528164556e-06,
      "loss": 0.0177,
      "step": 6848
    },
    {
      "epoch": 1.83668543845535,
      "grad_norm": 0.3131259614361377,
      "learning_rate": 3.9226259384163855e-06,
      "loss": 0.0229,
      "step": 6849
    },
    {
      "epoch": 1.8369536068651113,
      "grad_norm": 0.2783242141847321,
      "learning_rate": 3.921102453569264e-06,
      "loss": 0.0184,
      "step": 6850
    },
    {
      "epoch": 1.8372217752748727,
      "grad_norm": 0.229044604854004,
      "learning_rate": 3.919579073771532e-06,
      "loss": 0.0221,
      "step": 6851
    },
    {
      "epoch": 1.837489943684634,
      "grad_norm": 0.21689194706259882,
      "learning_rate": 3.918055799171518e-06,
      "loss": 0.02,
      "step": 6852
    },
    {
      "epoch": 1.8377581120943953,
      "grad_norm": 0.2758134826921344,
      "learning_rate": 3.9165326299175385e-06,
      "loss": 0.0245,
      "step": 6853
    },
    {
      "epoch": 1.8380262805041565,
      "grad_norm": 0.3092085392029247,
      "learning_rate": 3.9150095661579005e-06,
      "loss": 0.0276,
      "step": 6854
    },
    {
      "epoch": 1.838294448913918,
      "grad_norm": 0.1773941236689758,
      "learning_rate": 3.9134866080409e-06,
      "loss": 0.0173,
      "step": 6855
    },
    {
      "epoch": 1.8385626173236793,
      "grad_norm": 0.2930888952838018,
      "learning_rate": 3.911963755714823e-06,
      "loss": 0.0223,
      "step": 6856
    },
    {
      "epoch": 1.8388307857334407,
      "grad_norm": 0.24267788426849224,
      "learning_rate": 3.910441009327947e-06,
      "loss": 0.0217,
      "step": 6857
    },
    {
      "epoch": 1.839098954143202,
      "grad_norm": 0.2880526324637451,
      "learning_rate": 3.908918369028538e-06,
      "loss": 0.0219,
      "step": 6858
    },
    {
      "epoch": 1.8393671225529633,
      "grad_norm": 0.19992320634992655,
      "learning_rate": 3.90739583496485e-06,
      "loss": 0.0152,
      "step": 6859
    },
    {
      "epoch": 1.8396352909627245,
      "grad_norm": 0.22795088920161416,
      "learning_rate": 3.90587340728513e-06,
      "loss": 0.0132,
      "step": 6860
    },
    {
      "epoch": 1.839903459372486,
      "grad_norm": 0.38167979539488595,
      "learning_rate": 3.904351086137611e-06,
      "loss": 0.0193,
      "step": 6861
    },
    {
      "epoch": 1.8401716277822473,
      "grad_norm": 0.25737225202146746,
      "learning_rate": 3.902828871670518e-06,
      "loss": 0.0224,
      "step": 6862
    },
    {
      "epoch": 1.8404397961920087,
      "grad_norm": 0.30221183184085193,
      "learning_rate": 3.901306764032063e-06,
      "loss": 0.0221,
      "step": 6863
    },
    {
      "epoch": 1.8407079646017699,
      "grad_norm": 0.2529562720173654,
      "learning_rate": 3.8997847633704535e-06,
      "loss": 0.0272,
      "step": 6864
    },
    {
      "epoch": 1.8409761330115313,
      "grad_norm": 0.2397132250066566,
      "learning_rate": 3.898262869833879e-06,
      "loss": 0.0202,
      "step": 6865
    },
    {
      "epoch": 1.8412443014212925,
      "grad_norm": 0.2809433690288611,
      "learning_rate": 3.896741083570525e-06,
      "loss": 0.0268,
      "step": 6866
    },
    {
      "epoch": 1.8415124698310539,
      "grad_norm": 0.20066394068501658,
      "learning_rate": 3.895219404728561e-06,
      "loss": 0.0178,
      "step": 6867
    },
    {
      "epoch": 1.8417806382408153,
      "grad_norm": 0.17565602349233542,
      "learning_rate": 3.8936978334561516e-06,
      "loss": 0.0145,
      "step": 6868
    },
    {
      "epoch": 1.8420488066505767,
      "grad_norm": 0.4268600800034085,
      "learning_rate": 3.892176369901446e-06,
      "loss": 0.0271,
      "step": 6869
    },
    {
      "epoch": 1.8423169750603379,
      "grad_norm": 0.19654079281168427,
      "learning_rate": 3.890655014212588e-06,
      "loss": 0.021,
      "step": 6870
    },
    {
      "epoch": 1.8425851434700993,
      "grad_norm": 0.2041703701901406,
      "learning_rate": 3.889133766537705e-06,
      "loss": 0.0154,
      "step": 6871
    },
    {
      "epoch": 1.8428533118798605,
      "grad_norm": 0.23275420660408408,
      "learning_rate": 3.887612627024917e-06,
      "loss": 0.0199,
      "step": 6872
    },
    {
      "epoch": 1.8431214802896219,
      "grad_norm": 0.19357632367642727,
      "learning_rate": 3.8860915958223335e-06,
      "loss": 0.0191,
      "step": 6873
    },
    {
      "epoch": 1.8433896486993833,
      "grad_norm": 0.3224278366745231,
      "learning_rate": 3.884570673078058e-06,
      "loss": 0.027,
      "step": 6874
    },
    {
      "epoch": 1.8436578171091447,
      "grad_norm": 0.19959528199234783,
      "learning_rate": 3.883049858940172e-06,
      "loss": 0.0143,
      "step": 6875
    },
    {
      "epoch": 1.8439259855189059,
      "grad_norm": 0.15741918000248914,
      "learning_rate": 3.881529153556756e-06,
      "loss": 0.0162,
      "step": 6876
    },
    {
      "epoch": 1.8441941539286673,
      "grad_norm": 0.4075349315689493,
      "learning_rate": 3.880008557075875e-06,
      "loss": 0.0261,
      "step": 6877
    },
    {
      "epoch": 1.8444623223384284,
      "grad_norm": 0.31178568204909296,
      "learning_rate": 3.87848806964559e-06,
      "loss": 0.0207,
      "step": 6878
    },
    {
      "epoch": 1.8447304907481898,
      "grad_norm": 0.21134883750081107,
      "learning_rate": 3.8769676914139426e-06,
      "loss": 0.0188,
      "step": 6879
    },
    {
      "epoch": 1.8449986591579512,
      "grad_norm": 0.22902306980986897,
      "learning_rate": 3.87544742252897e-06,
      "loss": 0.0217,
      "step": 6880
    },
    {
      "epoch": 1.8452668275677127,
      "grad_norm": 0.240613679407034,
      "learning_rate": 3.873927263138696e-06,
      "loss": 0.027,
      "step": 6881
    },
    {
      "epoch": 1.8455349959774738,
      "grad_norm": 0.3063596773488676,
      "learning_rate": 3.872407213391135e-06,
      "loss": 0.029,
      "step": 6882
    },
    {
      "epoch": 1.845803164387235,
      "grad_norm": 0.2771864551051406,
      "learning_rate": 3.87088727343429e-06,
      "loss": 0.0279,
      "step": 6883
    },
    {
      "epoch": 1.8460713327969964,
      "grad_norm": 0.24117336067420936,
      "learning_rate": 3.869367443416152e-06,
      "loss": 0.0201,
      "step": 6884
    },
    {
      "epoch": 1.8463395012067578,
      "grad_norm": 0.8634318098260425,
      "learning_rate": 3.867847723484705e-06,
      "loss": 0.0431,
      "step": 6885
    },
    {
      "epoch": 1.8466076696165192,
      "grad_norm": 0.19649903365244265,
      "learning_rate": 3.866328113787923e-06,
      "loss": 0.0168,
      "step": 6886
    },
    {
      "epoch": 1.8468758380262806,
      "grad_norm": 0.3700210095681029,
      "learning_rate": 3.86480861447376e-06,
      "loss": 0.0287,
      "step": 6887
    },
    {
      "epoch": 1.8471440064360418,
      "grad_norm": 0.3337350163514652,
      "learning_rate": 3.863289225690166e-06,
      "loss": 0.013,
      "step": 6888
    },
    {
      "epoch": 1.847412174845803,
      "grad_norm": 0.2373124555373982,
      "learning_rate": 3.861769947585084e-06,
      "loss": 0.023,
      "step": 6889
    },
    {
      "epoch": 1.8476803432555644,
      "grad_norm": 0.22158872838084537,
      "learning_rate": 3.860250780306441e-06,
      "loss": 0.0237,
      "step": 6890
    },
    {
      "epoch": 1.8479485116653258,
      "grad_norm": 0.2453803896127285,
      "learning_rate": 3.858731724002153e-06,
      "loss": 0.0235,
      "step": 6891
    },
    {
      "epoch": 1.8482166800750872,
      "grad_norm": 0.2268365167463081,
      "learning_rate": 3.857212778820127e-06,
      "loss": 0.0219,
      "step": 6892
    },
    {
      "epoch": 1.8484848484848486,
      "grad_norm": 0.22553892905562506,
      "learning_rate": 3.8556939449082584e-06,
      "loss": 0.0257,
      "step": 6893
    },
    {
      "epoch": 1.8487530168946098,
      "grad_norm": 0.4539107942943889,
      "learning_rate": 3.854175222414432e-06,
      "loss": 0.0262,
      "step": 6894
    },
    {
      "epoch": 1.849021185304371,
      "grad_norm": 0.1796585338016776,
      "learning_rate": 3.852656611486522e-06,
      "loss": 0.0096,
      "step": 6895
    },
    {
      "epoch": 1.8492893537141324,
      "grad_norm": 0.27183537425558546,
      "learning_rate": 3.851138112272393e-06,
      "loss": 0.0282,
      "step": 6896
    },
    {
      "epoch": 1.8495575221238938,
      "grad_norm": 0.2463449117573995,
      "learning_rate": 3.849619724919895e-06,
      "loss": 0.0364,
      "step": 6897
    },
    {
      "epoch": 1.8498256905336552,
      "grad_norm": 0.19304805188821675,
      "learning_rate": 3.848101449576871e-06,
      "loss": 0.0201,
      "step": 6898
    },
    {
      "epoch": 1.8500938589434166,
      "grad_norm": 0.17187817169568295,
      "learning_rate": 3.84658328639115e-06,
      "loss": 0.0134,
      "step": 6899
    },
    {
      "epoch": 1.8503620273531778,
      "grad_norm": 0.17592823579543396,
      "learning_rate": 3.8450652355105525e-06,
      "loss": 0.0158,
      "step": 6900
    },
    {
      "epoch": 1.850630195762939,
      "grad_norm": 0.2616139902425976,
      "learning_rate": 3.843547297082886e-06,
      "loss": 0.0226,
      "step": 6901
    },
    {
      "epoch": 1.8508983641727004,
      "grad_norm": 0.30199025938468294,
      "learning_rate": 3.842029471255949e-06,
      "loss": 0.0246,
      "step": 6902
    },
    {
      "epoch": 1.8511665325824618,
      "grad_norm": 0.40386045452013797,
      "learning_rate": 3.840511758177528e-06,
      "loss": 0.0179,
      "step": 6903
    },
    {
      "epoch": 1.8514347009922232,
      "grad_norm": 0.22174483233751435,
      "learning_rate": 3.838994157995398e-06,
      "loss": 0.0309,
      "step": 6904
    },
    {
      "epoch": 1.8517028694019846,
      "grad_norm": 0.541906728761193,
      "learning_rate": 3.837476670857326e-06,
      "loss": 0.0354,
      "step": 6905
    },
    {
      "epoch": 1.8519710378117458,
      "grad_norm": 0.2746545247641939,
      "learning_rate": 3.835959296911063e-06,
      "loss": 0.0288,
      "step": 6906
    },
    {
      "epoch": 1.852239206221507,
      "grad_norm": 0.2634335445212595,
      "learning_rate": 3.834442036304353e-06,
      "loss": 0.0248,
      "step": 6907
    },
    {
      "epoch": 1.8525073746312684,
      "grad_norm": 0.1854678347013182,
      "learning_rate": 3.832924889184928e-06,
      "loss": 0.0137,
      "step": 6908
    },
    {
      "epoch": 1.8527755430410298,
      "grad_norm": 0.21435738902581836,
      "learning_rate": 3.831407855700508e-06,
      "loss": 0.0209,
      "step": 6909
    },
    {
      "epoch": 1.8530437114507912,
      "grad_norm": 0.5612920963719109,
      "learning_rate": 3.829890935998802e-06,
      "loss": 0.0212,
      "step": 6910
    },
    {
      "epoch": 1.8533118798605526,
      "grad_norm": 0.19901517860855003,
      "learning_rate": 3.82837413022751e-06,
      "loss": 0.0152,
      "step": 6911
    },
    {
      "epoch": 1.8535800482703138,
      "grad_norm": 0.2122049907532062,
      "learning_rate": 3.826857438534317e-06,
      "loss": 0.0172,
      "step": 6912
    },
    {
      "epoch": 1.853848216680075,
      "grad_norm": 0.29029020089342544,
      "learning_rate": 3.8253408610669015e-06,
      "loss": 0.0287,
      "step": 6913
    },
    {
      "epoch": 1.8541163850898363,
      "grad_norm": 0.21149485758919676,
      "learning_rate": 3.8238243979729275e-06,
      "loss": 0.0196,
      "step": 6914
    },
    {
      "epoch": 1.8543845534995977,
      "grad_norm": 0.35089621676942184,
      "learning_rate": 3.822308049400047e-06,
      "loss": 0.0215,
      "step": 6915
    },
    {
      "epoch": 1.8546527219093591,
      "grad_norm": 0.6843409134261673,
      "learning_rate": 3.820791815495908e-06,
      "loss": 0.0194,
      "step": 6916
    },
    {
      "epoch": 1.8549208903191206,
      "grad_norm": 0.32253935023872665,
      "learning_rate": 3.819275696408138e-06,
      "loss": 0.0286,
      "step": 6917
    },
    {
      "epoch": 1.8551890587288817,
      "grad_norm": 0.2526319835967422,
      "learning_rate": 3.817759692284358e-06,
      "loss": 0.0215,
      "step": 6918
    },
    {
      "epoch": 1.855457227138643,
      "grad_norm": 0.2588629061232179,
      "learning_rate": 3.816243803272179e-06,
      "loss": 0.0225,
      "step": 6919
    },
    {
      "epoch": 1.8557253955484043,
      "grad_norm": 0.22739556193210014,
      "learning_rate": 3.814728029519198e-06,
      "loss": 0.0187,
      "step": 6920
    },
    {
      "epoch": 1.8559935639581657,
      "grad_norm": 0.23295706503447042,
      "learning_rate": 3.8132123711730012e-06,
      "loss": 0.0176,
      "step": 6921
    },
    {
      "epoch": 1.8562617323679271,
      "grad_norm": 0.2347224468116582,
      "learning_rate": 3.811696828381168e-06,
      "loss": 0.0175,
      "step": 6922
    },
    {
      "epoch": 1.8565299007776885,
      "grad_norm": 0.2994484563086511,
      "learning_rate": 3.8101814012912575e-06,
      "loss": 0.0226,
      "step": 6923
    },
    {
      "epoch": 1.8567980691874497,
      "grad_norm": 0.27877140233863795,
      "learning_rate": 3.808666090050824e-06,
      "loss": 0.0203,
      "step": 6924
    },
    {
      "epoch": 1.857066237597211,
      "grad_norm": 0.23736863197131516,
      "learning_rate": 3.807150894807411e-06,
      "loss": 0.0241,
      "step": 6925
    },
    {
      "epoch": 1.8573344060069723,
      "grad_norm": 0.25013429411361215,
      "learning_rate": 3.805635815708548e-06,
      "loss": 0.0305,
      "step": 6926
    },
    {
      "epoch": 1.8576025744167337,
      "grad_norm": 0.26634989133974035,
      "learning_rate": 3.804120852901756e-06,
      "loss": 0.0237,
      "step": 6927
    },
    {
      "epoch": 1.8578707428264951,
      "grad_norm": 0.28896852925463723,
      "learning_rate": 3.8026060065345407e-06,
      "loss": 0.0189,
      "step": 6928
    },
    {
      "epoch": 1.8581389112362565,
      "grad_norm": 0.2528866207024745,
      "learning_rate": 3.8010912767543993e-06,
      "loss": 0.0241,
      "step": 6929
    },
    {
      "epoch": 1.8584070796460177,
      "grad_norm": 0.24553085521626805,
      "learning_rate": 3.799576663708818e-06,
      "loss": 0.0206,
      "step": 6930
    },
    {
      "epoch": 1.8586752480557789,
      "grad_norm": 0.4310048778041812,
      "learning_rate": 3.79806216754527e-06,
      "loss": 0.0257,
      "step": 6931
    },
    {
      "epoch": 1.8589434164655403,
      "grad_norm": 0.2862437213992107,
      "learning_rate": 3.7965477884112156e-06,
      "loss": 0.0207,
      "step": 6932
    },
    {
      "epoch": 1.8592115848753017,
      "grad_norm": 0.23133394353864725,
      "learning_rate": 3.79503352645411e-06,
      "loss": 0.0264,
      "step": 6933
    },
    {
      "epoch": 1.859479753285063,
      "grad_norm": 0.20907174962762276,
      "learning_rate": 3.793519381821392e-06,
      "loss": 0.0202,
      "step": 6934
    },
    {
      "epoch": 1.8597479216948245,
      "grad_norm": 0.23161348247079522,
      "learning_rate": 3.7920053546604875e-06,
      "loss": 0.0223,
      "step": 6935
    },
    {
      "epoch": 1.8600160901045857,
      "grad_norm": 0.19155366008762223,
      "learning_rate": 3.790491445118812e-06,
      "loss": 0.0179,
      "step": 6936
    },
    {
      "epoch": 1.8602842585143469,
      "grad_norm": 0.4005368812297641,
      "learning_rate": 3.788977653343776e-06,
      "loss": 0.0251,
      "step": 6937
    },
    {
      "epoch": 1.8605524269241083,
      "grad_norm": 0.25845495421528586,
      "learning_rate": 3.78746397948277e-06,
      "loss": 0.0193,
      "step": 6938
    },
    {
      "epoch": 1.8608205953338697,
      "grad_norm": 0.44631407335768314,
      "learning_rate": 3.7859504236831766e-06,
      "loss": 0.0179,
      "step": 6939
    },
    {
      "epoch": 1.861088763743631,
      "grad_norm": 0.27091810661515703,
      "learning_rate": 3.784436986092368e-06,
      "loss": 0.0165,
      "step": 6940
    },
    {
      "epoch": 1.8613569321533925,
      "grad_norm": 0.2500962624375777,
      "learning_rate": 3.7829236668577025e-06,
      "loss": 0.0283,
      "step": 6941
    },
    {
      "epoch": 1.8616251005631537,
      "grad_norm": 0.24769954266746042,
      "learning_rate": 3.7814104661265272e-06,
      "loss": 0.0194,
      "step": 6942
    },
    {
      "epoch": 1.8618932689729148,
      "grad_norm": 0.2648395917941033,
      "learning_rate": 3.7798973840461805e-06,
      "loss": 0.0246,
      "step": 6943
    },
    {
      "epoch": 1.8621614373826763,
      "grad_norm": 0.25525958801830617,
      "learning_rate": 3.7783844207639864e-06,
      "loss": 0.0256,
      "step": 6944
    },
    {
      "epoch": 1.8624296057924377,
      "grad_norm": 0.2447714175444359,
      "learning_rate": 3.776871576427258e-06,
      "loss": 0.0189,
      "step": 6945
    },
    {
      "epoch": 1.862697774202199,
      "grad_norm": 0.527679564382456,
      "learning_rate": 3.7753588511832973e-06,
      "loss": 0.0264,
      "step": 6946
    },
    {
      "epoch": 1.8629659426119605,
      "grad_norm": 0.19437480485350803,
      "learning_rate": 3.7738462451793932e-06,
      "loss": 0.0222,
      "step": 6947
    },
    {
      "epoch": 1.8632341110217217,
      "grad_norm": 0.24861708777404487,
      "learning_rate": 3.772333758562825e-06,
      "loss": 0.0224,
      "step": 6948
    },
    {
      "epoch": 1.8635022794314828,
      "grad_norm": 0.20935678403845923,
      "learning_rate": 3.7708213914808595e-06,
      "loss": 0.0155,
      "step": 6949
    },
    {
      "epoch": 1.8637704478412442,
      "grad_norm": 0.20856999955869213,
      "learning_rate": 3.7693091440807517e-06,
      "loss": 0.0212,
      "step": 6950
    },
    {
      "epoch": 1.8640386162510056,
      "grad_norm": 0.23998810518278219,
      "learning_rate": 3.7677970165097444e-06,
      "loss": 0.0199,
      "step": 6951
    },
    {
      "epoch": 1.864306784660767,
      "grad_norm": 0.23121167793853545,
      "learning_rate": 3.7662850089150693e-06,
      "loss": 0.0281,
      "step": 6952
    },
    {
      "epoch": 1.8645749530705285,
      "grad_norm": 0.32107159270748564,
      "learning_rate": 3.764773121443949e-06,
      "loss": 0.021,
      "step": 6953
    },
    {
      "epoch": 1.8648431214802896,
      "grad_norm": 0.16394119503769808,
      "learning_rate": 3.7632613542435904e-06,
      "loss": 0.0171,
      "step": 6954
    },
    {
      "epoch": 1.8651112898900508,
      "grad_norm": 0.22308678785144828,
      "learning_rate": 3.76174970746119e-06,
      "loss": 0.0166,
      "step": 6955
    },
    {
      "epoch": 1.8653794582998122,
      "grad_norm": 0.2607107652943991,
      "learning_rate": 3.760238181243933e-06,
      "loss": 0.0184,
      "step": 6956
    },
    {
      "epoch": 1.8656476267095736,
      "grad_norm": 0.2805988845489507,
      "learning_rate": 3.758726775738993e-06,
      "loss": 0.0258,
      "step": 6957
    },
    {
      "epoch": 1.865915795119335,
      "grad_norm": 0.2794543660113636,
      "learning_rate": 3.757215491093532e-06,
      "loss": 0.0224,
      "step": 6958
    },
    {
      "epoch": 1.8661839635290964,
      "grad_norm": 0.1937611782214648,
      "learning_rate": 3.7557043274546983e-06,
      "loss": 0.0167,
      "step": 6959
    },
    {
      "epoch": 1.8664521319388576,
      "grad_norm": 0.23307789623691946,
      "learning_rate": 3.754193284969631e-06,
      "loss": 0.0183,
      "step": 6960
    },
    {
      "epoch": 1.8667203003486188,
      "grad_norm": 0.18633256546762245,
      "learning_rate": 3.752682363785456e-06,
      "loss": 0.0249,
      "step": 6961
    },
    {
      "epoch": 1.8669884687583802,
      "grad_norm": 0.21827672780996532,
      "learning_rate": 3.7511715640492867e-06,
      "loss": 0.018,
      "step": 6962
    },
    {
      "epoch": 1.8672566371681416,
      "grad_norm": 0.197992716588563,
      "learning_rate": 3.749660885908226e-06,
      "loss": 0.0208,
      "step": 6963
    },
    {
      "epoch": 1.867524805577903,
      "grad_norm": 0.35439091852439303,
      "learning_rate": 3.7481503295093667e-06,
      "loss": 0.0342,
      "step": 6964
    },
    {
      "epoch": 1.8677929739876644,
      "grad_norm": 0.218978327780633,
      "learning_rate": 3.7466398949997864e-06,
      "loss": 0.0236,
      "step": 6965
    },
    {
      "epoch": 1.8680611423974256,
      "grad_norm": 0.34114200973506364,
      "learning_rate": 3.745129582526552e-06,
      "loss": 0.0192,
      "step": 6966
    },
    {
      "epoch": 1.8683293108071868,
      "grad_norm": 0.23455429643048675,
      "learning_rate": 3.7436193922367183e-06,
      "loss": 0.022,
      "step": 6967
    },
    {
      "epoch": 1.8685974792169482,
      "grad_norm": 0.19155953919726126,
      "learning_rate": 3.7421093242773284e-06,
      "loss": 0.0108,
      "step": 6968
    },
    {
      "epoch": 1.8688656476267096,
      "grad_norm": 0.22619788885285666,
      "learning_rate": 3.740599378795413e-06,
      "loss": 0.0169,
      "step": 6969
    },
    {
      "epoch": 1.869133816036471,
      "grad_norm": 0.1958097300640288,
      "learning_rate": 3.7390895559379956e-06,
      "loss": 0.019,
      "step": 6970
    },
    {
      "epoch": 1.8694019844462322,
      "grad_norm": 0.2888452949974741,
      "learning_rate": 3.737579855852078e-06,
      "loss": 0.0231,
      "step": 6971
    },
    {
      "epoch": 1.8696701528559936,
      "grad_norm": 0.2187344420831675,
      "learning_rate": 3.7360702786846594e-06,
      "loss": 0.0182,
      "step": 6972
    },
    {
      "epoch": 1.8699383212657548,
      "grad_norm": 0.19178801091015069,
      "learning_rate": 3.73456082458272e-06,
      "loss": 0.0174,
      "step": 6973
    },
    {
      "epoch": 1.8702064896755162,
      "grad_norm": 0.2697875391590228,
      "learning_rate": 3.733051493693235e-06,
      "loss": 0.0271,
      "step": 6974
    },
    {
      "epoch": 1.8704746580852776,
      "grad_norm": 0.16702355312191433,
      "learning_rate": 3.7315422861631623e-06,
      "loss": 0.0125,
      "step": 6975
    },
    {
      "epoch": 1.870742826495039,
      "grad_norm": 0.2753655706747481,
      "learning_rate": 3.7300332021394494e-06,
      "loss": 0.0284,
      "step": 6976
    },
    {
      "epoch": 1.8710109949048002,
      "grad_norm": 0.29307470070896974,
      "learning_rate": 3.7285242417690315e-06,
      "loss": 0.0239,
      "step": 6977
    },
    {
      "epoch": 1.8712791633145616,
      "grad_norm": 0.6665346557541083,
      "learning_rate": 3.727015405198833e-06,
      "loss": 0.0315,
      "step": 6978
    },
    {
      "epoch": 1.8715473317243227,
      "grad_norm": 0.28816724856558723,
      "learning_rate": 3.725506692575764e-06,
      "loss": 0.0222,
      "step": 6979
    },
    {
      "epoch": 1.8718155001340842,
      "grad_norm": 0.2638693850188307,
      "learning_rate": 3.723998104046725e-06,
      "loss": 0.0266,
      "step": 6980
    },
    {
      "epoch": 1.8720836685438456,
      "grad_norm": 0.18862376917066837,
      "learning_rate": 3.722489639758603e-06,
      "loss": 0.0147,
      "step": 6981
    },
    {
      "epoch": 1.872351836953607,
      "grad_norm": 0.24689150124554218,
      "learning_rate": 3.7209812998582753e-06,
      "loss": 0.0172,
      "step": 6982
    },
    {
      "epoch": 1.8726200053633681,
      "grad_norm": 0.292614102732543,
      "learning_rate": 3.719473084492601e-06,
      "loss": 0.0243,
      "step": 6983
    },
    {
      "epoch": 1.8728881737731295,
      "grad_norm": 0.20445945568691337,
      "learning_rate": 3.7179649938084305e-06,
      "loss": 0.0142,
      "step": 6984
    },
    {
      "epoch": 1.8731563421828907,
      "grad_norm": 0.27538521797114024,
      "learning_rate": 3.7164570279526067e-06,
      "loss": 0.025,
      "step": 6985
    },
    {
      "epoch": 1.8734245105926521,
      "grad_norm": 0.21288748774711,
      "learning_rate": 3.7149491870719546e-06,
      "loss": 0.0201,
      "step": 6986
    },
    {
      "epoch": 1.8736926790024135,
      "grad_norm": 0.3490647517692959,
      "learning_rate": 3.7134414713132883e-06,
      "loss": 0.0195,
      "step": 6987
    },
    {
      "epoch": 1.873960847412175,
      "grad_norm": 0.2699833440703194,
      "learning_rate": 3.71193388082341e-06,
      "loss": 0.0271,
      "step": 6988
    },
    {
      "epoch": 1.8742290158219361,
      "grad_norm": 0.21326003424104192,
      "learning_rate": 3.71042641574911e-06,
      "loss": 0.0197,
      "step": 6989
    },
    {
      "epoch": 1.8744971842316975,
      "grad_norm": 0.22691999511398894,
      "learning_rate": 3.7089190762371653e-06,
      "loss": 0.0153,
      "step": 6990
    },
    {
      "epoch": 1.8747653526414587,
      "grad_norm": 0.1958500585735867,
      "learning_rate": 3.707411862434343e-06,
      "loss": 0.0164,
      "step": 6991
    },
    {
      "epoch": 1.8750335210512201,
      "grad_norm": 0.21554136204973157,
      "learning_rate": 3.705904774487396e-06,
      "loss": 0.0186,
      "step": 6992
    },
    {
      "epoch": 1.8753016894609815,
      "grad_norm": 0.2693994121691686,
      "learning_rate": 3.7043978125430656e-06,
      "loss": 0.021,
      "step": 6993
    },
    {
      "epoch": 1.875569857870743,
      "grad_norm": 0.3652888290555059,
      "learning_rate": 3.7028909767480805e-06,
      "loss": 0.0295,
      "step": 6994
    },
    {
      "epoch": 1.8758380262805041,
      "grad_norm": 0.2914650508213058,
      "learning_rate": 3.7013842672491577e-06,
      "loss": 0.0342,
      "step": 6995
    },
    {
      "epoch": 1.8761061946902655,
      "grad_norm": 0.21938719293537903,
      "learning_rate": 3.6998776841930006e-06,
      "loss": 0.0215,
      "step": 6996
    },
    {
      "epoch": 1.8763743631000267,
      "grad_norm": 0.4920219307855389,
      "learning_rate": 3.698371227726302e-06,
      "loss": 0.0265,
      "step": 6997
    },
    {
      "epoch": 1.876642531509788,
      "grad_norm": 0.3262475262146318,
      "learning_rate": 3.69686489799574e-06,
      "loss": 0.0206,
      "step": 6998
    },
    {
      "epoch": 1.8769106999195495,
      "grad_norm": 0.27683135678380505,
      "learning_rate": 3.6953586951479834e-06,
      "loss": 0.0178,
      "step": 6999
    },
    {
      "epoch": 1.877178868329311,
      "grad_norm": 0.21968624320647417,
      "learning_rate": 3.6938526193296873e-06,
      "loss": 0.0185,
      "step": 7000
    },
    {
      "epoch": 1.877447036739072,
      "grad_norm": 0.4556426087693223,
      "learning_rate": 3.6923466706874933e-06,
      "loss": 0.0159,
      "step": 7001
    },
    {
      "epoch": 1.8777152051488335,
      "grad_norm": 0.29466512473219314,
      "learning_rate": 3.6908408493680326e-06,
      "loss": 0.0295,
      "step": 7002
    },
    {
      "epoch": 1.8779833735585947,
      "grad_norm": 0.7259532545691287,
      "learning_rate": 3.689335155517923e-06,
      "loss": 0.0219,
      "step": 7003
    },
    {
      "epoch": 1.878251541968356,
      "grad_norm": 0.24147129949229526,
      "learning_rate": 3.687829589283769e-06,
      "loss": 0.0335,
      "step": 7004
    },
    {
      "epoch": 1.8785197103781175,
      "grad_norm": 0.31874545322073783,
      "learning_rate": 3.686324150812165e-06,
      "loss": 0.0441,
      "step": 7005
    },
    {
      "epoch": 1.878787878787879,
      "grad_norm": 0.23483730295668898,
      "learning_rate": 3.6848188402496897e-06,
      "loss": 0.0244,
      "step": 7006
    },
    {
      "epoch": 1.87905604719764,
      "grad_norm": 0.2111676055028106,
      "learning_rate": 3.6833136577429142e-06,
      "loss": 0.0144,
      "step": 7007
    },
    {
      "epoch": 1.8793242156074015,
      "grad_norm": 0.26821349236064757,
      "learning_rate": 3.6818086034383904e-06,
      "loss": 0.0196,
      "step": 7008
    },
    {
      "epoch": 1.8795923840171627,
      "grad_norm": 0.3146965854347319,
      "learning_rate": 3.6803036774826633e-06,
      "loss": 0.0236,
      "step": 7009
    },
    {
      "epoch": 1.879860552426924,
      "grad_norm": 0.33443018022702936,
      "learning_rate": 3.678798880022264e-06,
      "loss": 0.0437,
      "step": 7010
    },
    {
      "epoch": 1.8801287208366855,
      "grad_norm": 0.452089724738561,
      "learning_rate": 3.677294211203708e-06,
      "loss": 0.026,
      "step": 7011
    },
    {
      "epoch": 1.8803968892464469,
      "grad_norm": 0.48043155991685377,
      "learning_rate": 3.675789671173504e-06,
      "loss": 0.0253,
      "step": 7012
    },
    {
      "epoch": 1.880665057656208,
      "grad_norm": 0.17162679145973483,
      "learning_rate": 3.674285260078145e-06,
      "loss": 0.0147,
      "step": 7013
    },
    {
      "epoch": 1.8809332260659695,
      "grad_norm": 0.20991801599544427,
      "learning_rate": 3.67278097806411e-06,
      "loss": 0.0198,
      "step": 7014
    },
    {
      "epoch": 1.8812013944757306,
      "grad_norm": 0.24309580096810576,
      "learning_rate": 3.6712768252778675e-06,
      "loss": 0.0177,
      "step": 7015
    },
    {
      "epoch": 1.881469562885492,
      "grad_norm": 0.2082945742775015,
      "learning_rate": 3.669772801865873e-06,
      "loss": 0.018,
      "step": 7016
    },
    {
      "epoch": 1.8817377312952535,
      "grad_norm": 0.2499281183632208,
      "learning_rate": 3.668268907974567e-06,
      "loss": 0.0201,
      "step": 7017
    },
    {
      "epoch": 1.8820058997050149,
      "grad_norm": 0.22758519362615992,
      "learning_rate": 3.6667651437503836e-06,
      "loss": 0.0234,
      "step": 7018
    },
    {
      "epoch": 1.882274068114776,
      "grad_norm": 0.2714685689520811,
      "learning_rate": 3.6652615093397414e-06,
      "loss": 0.0242,
      "step": 7019
    },
    {
      "epoch": 1.8825422365245374,
      "grad_norm": 0.21201332235644643,
      "learning_rate": 3.663758004889039e-06,
      "loss": 0.0142,
      "step": 7020
    },
    {
      "epoch": 1.8828104049342986,
      "grad_norm": 0.370982055867651,
      "learning_rate": 3.6622546305446714e-06,
      "loss": 0.0387,
      "step": 7021
    },
    {
      "epoch": 1.88307857334406,
      "grad_norm": 0.32092564669169865,
      "learning_rate": 3.6607513864530197e-06,
      "loss": 0.0189,
      "step": 7022
    },
    {
      "epoch": 1.8833467417538214,
      "grad_norm": 0.2446124574540049,
      "learning_rate": 3.6592482727604508e-06,
      "loss": 0.0238,
      "step": 7023
    },
    {
      "epoch": 1.8836149101635828,
      "grad_norm": 0.2457979077141817,
      "learning_rate": 3.657745289613317e-06,
      "loss": 0.0197,
      "step": 7024
    },
    {
      "epoch": 1.883883078573344,
      "grad_norm": 0.2129551794920912,
      "learning_rate": 3.656242437157961e-06,
      "loss": 0.0143,
      "step": 7025
    },
    {
      "epoch": 1.8841512469831054,
      "grad_norm": 0.20616959603395404,
      "learning_rate": 3.6547397155407104e-06,
      "loss": 0.0179,
      "step": 7026
    },
    {
      "epoch": 1.8844194153928666,
      "grad_norm": 0.3634533808171303,
      "learning_rate": 3.6532371249078834e-06,
      "loss": 0.0198,
      "step": 7027
    },
    {
      "epoch": 1.884687583802628,
      "grad_norm": 0.25091978318769564,
      "learning_rate": 3.6517346654057804e-06,
      "loss": 0.0191,
      "step": 7028
    },
    {
      "epoch": 1.8849557522123894,
      "grad_norm": 0.26318757034769874,
      "learning_rate": 3.6502323371806947e-06,
      "loss": 0.0174,
      "step": 7029
    },
    {
      "epoch": 1.8852239206221508,
      "grad_norm": 0.19227605880187817,
      "learning_rate": 3.6487301403789024e-06,
      "loss": 0.017,
      "step": 7030
    },
    {
      "epoch": 1.885492089031912,
      "grad_norm": 0.25021704873425865,
      "learning_rate": 3.647228075146671e-06,
      "loss": 0.0204,
      "step": 7031
    },
    {
      "epoch": 1.8857602574416734,
      "grad_norm": 0.2698683286019005,
      "learning_rate": 3.645726141630247e-06,
      "loss": 0.031,
      "step": 7032
    },
    {
      "epoch": 1.8860284258514346,
      "grad_norm": 0.2286795448611874,
      "learning_rate": 3.6442243399758748e-06,
      "loss": 0.0264,
      "step": 7033
    },
    {
      "epoch": 1.886296594261196,
      "grad_norm": 0.2954885386849269,
      "learning_rate": 3.6427226703297785e-06,
      "loss": 0.0328,
      "step": 7034
    },
    {
      "epoch": 1.8865647626709574,
      "grad_norm": 0.34084282901186963,
      "learning_rate": 3.641221132838173e-06,
      "loss": 0.0244,
      "step": 7035
    },
    {
      "epoch": 1.8868329310807188,
      "grad_norm": 0.33870234616780215,
      "learning_rate": 3.6397197276472574e-06,
      "loss": 0.0276,
      "step": 7036
    },
    {
      "epoch": 1.88710109949048,
      "grad_norm": 0.2178317180176426,
      "learning_rate": 3.638218454903221e-06,
      "loss": 0.0191,
      "step": 7037
    },
    {
      "epoch": 1.8873692679002414,
      "grad_norm": 0.2569043496159182,
      "learning_rate": 3.636717314752237e-06,
      "loss": 0.0293,
      "step": 7038
    },
    {
      "epoch": 1.8876374363100026,
      "grad_norm": 0.2465427753762059,
      "learning_rate": 3.6352163073404695e-06,
      "loss": 0.0178,
      "step": 7039
    },
    {
      "epoch": 1.887905604719764,
      "grad_norm": 0.3143826234022342,
      "learning_rate": 3.6337154328140665e-06,
      "loss": 0.0251,
      "step": 7040
    },
    {
      "epoch": 1.8881737731295254,
      "grad_norm": 0.2935050413340353,
      "learning_rate": 3.632214691319165e-06,
      "loss": 0.036,
      "step": 7041
    },
    {
      "epoch": 1.8884419415392868,
      "grad_norm": 0.23619078363731105,
      "learning_rate": 3.6307140830018874e-06,
      "loss": 0.0186,
      "step": 7042
    },
    {
      "epoch": 1.888710109949048,
      "grad_norm": 0.18216337134079702,
      "learning_rate": 3.629213608008345e-06,
      "loss": 0.0143,
      "step": 7043
    },
    {
      "epoch": 1.8889782783588094,
      "grad_norm": 0.2933549512345838,
      "learning_rate": 3.6277132664846338e-06,
      "loss": 0.0215,
      "step": 7044
    },
    {
      "epoch": 1.8892464467685706,
      "grad_norm": 0.32323796260027526,
      "learning_rate": 3.6262130585768377e-06,
      "loss": 0.024,
      "step": 7045
    },
    {
      "epoch": 1.889514615178332,
      "grad_norm": 0.3519092971036347,
      "learning_rate": 3.624712984431029e-06,
      "loss": 0.0277,
      "step": 7046
    },
    {
      "epoch": 1.8897827835880934,
      "grad_norm": 0.22728157075217575,
      "learning_rate": 3.623213044193266e-06,
      "loss": 0.0158,
      "step": 7047
    },
    {
      "epoch": 1.8900509519978548,
      "grad_norm": 0.23834144482277606,
      "learning_rate": 3.621713238009593e-06,
      "loss": 0.0181,
      "step": 7048
    },
    {
      "epoch": 1.890319120407616,
      "grad_norm": 0.27202791484022565,
      "learning_rate": 3.6202135660260424e-06,
      "loss": 0.0242,
      "step": 7049
    },
    {
      "epoch": 1.8905872888173774,
      "grad_norm": 0.28648044472606465,
      "learning_rate": 3.6187140283886346e-06,
      "loss": 0.0272,
      "step": 7050
    },
    {
      "epoch": 1.8908554572271385,
      "grad_norm": 0.21874665828440795,
      "learning_rate": 3.6172146252433736e-06,
      "loss": 0.0184,
      "step": 7051
    },
    {
      "epoch": 1.8911236256369,
      "grad_norm": 0.24518167430022472,
      "learning_rate": 3.6157153567362548e-06,
      "loss": 0.025,
      "step": 7052
    },
    {
      "epoch": 1.8913917940466614,
      "grad_norm": 0.22912996093580784,
      "learning_rate": 3.6142162230132554e-06,
      "loss": 0.024,
      "step": 7053
    },
    {
      "epoch": 1.8916599624564228,
      "grad_norm": 0.3031929104786631,
      "learning_rate": 3.6127172242203433e-06,
      "loss": 0.0154,
      "step": 7054
    },
    {
      "epoch": 1.891928130866184,
      "grad_norm": 0.24333704048729993,
      "learning_rate": 3.6112183605034734e-06,
      "loss": 0.0181,
      "step": 7055
    },
    {
      "epoch": 1.8921962992759453,
      "grad_norm": 0.26608979603609817,
      "learning_rate": 3.609719632008584e-06,
      "loss": 0.0177,
      "step": 7056
    },
    {
      "epoch": 1.8924644676857065,
      "grad_norm": 0.2819566043907426,
      "learning_rate": 3.608221038881602e-06,
      "loss": 0.0265,
      "step": 7057
    },
    {
      "epoch": 1.892732636095468,
      "grad_norm": 0.3278746292151866,
      "learning_rate": 3.6067225812684427e-06,
      "loss": 0.0224,
      "step": 7058
    },
    {
      "epoch": 1.8930008045052293,
      "grad_norm": 0.25462131932433885,
      "learning_rate": 3.605224259315005e-06,
      "loss": 0.0279,
      "step": 7059
    },
    {
      "epoch": 1.8932689729149907,
      "grad_norm": 0.24428452901546202,
      "learning_rate": 3.6037260731671797e-06,
      "loss": 0.0216,
      "step": 7060
    },
    {
      "epoch": 1.893537141324752,
      "grad_norm": 0.2336104719904052,
      "learning_rate": 3.6022280229708393e-06,
      "loss": 0.0187,
      "step": 7061
    },
    {
      "epoch": 1.893805309734513,
      "grad_norm": 0.25492555326375443,
      "learning_rate": 3.600730108871846e-06,
      "loss": 0.0252,
      "step": 7062
    },
    {
      "epoch": 1.8940734781442745,
      "grad_norm": 0.20146735109100866,
      "learning_rate": 3.599232331016046e-06,
      "loss": 0.0134,
      "step": 7063
    },
    {
      "epoch": 1.894341646554036,
      "grad_norm": 0.19210063985976,
      "learning_rate": 3.597734689549276e-06,
      "loss": 0.0153,
      "step": 7064
    },
    {
      "epoch": 1.8946098149637973,
      "grad_norm": 0.35925790424416315,
      "learning_rate": 3.5962371846173542e-06,
      "loss": 0.0238,
      "step": 7065
    },
    {
      "epoch": 1.8948779833735587,
      "grad_norm": 0.25860596809124786,
      "learning_rate": 3.594739816366093e-06,
      "loss": 0.0214,
      "step": 7066
    },
    {
      "epoch": 1.89514615178332,
      "grad_norm": 0.18475219782595212,
      "learning_rate": 3.593242584941286e-06,
      "loss": 0.0145,
      "step": 7067
    },
    {
      "epoch": 1.895414320193081,
      "grad_norm": 0.19765040423060004,
      "learning_rate": 3.591745490488712e-06,
      "loss": 0.013,
      "step": 7068
    },
    {
      "epoch": 1.8956824886028425,
      "grad_norm": 0.2685459568541738,
      "learning_rate": 3.59024853315414e-06,
      "loss": 0.0266,
      "step": 7069
    },
    {
      "epoch": 1.895950657012604,
      "grad_norm": 0.19851357402496145,
      "learning_rate": 3.588751713083326e-06,
      "loss": 0.0174,
      "step": 7070
    },
    {
      "epoch": 1.8962188254223653,
      "grad_norm": 0.3022403495807125,
      "learning_rate": 3.587255030422011e-06,
      "loss": 0.0292,
      "step": 7071
    },
    {
      "epoch": 1.8964869938321267,
      "grad_norm": 0.19896067677700732,
      "learning_rate": 3.5857584853159235e-06,
      "loss": 0.0182,
      "step": 7072
    },
    {
      "epoch": 1.896755162241888,
      "grad_norm": 0.3307489908240372,
      "learning_rate": 3.5842620779107774e-06,
      "loss": 0.0203,
      "step": 7073
    },
    {
      "epoch": 1.897023330651649,
      "grad_norm": 0.2472748316509056,
      "learning_rate": 3.5827658083522737e-06,
      "loss": 0.0243,
      "step": 7074
    },
    {
      "epoch": 1.8972914990614105,
      "grad_norm": 0.19046092220160468,
      "learning_rate": 3.5812696767861012e-06,
      "loss": 0.0148,
      "step": 7075
    },
    {
      "epoch": 1.8975596674711719,
      "grad_norm": 0.24328357820071297,
      "learning_rate": 3.5797736833579315e-06,
      "loss": 0.0289,
      "step": 7076
    },
    {
      "epoch": 1.8978278358809333,
      "grad_norm": 0.2859728910214367,
      "learning_rate": 3.578277828213429e-06,
      "loss": 0.0272,
      "step": 7077
    },
    {
      "epoch": 1.8980960042906947,
      "grad_norm": 0.24579901952801445,
      "learning_rate": 3.5767821114982402e-06,
      "loss": 0.0224,
      "step": 7078
    },
    {
      "epoch": 1.8983641727004559,
      "grad_norm": 0.36692602768068056,
      "learning_rate": 3.5752865333580007e-06,
      "loss": 0.0219,
      "step": 7079
    },
    {
      "epoch": 1.898632341110217,
      "grad_norm": 0.18376942084347292,
      "learning_rate": 3.573791093938326e-06,
      "loss": 0.0148,
      "step": 7080
    },
    {
      "epoch": 1.8989005095199785,
      "grad_norm": 0.20212165538605545,
      "learning_rate": 3.572295793384826e-06,
      "loss": 0.0218,
      "step": 7081
    },
    {
      "epoch": 1.8991686779297399,
      "grad_norm": 0.22785970856467372,
      "learning_rate": 3.5708006318430947e-06,
      "loss": 0.0182,
      "step": 7082
    },
    {
      "epoch": 1.8994368463395013,
      "grad_norm": 0.3583667876106721,
      "learning_rate": 3.569305609458712e-06,
      "loss": 0.0271,
      "step": 7083
    },
    {
      "epoch": 1.8997050147492627,
      "grad_norm": 0.22805695506938622,
      "learning_rate": 3.5678107263772428e-06,
      "loss": 0.0193,
      "step": 7084
    },
    {
      "epoch": 1.8999731831590239,
      "grad_norm": 0.20571963827933853,
      "learning_rate": 3.566315982744241e-06,
      "loss": 0.0194,
      "step": 7085
    },
    {
      "epoch": 1.900241351568785,
      "grad_norm": 0.27578171672999596,
      "learning_rate": 3.5648213787052443e-06,
      "loss": 0.0212,
      "step": 7086
    },
    {
      "epoch": 1.9005095199785464,
      "grad_norm": 0.2469544742118166,
      "learning_rate": 3.5633269144057816e-06,
      "loss": 0.0213,
      "step": 7087
    },
    {
      "epoch": 1.9007776883883079,
      "grad_norm": 0.23025679927357307,
      "learning_rate": 3.5618325899913617e-06,
      "loss": 0.0214,
      "step": 7088
    },
    {
      "epoch": 1.9010458567980693,
      "grad_norm": 0.24061436892342547,
      "learning_rate": 3.560338405607485e-06,
      "loss": 0.0233,
      "step": 7089
    },
    {
      "epoch": 1.9013140252078307,
      "grad_norm": 0.24674169746577665,
      "learning_rate": 3.558844361399636e-06,
      "loss": 0.0211,
      "step": 7090
    },
    {
      "epoch": 1.9015821936175918,
      "grad_norm": 0.2478521225693802,
      "learning_rate": 3.5573504575132856e-06,
      "loss": 0.0223,
      "step": 7091
    },
    {
      "epoch": 1.901850362027353,
      "grad_norm": 0.2501985639402288,
      "learning_rate": 3.5558566940938906e-06,
      "loss": 0.019,
      "step": 7092
    },
    {
      "epoch": 1.9021185304371144,
      "grad_norm": 0.23993871837771222,
      "learning_rate": 3.554363071286895e-06,
      "loss": 0.025,
      "step": 7093
    },
    {
      "epoch": 1.9023866988468758,
      "grad_norm": 0.3510888308825505,
      "learning_rate": 3.5528695892377287e-06,
      "loss": 0.0269,
      "step": 7094
    },
    {
      "epoch": 1.9026548672566372,
      "grad_norm": 0.31294959490497587,
      "learning_rate": 3.5513762480918084e-06,
      "loss": 0.0175,
      "step": 7095
    },
    {
      "epoch": 1.9029230356663986,
      "grad_norm": 0.2816340358035628,
      "learning_rate": 3.5498830479945367e-06,
      "loss": 0.0239,
      "step": 7096
    },
    {
      "epoch": 1.9031912040761598,
      "grad_norm": 0.17500505971658042,
      "learning_rate": 3.548389989091302e-06,
      "loss": 0.0166,
      "step": 7097
    },
    {
      "epoch": 1.903459372485921,
      "grad_norm": 0.29381925972440326,
      "learning_rate": 3.546897071527481e-06,
      "loss": 0.0223,
      "step": 7098
    },
    {
      "epoch": 1.9037275408956824,
      "grad_norm": 0.33782898747479334,
      "learning_rate": 3.5454042954484345e-06,
      "loss": 0.0202,
      "step": 7099
    },
    {
      "epoch": 1.9039957093054438,
      "grad_norm": 0.25456411633782733,
      "learning_rate": 3.5439116609995095e-06,
      "loss": 0.0244,
      "step": 7100
    },
    {
      "epoch": 1.9042638777152052,
      "grad_norm": 0.29473023530956094,
      "learning_rate": 3.54241916832604e-06,
      "loss": 0.0223,
      "step": 7101
    },
    {
      "epoch": 1.9045320461249666,
      "grad_norm": 0.27121348732917167,
      "learning_rate": 3.540926817573347e-06,
      "loss": 0.0314,
      "step": 7102
    },
    {
      "epoch": 1.9048002145347278,
      "grad_norm": 0.29320278577891506,
      "learning_rate": 3.539434608886737e-06,
      "loss": 0.0236,
      "step": 7103
    },
    {
      "epoch": 1.905068382944489,
      "grad_norm": 0.25023252016743847,
      "learning_rate": 3.5379425424115e-06,
      "loss": 0.0163,
      "step": 7104
    },
    {
      "epoch": 1.9053365513542504,
      "grad_norm": 0.23036393735880645,
      "learning_rate": 3.536450618292916e-06,
      "loss": 0.0204,
      "step": 7105
    },
    {
      "epoch": 1.9056047197640118,
      "grad_norm": 0.33599322456956554,
      "learning_rate": 3.5349588366762493e-06,
      "loss": 0.0275,
      "step": 7106
    },
    {
      "epoch": 1.9058728881737732,
      "grad_norm": 0.2749105394421017,
      "learning_rate": 3.5334671977067504e-06,
      "loss": 0.0206,
      "step": 7107
    },
    {
      "epoch": 1.9061410565835346,
      "grad_norm": 0.2873911744369586,
      "learning_rate": 3.531975701529658e-06,
      "loss": 0.019,
      "step": 7108
    },
    {
      "epoch": 1.9064092249932958,
      "grad_norm": 0.3030079154593814,
      "learning_rate": 3.5304843482901942e-06,
      "loss": 0.0234,
      "step": 7109
    },
    {
      "epoch": 1.906677393403057,
      "grad_norm": 0.24472037067753993,
      "learning_rate": 3.5289931381335672e-06,
      "loss": 0.0209,
      "step": 7110
    },
    {
      "epoch": 1.9069455618128184,
      "grad_norm": 0.32036583600200136,
      "learning_rate": 3.5275020712049736e-06,
      "loss": 0.0259,
      "step": 7111
    },
    {
      "epoch": 1.9072137302225798,
      "grad_norm": 0.2928462312796386,
      "learning_rate": 3.5260111476495935e-06,
      "loss": 0.0218,
      "step": 7112
    },
    {
      "epoch": 1.9074818986323412,
      "grad_norm": 0.21290740136466552,
      "learning_rate": 3.524520367612595e-06,
      "loss": 0.0163,
      "step": 7113
    },
    {
      "epoch": 1.9077500670421026,
      "grad_norm": 0.28203123533284463,
      "learning_rate": 3.523029731239129e-06,
      "loss": 0.0175,
      "step": 7114
    },
    {
      "epoch": 1.9080182354518638,
      "grad_norm": 0.3867072447200513,
      "learning_rate": 3.5215392386743385e-06,
      "loss": 0.0224,
      "step": 7115
    },
    {
      "epoch": 1.908286403861625,
      "grad_norm": 0.2224486449250359,
      "learning_rate": 3.52004889006335e-06,
      "loss": 0.0205,
      "step": 7116
    },
    {
      "epoch": 1.9085545722713864,
      "grad_norm": 0.22106275986429766,
      "learning_rate": 3.5185586855512677e-06,
      "loss": 0.0174,
      "step": 7117
    },
    {
      "epoch": 1.9088227406811478,
      "grad_norm": 0.25210774165645616,
      "learning_rate": 3.5170686252831942e-06,
      "loss": 0.0157,
      "step": 7118
    },
    {
      "epoch": 1.9090909090909092,
      "grad_norm": 0.3298744388389556,
      "learning_rate": 3.5155787094042113e-06,
      "loss": 0.0285,
      "step": 7119
    },
    {
      "epoch": 1.9093590775006706,
      "grad_norm": 0.2592173864842358,
      "learning_rate": 3.514088938059389e-06,
      "loss": 0.0246,
      "step": 7120
    },
    {
      "epoch": 1.9096272459104318,
      "grad_norm": 0.23042473043732348,
      "learning_rate": 3.5125993113937816e-06,
      "loss": 0.0195,
      "step": 7121
    },
    {
      "epoch": 1.909895414320193,
      "grad_norm": 0.23237998100882748,
      "learning_rate": 3.5111098295524303e-06,
      "loss": 0.018,
      "step": 7122
    },
    {
      "epoch": 1.9101635827299543,
      "grad_norm": 0.26747933204721447,
      "learning_rate": 3.5096204926803623e-06,
      "loss": 0.0253,
      "step": 7123
    },
    {
      "epoch": 1.9104317511397158,
      "grad_norm": 0.2990659677619925,
      "learning_rate": 3.5081313009225882e-06,
      "loss": 0.0205,
      "step": 7124
    },
    {
      "epoch": 1.9106999195494772,
      "grad_norm": 0.2168266502087794,
      "learning_rate": 3.5066422544241096e-06,
      "loss": 0.019,
      "step": 7125
    },
    {
      "epoch": 1.9109680879592386,
      "grad_norm": 0.2390449036304791,
      "learning_rate": 3.505153353329911e-06,
      "loss": 0.0226,
      "step": 7126
    },
    {
      "epoch": 1.9112362563689997,
      "grad_norm": 0.24053754421376564,
      "learning_rate": 3.5036645977849605e-06,
      "loss": 0.0226,
      "step": 7127
    },
    {
      "epoch": 1.911504424778761,
      "grad_norm": 0.3254712317406596,
      "learning_rate": 3.5021759879342167e-06,
      "loss": 0.0271,
      "step": 7128
    },
    {
      "epoch": 1.9117725931885223,
      "grad_norm": 0.1958839737995743,
      "learning_rate": 3.5006875239226196e-06,
      "loss": 0.0182,
      "step": 7129
    },
    {
      "epoch": 1.9120407615982837,
      "grad_norm": 0.2412095689354165,
      "learning_rate": 3.4991992058950973e-06,
      "loss": 0.02,
      "step": 7130
    },
    {
      "epoch": 1.9123089300080451,
      "grad_norm": 0.3022310793339784,
      "learning_rate": 3.497711033996564e-06,
      "loss": 0.0172,
      "step": 7131
    },
    {
      "epoch": 1.9125770984178065,
      "grad_norm": 0.2527084266574049,
      "learning_rate": 3.496223008371919e-06,
      "loss": 0.0149,
      "step": 7132
    },
    {
      "epoch": 1.9128452668275677,
      "grad_norm": 0.25208367710569185,
      "learning_rate": 3.494735129166047e-06,
      "loss": 0.0261,
      "step": 7133
    },
    {
      "epoch": 1.913113435237329,
      "grad_norm": 0.4008542264004503,
      "learning_rate": 3.4932473965238177e-06,
      "loss": 0.0267,
      "step": 7134
    },
    {
      "epoch": 1.9133816036470903,
      "grad_norm": 0.1965505746694254,
      "learning_rate": 3.491759810590091e-06,
      "loss": 0.0131,
      "step": 7135
    },
    {
      "epoch": 1.9136497720568517,
      "grad_norm": 0.38812085596641105,
      "learning_rate": 3.4902723715097064e-06,
      "loss": 0.0303,
      "step": 7136
    },
    {
      "epoch": 1.9139179404666131,
      "grad_norm": 0.33912552162151005,
      "learning_rate": 3.4887850794274934e-06,
      "loss": 0.0316,
      "step": 7137
    },
    {
      "epoch": 1.9141861088763745,
      "grad_norm": 0.26935010123984887,
      "learning_rate": 3.487297934488264e-06,
      "loss": 0.0218,
      "step": 7138
    },
    {
      "epoch": 1.9144542772861357,
      "grad_norm": 0.37203557446645347,
      "learning_rate": 3.48581093683682e-06,
      "loss": 0.0221,
      "step": 7139
    },
    {
      "epoch": 1.914722445695897,
      "grad_norm": 0.22235062689065008,
      "learning_rate": 3.484324086617945e-06,
      "loss": 0.018,
      "step": 7140
    },
    {
      "epoch": 1.9149906141056583,
      "grad_norm": 0.2279918453222769,
      "learning_rate": 3.4828373839764084e-06,
      "loss": 0.0177,
      "step": 7141
    },
    {
      "epoch": 1.9152587825154197,
      "grad_norm": 0.22661893251707593,
      "learning_rate": 3.4813508290569685e-06,
      "loss": 0.0208,
      "step": 7142
    },
    {
      "epoch": 1.915526950925181,
      "grad_norm": 0.27673828332223194,
      "learning_rate": 3.4798644220043663e-06,
      "loss": 0.0189,
      "step": 7143
    },
    {
      "epoch": 1.9157951193349425,
      "grad_norm": 0.3117087018257065,
      "learning_rate": 3.4783781629633295e-06,
      "loss": 0.0231,
      "step": 7144
    },
    {
      "epoch": 1.9160632877447037,
      "grad_norm": 0.2831047371764861,
      "learning_rate": 3.47689205207857e-06,
      "loss": 0.0254,
      "step": 7145
    },
    {
      "epoch": 1.9163314561544649,
      "grad_norm": 0.27731941056345627,
      "learning_rate": 3.475406089494788e-06,
      "loss": 0.0207,
      "step": 7146
    },
    {
      "epoch": 1.9165996245642263,
      "grad_norm": 0.32552537424055394,
      "learning_rate": 3.473920275356668e-06,
      "loss": 0.033,
      "step": 7147
    },
    {
      "epoch": 1.9168677929739877,
      "grad_norm": 0.32894881999074593,
      "learning_rate": 3.472434609808879e-06,
      "loss": 0.0243,
      "step": 7148
    },
    {
      "epoch": 1.917135961383749,
      "grad_norm": 0.17805837741423605,
      "learning_rate": 3.4709490929960763e-06,
      "loss": 0.0138,
      "step": 7149
    },
    {
      "epoch": 1.9174041297935103,
      "grad_norm": 0.3966218260711857,
      "learning_rate": 3.469463725062901e-06,
      "loss": 0.0372,
      "step": 7150
    },
    {
      "epoch": 1.9176722982032717,
      "grad_norm": 0.27974986981162187,
      "learning_rate": 3.4679785061539786e-06,
      "loss": 0.0245,
      "step": 7151
    },
    {
      "epoch": 1.9179404666130329,
      "grad_norm": 0.25413126519346585,
      "learning_rate": 3.466493436413924e-06,
      "loss": 0.0231,
      "step": 7152
    },
    {
      "epoch": 1.9182086350227943,
      "grad_norm": 0.21716194407860884,
      "learning_rate": 3.4650085159873303e-06,
      "loss": 0.0212,
      "step": 7153
    },
    {
      "epoch": 1.9184768034325557,
      "grad_norm": 0.26567614042245846,
      "learning_rate": 3.4635237450187825e-06,
      "loss": 0.0261,
      "step": 7154
    },
    {
      "epoch": 1.918744971842317,
      "grad_norm": 0.4032409840009527,
      "learning_rate": 3.462039123652847e-06,
      "loss": 0.0229,
      "step": 7155
    },
    {
      "epoch": 1.9190131402520783,
      "grad_norm": 0.2961964304892529,
      "learning_rate": 3.4605546520340797e-06,
      "loss": 0.0257,
      "step": 7156
    },
    {
      "epoch": 1.9192813086618397,
      "grad_norm": 0.24278901595220345,
      "learning_rate": 3.4590703303070196e-06,
      "loss": 0.0189,
      "step": 7157
    },
    {
      "epoch": 1.9195494770716008,
      "grad_norm": 0.2304473108782152,
      "learning_rate": 3.4575861586161895e-06,
      "loss": 0.0159,
      "step": 7158
    },
    {
      "epoch": 1.9198176454813622,
      "grad_norm": 0.18761957800106538,
      "learning_rate": 3.4561021371061e-06,
      "loss": 0.0142,
      "step": 7159
    },
    {
      "epoch": 1.9200858138911236,
      "grad_norm": 0.38824301838507186,
      "learning_rate": 3.454618265921247e-06,
      "loss": 0.0226,
      "step": 7160
    },
    {
      "epoch": 1.920353982300885,
      "grad_norm": 0.2276031532765234,
      "learning_rate": 3.4531345452061105e-06,
      "loss": 0.0224,
      "step": 7161
    },
    {
      "epoch": 1.9206221507106462,
      "grad_norm": 0.2540901591824478,
      "learning_rate": 3.4516509751051554e-06,
      "loss": 0.0215,
      "step": 7162
    },
    {
      "epoch": 1.9208903191204076,
      "grad_norm": 0.28058980858823557,
      "learning_rate": 3.4501675557628347e-06,
      "loss": 0.0237,
      "step": 7163
    },
    {
      "epoch": 1.9211584875301688,
      "grad_norm": 0.17620710019289818,
      "learning_rate": 3.448684287323587e-06,
      "loss": 0.0186,
      "step": 7164
    },
    {
      "epoch": 1.9214266559399302,
      "grad_norm": 0.2075668144541961,
      "learning_rate": 3.4472011699318274e-06,
      "loss": 0.0155,
      "step": 7165
    },
    {
      "epoch": 1.9216948243496916,
      "grad_norm": 0.29413577551278464,
      "learning_rate": 3.44571820373197e-06,
      "loss": 0.0223,
      "step": 7166
    },
    {
      "epoch": 1.921962992759453,
      "grad_norm": 0.2803808805642251,
      "learning_rate": 3.444235388868403e-06,
      "loss": 0.023,
      "step": 7167
    },
    {
      "epoch": 1.9222311611692142,
      "grad_norm": 0.3094673471903735,
      "learning_rate": 3.4427527254855064e-06,
      "loss": 0.019,
      "step": 7168
    },
    {
      "epoch": 1.9224993295789756,
      "grad_norm": 0.2814274784983791,
      "learning_rate": 3.4412702137276426e-06,
      "loss": 0.0157,
      "step": 7169
    },
    {
      "epoch": 1.9227674979887368,
      "grad_norm": 0.28985489314384305,
      "learning_rate": 3.43978785373916e-06,
      "loss": 0.0243,
      "step": 7170
    },
    {
      "epoch": 1.9230356663984982,
      "grad_norm": 0.24281668017009025,
      "learning_rate": 3.4383056456643917e-06,
      "loss": 0.0197,
      "step": 7171
    },
    {
      "epoch": 1.9233038348082596,
      "grad_norm": 0.2497423913142394,
      "learning_rate": 3.4368235896476547e-06,
      "loss": 0.0243,
      "step": 7172
    },
    {
      "epoch": 1.923572003218021,
      "grad_norm": 0.2671980177168584,
      "learning_rate": 3.4353416858332562e-06,
      "loss": 0.0283,
      "step": 7173
    },
    {
      "epoch": 1.9238401716277822,
      "grad_norm": 0.29430809419339904,
      "learning_rate": 3.4338599343654833e-06,
      "loss": 0.0264,
      "step": 7174
    },
    {
      "epoch": 1.9241083400375436,
      "grad_norm": 0.24272818818406977,
      "learning_rate": 3.432378335388611e-06,
      "loss": 0.0236,
      "step": 7175
    },
    {
      "epoch": 1.9243765084473048,
      "grad_norm": 0.30702434915932736,
      "learning_rate": 3.430896889046899e-06,
      "loss": 0.0238,
      "step": 7176
    },
    {
      "epoch": 1.9246446768570662,
      "grad_norm": 0.24330676953622418,
      "learning_rate": 3.42941559548459e-06,
      "loss": 0.0209,
      "step": 7177
    },
    {
      "epoch": 1.9249128452668276,
      "grad_norm": 0.24873215986675828,
      "learning_rate": 3.427934454845915e-06,
      "loss": 0.0232,
      "step": 7178
    },
    {
      "epoch": 1.925181013676589,
      "grad_norm": 0.24657372971640107,
      "learning_rate": 3.4264534672750884e-06,
      "loss": 0.0255,
      "step": 7179
    },
    {
      "epoch": 1.9254491820863502,
      "grad_norm": 0.2188551181062495,
      "learning_rate": 3.4249726329163097e-06,
      "loss": 0.0185,
      "step": 7180
    },
    {
      "epoch": 1.9257173504961116,
      "grad_norm": 0.32082288628458633,
      "learning_rate": 3.4234919519137644e-06,
      "loss": 0.0218,
      "step": 7181
    },
    {
      "epoch": 1.9259855189058728,
      "grad_norm": 0.2394315828054108,
      "learning_rate": 3.422011424411621e-06,
      "loss": 0.0281,
      "step": 7182
    },
    {
      "epoch": 1.9262536873156342,
      "grad_norm": 0.30388306820669747,
      "learning_rate": 3.420531050554036e-06,
      "loss": 0.0221,
      "step": 7183
    },
    {
      "epoch": 1.9265218557253956,
      "grad_norm": 0.2618962040986585,
      "learning_rate": 3.41905083048515e-06,
      "loss": 0.0274,
      "step": 7184
    },
    {
      "epoch": 1.926790024135157,
      "grad_norm": 0.227393532735514,
      "learning_rate": 3.417570764349087e-06,
      "loss": 0.018,
      "step": 7185
    },
    {
      "epoch": 1.9270581925449182,
      "grad_norm": 0.2678705892558985,
      "learning_rate": 3.4160908522899576e-06,
      "loss": 0.023,
      "step": 7186
    },
    {
      "epoch": 1.9273263609546796,
      "grad_norm": 0.3614885657950307,
      "learning_rate": 3.4146110944518566e-06,
      "loss": 0.0378,
      "step": 7187
    },
    {
      "epoch": 1.9275945293644408,
      "grad_norm": 0.2132148171696766,
      "learning_rate": 3.413131490978866e-06,
      "loss": 0.0209,
      "step": 7188
    },
    {
      "epoch": 1.9278626977742022,
      "grad_norm": 0.21854427691784456,
      "learning_rate": 3.411652042015047e-06,
      "loss": 0.014,
      "step": 7189
    },
    {
      "epoch": 1.9281308661839636,
      "grad_norm": 0.2494560387101364,
      "learning_rate": 3.4101727477044522e-06,
      "loss": 0.0202,
      "step": 7190
    },
    {
      "epoch": 1.928399034593725,
      "grad_norm": 0.24560861689224098,
      "learning_rate": 3.408693608191116e-06,
      "loss": 0.0187,
      "step": 7191
    },
    {
      "epoch": 1.9286672030034862,
      "grad_norm": 0.3342363343579039,
      "learning_rate": 3.4072146236190582e-06,
      "loss": 0.0204,
      "step": 7192
    },
    {
      "epoch": 1.9289353714132476,
      "grad_norm": 0.28751513464329487,
      "learning_rate": 3.4057357941322823e-06,
      "loss": 0.0222,
      "step": 7193
    },
    {
      "epoch": 1.9292035398230087,
      "grad_norm": 0.2585111833717884,
      "learning_rate": 3.404257119874781e-06,
      "loss": 0.0184,
      "step": 7194
    },
    {
      "epoch": 1.9294717082327701,
      "grad_norm": 0.22808487868361046,
      "learning_rate": 3.4027786009905272e-06,
      "loss": 0.0153,
      "step": 7195
    },
    {
      "epoch": 1.9297398766425315,
      "grad_norm": 0.19831924775167226,
      "learning_rate": 3.4013002376234805e-06,
      "loss": 0.0195,
      "step": 7196
    },
    {
      "epoch": 1.930008045052293,
      "grad_norm": 0.322005055277257,
      "learning_rate": 3.3998220299175855e-06,
      "loss": 0.0258,
      "step": 7197
    },
    {
      "epoch": 1.9302762134620541,
      "grad_norm": 0.23385855106823822,
      "learning_rate": 3.39834397801677e-06,
      "loss": 0.0288,
      "step": 7198
    },
    {
      "epoch": 1.9305443818718155,
      "grad_norm": 0.41744507805904746,
      "learning_rate": 3.3968660820649492e-06,
      "loss": 0.0648,
      "step": 7199
    },
    {
      "epoch": 1.9308125502815767,
      "grad_norm": 0.311327903330626,
      "learning_rate": 3.3953883422060247e-06,
      "loss": 0.0218,
      "step": 7200
    },
    {
      "epoch": 1.9310807186913381,
      "grad_norm": 0.27739334145761296,
      "learning_rate": 3.3939107585838747e-06,
      "loss": 0.0338,
      "step": 7201
    },
    {
      "epoch": 1.9313488871010995,
      "grad_norm": 0.2541434844328274,
      "learning_rate": 3.39243333134237e-06,
      "loss": 0.023,
      "step": 7202
    },
    {
      "epoch": 1.931617055510861,
      "grad_norm": 0.23381246788016055,
      "learning_rate": 3.3909560606253632e-06,
      "loss": 0.0172,
      "step": 7203
    },
    {
      "epoch": 1.9318852239206221,
      "grad_norm": 0.2968184767314714,
      "learning_rate": 3.3894789465766935e-06,
      "loss": 0.0261,
      "step": 7204
    },
    {
      "epoch": 1.9321533923303835,
      "grad_norm": 0.2658141494372896,
      "learning_rate": 3.388001989340183e-06,
      "loss": 0.0295,
      "step": 7205
    },
    {
      "epoch": 1.9324215607401447,
      "grad_norm": 0.2074266215311971,
      "learning_rate": 3.3865251890596395e-06,
      "loss": 0.0162,
      "step": 7206
    },
    {
      "epoch": 1.932689729149906,
      "grad_norm": 0.22734488754899393,
      "learning_rate": 3.385048545878854e-06,
      "loss": 0.0206,
      "step": 7207
    },
    {
      "epoch": 1.9329578975596675,
      "grad_norm": 0.24505164404169855,
      "learning_rate": 3.3835720599416042e-06,
      "loss": 0.0246,
      "step": 7208
    },
    {
      "epoch": 1.933226065969429,
      "grad_norm": 0.29462017521221207,
      "learning_rate": 3.3820957313916516e-06,
      "loss": 0.0245,
      "step": 7209
    },
    {
      "epoch": 1.93349423437919,
      "grad_norm": 0.30073703615901964,
      "learning_rate": 3.380619560372741e-06,
      "loss": 0.0193,
      "step": 7210
    },
    {
      "epoch": 1.9337624027889515,
      "grad_norm": 0.29354234197600404,
      "learning_rate": 3.379143547028605e-06,
      "loss": 0.022,
      "step": 7211
    },
    {
      "epoch": 1.9340305711987127,
      "grad_norm": 0.20948275570957947,
      "learning_rate": 3.3776676915029615e-06,
      "loss": 0.0168,
      "step": 7212
    },
    {
      "epoch": 1.934298739608474,
      "grad_norm": 0.2468844181637911,
      "learning_rate": 3.376191993939504e-06,
      "loss": 0.0243,
      "step": 7213
    },
    {
      "epoch": 1.9345669080182355,
      "grad_norm": 0.18626674621039455,
      "learning_rate": 3.374716454481922e-06,
      "loss": 0.0175,
      "step": 7214
    },
    {
      "epoch": 1.934835076427997,
      "grad_norm": 0.24787908005964948,
      "learning_rate": 3.3732410732738843e-06,
      "loss": 0.0187,
      "step": 7215
    },
    {
      "epoch": 1.935103244837758,
      "grad_norm": 0.36552051754551834,
      "learning_rate": 3.371765850459043e-06,
      "loss": 0.0342,
      "step": 7216
    },
    {
      "epoch": 1.9353714132475195,
      "grad_norm": 0.26048597673486495,
      "learning_rate": 3.3702907861810393e-06,
      "loss": 0.0251,
      "step": 7217
    },
    {
      "epoch": 1.9356395816572807,
      "grad_norm": 0.22184261423279286,
      "learning_rate": 3.368815880583494e-06,
      "loss": 0.026,
      "step": 7218
    },
    {
      "epoch": 1.935907750067042,
      "grad_norm": 0.17910308693188762,
      "learning_rate": 3.3673411338100163e-06,
      "loss": 0.0135,
      "step": 7219
    },
    {
      "epoch": 1.9361759184768035,
      "grad_norm": 0.29436803644453036,
      "learning_rate": 3.365866546004197e-06,
      "loss": 0.0275,
      "step": 7220
    },
    {
      "epoch": 1.9364440868865649,
      "grad_norm": 0.2113852297753171,
      "learning_rate": 3.364392117309614e-06,
      "loss": 0.0221,
      "step": 7221
    },
    {
      "epoch": 1.936712255296326,
      "grad_norm": 0.2139667079589505,
      "learning_rate": 3.3629178478698284e-06,
      "loss": 0.0116,
      "step": 7222
    },
    {
      "epoch": 1.9369804237060875,
      "grad_norm": 0.24418555802654415,
      "learning_rate": 3.3614437378283854e-06,
      "loss": 0.0256,
      "step": 7223
    },
    {
      "epoch": 1.9372485921158487,
      "grad_norm": 0.2961938992212667,
      "learning_rate": 3.3599697873288163e-06,
      "loss": 0.0288,
      "step": 7224
    },
    {
      "epoch": 1.93751676052561,
      "grad_norm": 0.21618108159324634,
      "learning_rate": 3.358495996514634e-06,
      "loss": 0.0224,
      "step": 7225
    },
    {
      "epoch": 1.9377849289353715,
      "grad_norm": 0.3339499853342691,
      "learning_rate": 3.3570223655293388e-06,
      "loss": 0.0227,
      "step": 7226
    },
    {
      "epoch": 1.9380530973451329,
      "grad_norm": 0.23482922100773249,
      "learning_rate": 3.3555488945164127e-06,
      "loss": 0.0189,
      "step": 7227
    },
    {
      "epoch": 1.938321265754894,
      "grad_norm": 0.21011744169434862,
      "learning_rate": 3.354075583619326e-06,
      "loss": 0.0171,
      "step": 7228
    },
    {
      "epoch": 1.9385894341646555,
      "grad_norm": 0.21357533378241214,
      "learning_rate": 3.3526024329815293e-06,
      "loss": 0.0237,
      "step": 7229
    },
    {
      "epoch": 1.9388576025744166,
      "grad_norm": 0.33637065603835076,
      "learning_rate": 3.351129442746459e-06,
      "loss": 0.0237,
      "step": 7230
    },
    {
      "epoch": 1.939125770984178,
      "grad_norm": 0.3036592755976574,
      "learning_rate": 3.349656613057538e-06,
      "loss": 0.0303,
      "step": 7231
    },
    {
      "epoch": 1.9393939393939394,
      "grad_norm": 0.24233432752626646,
      "learning_rate": 3.3481839440581714e-06,
      "loss": 0.0164,
      "step": 7232
    },
    {
      "epoch": 1.9396621078037009,
      "grad_norm": 0.22201470923065794,
      "learning_rate": 3.3467114358917484e-06,
      "loss": 0.0159,
      "step": 7233
    },
    {
      "epoch": 1.939930276213462,
      "grad_norm": 0.24328650220008333,
      "learning_rate": 3.345239088701644e-06,
      "loss": 0.0177,
      "step": 7234
    },
    {
      "epoch": 1.9401984446232234,
      "grad_norm": 0.19824929278595427,
      "learning_rate": 3.343766902631216e-06,
      "loss": 0.02,
      "step": 7235
    },
    {
      "epoch": 1.9404666130329846,
      "grad_norm": 0.28296271181276195,
      "learning_rate": 3.3422948778238074e-06,
      "loss": 0.0251,
      "step": 7236
    },
    {
      "epoch": 1.940734781442746,
      "grad_norm": 1.7159774157584902,
      "learning_rate": 3.340823014422746e-06,
      "loss": 0.0228,
      "step": 7237
    },
    {
      "epoch": 1.9410029498525074,
      "grad_norm": 0.4766270752010277,
      "learning_rate": 3.339351312571342e-06,
      "loss": 0.0188,
      "step": 7238
    },
    {
      "epoch": 1.9412711182622688,
      "grad_norm": 0.6164525604350836,
      "learning_rate": 3.337879772412892e-06,
      "loss": 0.0278,
      "step": 7239
    },
    {
      "epoch": 1.94153928667203,
      "grad_norm": 0.24052239713881798,
      "learning_rate": 3.3364083940906754e-06,
      "loss": 0.0219,
      "step": 7240
    },
    {
      "epoch": 1.9418074550817912,
      "grad_norm": 0.20742263430130703,
      "learning_rate": 3.3349371777479556e-06,
      "loss": 0.0189,
      "step": 7241
    },
    {
      "epoch": 1.9420756234915526,
      "grad_norm": 0.28813426158630956,
      "learning_rate": 3.3334661235279836e-06,
      "loss": 0.0189,
      "step": 7242
    },
    {
      "epoch": 1.942343791901314,
      "grad_norm": 0.2601999809015,
      "learning_rate": 3.33199523157399e-06,
      "loss": 0.0246,
      "step": 7243
    },
    {
      "epoch": 1.9426119603110754,
      "grad_norm": 0.45000488648307824,
      "learning_rate": 3.3305245020291933e-06,
      "loss": 0.021,
      "step": 7244
    },
    {
      "epoch": 1.9428801287208368,
      "grad_norm": 0.2791839331735888,
      "learning_rate": 3.3290539350367925e-06,
      "loss": 0.0238,
      "step": 7245
    },
    {
      "epoch": 1.943148297130598,
      "grad_norm": 0.936226692962895,
      "learning_rate": 3.327583530739974e-06,
      "loss": 0.027,
      "step": 7246
    },
    {
      "epoch": 1.9434164655403592,
      "grad_norm": 0.23492671749295022,
      "learning_rate": 3.3261132892819057e-06,
      "loss": 0.0204,
      "step": 7247
    },
    {
      "epoch": 1.9436846339501206,
      "grad_norm": 0.27128458575834713,
      "learning_rate": 3.3246432108057436e-06,
      "loss": 0.0245,
      "step": 7248
    },
    {
      "epoch": 1.943952802359882,
      "grad_norm": 0.6157341757689855,
      "learning_rate": 3.323173295454626e-06,
      "loss": 0.0285,
      "step": 7249
    },
    {
      "epoch": 1.9442209707696434,
      "grad_norm": 0.28910675168210076,
      "learning_rate": 3.321703543371671e-06,
      "loss": 0.0163,
      "step": 7250
    },
    {
      "epoch": 1.9444891391794048,
      "grad_norm": 0.3279579902447013,
      "learning_rate": 3.320233954699985e-06,
      "loss": 0.0241,
      "step": 7251
    },
    {
      "epoch": 1.944757307589166,
      "grad_norm": 0.3275048533410263,
      "learning_rate": 3.31876452958266e-06,
      "loss": 0.0211,
      "step": 7252
    },
    {
      "epoch": 1.9450254759989272,
      "grad_norm": 0.3711006868317802,
      "learning_rate": 3.317295268162769e-06,
      "loss": 0.0229,
      "step": 7253
    },
    {
      "epoch": 1.9452936444086886,
      "grad_norm": 0.2550440119806674,
      "learning_rate": 3.3158261705833706e-06,
      "loss": 0.0244,
      "step": 7254
    },
    {
      "epoch": 1.94556181281845,
      "grad_norm": 0.225557914235331,
      "learning_rate": 3.3143572369875064e-06,
      "loss": 0.0231,
      "step": 7255
    },
    {
      "epoch": 1.9458299812282114,
      "grad_norm": 0.23567582128649975,
      "learning_rate": 3.312888467518203e-06,
      "loss": 0.0241,
      "step": 7256
    },
    {
      "epoch": 1.9460981496379728,
      "grad_norm": 0.3077048960104107,
      "learning_rate": 3.31141986231847e-06,
      "loss": 0.0225,
      "step": 7257
    },
    {
      "epoch": 1.946366318047734,
      "grad_norm": 0.4478250638408225,
      "learning_rate": 3.309951421531301e-06,
      "loss": 0.0284,
      "step": 7258
    },
    {
      "epoch": 1.9466344864574952,
      "grad_norm": 0.25523355166992107,
      "learning_rate": 3.3084831452996768e-06,
      "loss": 0.0165,
      "step": 7259
    },
    {
      "epoch": 1.9469026548672566,
      "grad_norm": 0.1947316463659574,
      "learning_rate": 3.3070150337665573e-06,
      "loss": 0.019,
      "step": 7260
    },
    {
      "epoch": 1.947170823277018,
      "grad_norm": 0.24770198588914882,
      "learning_rate": 3.3055470870748898e-06,
      "loss": 0.02,
      "step": 7261
    },
    {
      "epoch": 1.9474389916867794,
      "grad_norm": 0.2500610610571737,
      "learning_rate": 3.3040793053676035e-06,
      "loss": 0.0196,
      "step": 7262
    },
    {
      "epoch": 1.9477071600965408,
      "grad_norm": 0.29429674818239904,
      "learning_rate": 3.302611688787612e-06,
      "loss": 0.0238,
      "step": 7263
    },
    {
      "epoch": 1.947975328506302,
      "grad_norm": 0.3142482582260968,
      "learning_rate": 3.3011442374778137e-06,
      "loss": 0.0225,
      "step": 7264
    },
    {
      "epoch": 1.9482434969160631,
      "grad_norm": 0.26214611116363046,
      "learning_rate": 3.299676951581091e-06,
      "loss": 0.0162,
      "step": 7265
    },
    {
      "epoch": 1.9485116653258245,
      "grad_norm": 0.25188292884309677,
      "learning_rate": 3.2982098312403094e-06,
      "loss": 0.0237,
      "step": 7266
    },
    {
      "epoch": 1.948779833735586,
      "grad_norm": 0.27074429397873917,
      "learning_rate": 3.2967428765983185e-06,
      "loss": 0.0157,
      "step": 7267
    },
    {
      "epoch": 1.9490480021453473,
      "grad_norm": 0.267728122834913,
      "learning_rate": 3.29527608779795e-06,
      "loss": 0.0209,
      "step": 7268
    },
    {
      "epoch": 1.9493161705551088,
      "grad_norm": 0.2827254870826069,
      "learning_rate": 3.293809464982024e-06,
      "loss": 0.0176,
      "step": 7269
    },
    {
      "epoch": 1.94958433896487,
      "grad_norm": 0.22768053765300514,
      "learning_rate": 3.292343008293341e-06,
      "loss": 0.0177,
      "step": 7270
    },
    {
      "epoch": 1.9498525073746311,
      "grad_norm": 0.19869803234542954,
      "learning_rate": 3.290876717874686e-06,
      "loss": 0.0157,
      "step": 7271
    },
    {
      "epoch": 1.9501206757843925,
      "grad_norm": 0.23543443144286302,
      "learning_rate": 3.289410593868826e-06,
      "loss": 0.0193,
      "step": 7272
    },
    {
      "epoch": 1.950388844194154,
      "grad_norm": 0.26468877854466416,
      "learning_rate": 3.287944636418517e-06,
      "loss": 0.0262,
      "step": 7273
    },
    {
      "epoch": 1.9506570126039153,
      "grad_norm": 1.3406117055371682,
      "learning_rate": 3.286478845666492e-06,
      "loss": 0.0204,
      "step": 7274
    },
    {
      "epoch": 1.9509251810136767,
      "grad_norm": 0.22614158940232104,
      "learning_rate": 3.285013221755472e-06,
      "loss": 0.0207,
      "step": 7275
    },
    {
      "epoch": 1.951193349423438,
      "grad_norm": 0.28261212982529965,
      "learning_rate": 3.2835477648281624e-06,
      "loss": 0.026,
      "step": 7276
    },
    {
      "epoch": 1.951461517833199,
      "grad_norm": 0.19577555641572983,
      "learning_rate": 3.282082475027249e-06,
      "loss": 0.016,
      "step": 7277
    },
    {
      "epoch": 1.9517296862429605,
      "grad_norm": 0.21446217015446803,
      "learning_rate": 3.280617352495403e-06,
      "loss": 0.0245,
      "step": 7278
    },
    {
      "epoch": 1.951997854652722,
      "grad_norm": 0.5695721712053765,
      "learning_rate": 3.279152397375282e-06,
      "loss": 0.031,
      "step": 7279
    },
    {
      "epoch": 1.9522660230624833,
      "grad_norm": 0.22887071708210222,
      "learning_rate": 3.277687609809523e-06,
      "loss": 0.0242,
      "step": 7280
    },
    {
      "epoch": 1.9525341914722447,
      "grad_norm": 0.17226352459104693,
      "learning_rate": 3.2762229899407494e-06,
      "loss": 0.0153,
      "step": 7281
    },
    {
      "epoch": 1.952802359882006,
      "grad_norm": 1.95857696256105,
      "learning_rate": 3.274758537911566e-06,
      "loss": 0.0288,
      "step": 7282
    },
    {
      "epoch": 1.953070528291767,
      "grad_norm": 0.23704344986946801,
      "learning_rate": 3.273294253864564e-06,
      "loss": 0.0228,
      "step": 7283
    },
    {
      "epoch": 1.9533386967015285,
      "grad_norm": 0.2599262793136385,
      "learning_rate": 3.271830137942316e-06,
      "loss": 0.0227,
      "step": 7284
    },
    {
      "epoch": 1.95360686511129,
      "grad_norm": 0.2406656249609967,
      "learning_rate": 3.2703661902873797e-06,
      "loss": 0.0226,
      "step": 7285
    },
    {
      "epoch": 1.9538750335210513,
      "grad_norm": 0.2951081278937298,
      "learning_rate": 3.2689024110422946e-06,
      "loss": 0.0237,
      "step": 7286
    },
    {
      "epoch": 1.9541432019308127,
      "grad_norm": 0.2986252436390067,
      "learning_rate": 3.267438800349586e-06,
      "loss": 0.0189,
      "step": 7287
    },
    {
      "epoch": 1.9544113703405739,
      "grad_norm": 0.2089620551078566,
      "learning_rate": 3.2659753583517613e-06,
      "loss": 0.0153,
      "step": 7288
    },
    {
      "epoch": 1.954679538750335,
      "grad_norm": 0.2374683369847728,
      "learning_rate": 3.2645120851913116e-06,
      "loss": 0.0163,
      "step": 7289
    },
    {
      "epoch": 1.9549477071600965,
      "grad_norm": 0.2110732831376364,
      "learning_rate": 3.2630489810107125e-06,
      "loss": 0.0194,
      "step": 7290
    },
    {
      "epoch": 1.9552158755698579,
      "grad_norm": 0.3688689667573004,
      "learning_rate": 3.261586045952424e-06,
      "loss": 0.0284,
      "step": 7291
    },
    {
      "epoch": 1.9554840439796193,
      "grad_norm": 0.24806684077869232,
      "learning_rate": 3.2601232801588857e-06,
      "loss": 0.0217,
      "step": 7292
    },
    {
      "epoch": 1.9557522123893807,
      "grad_norm": 0.2438056003496756,
      "learning_rate": 3.258660683772525e-06,
      "loss": 0.0159,
      "step": 7293
    },
    {
      "epoch": 1.9560203807991419,
      "grad_norm": 0.4361051095382746,
      "learning_rate": 3.25719825693575e-06,
      "loss": 0.0209,
      "step": 7294
    },
    {
      "epoch": 1.956288549208903,
      "grad_norm": 0.21364374597346794,
      "learning_rate": 3.255735999790953e-06,
      "loss": 0.0175,
      "step": 7295
    },
    {
      "epoch": 1.9565567176186645,
      "grad_norm": 0.3588876401576768,
      "learning_rate": 3.2542739124805113e-06,
      "loss": 0.022,
      "step": 7296
    },
    {
      "epoch": 1.9568248860284259,
      "grad_norm": 0.2799114898841924,
      "learning_rate": 3.252811995146786e-06,
      "loss": 0.0203,
      "step": 7297
    },
    {
      "epoch": 1.9570930544381873,
      "grad_norm": 0.2814971297534194,
      "learning_rate": 3.251350247932117e-06,
      "loss": 0.0265,
      "step": 7298
    },
    {
      "epoch": 1.9573612228479487,
      "grad_norm": 0.24009168386660548,
      "learning_rate": 3.2498886709788298e-06,
      "loss": 0.0203,
      "step": 7299
    },
    {
      "epoch": 1.9576293912577098,
      "grad_norm": 0.17919401403269225,
      "learning_rate": 3.248427264429237e-06,
      "loss": 0.015,
      "step": 7300
    },
    {
      "epoch": 1.957897559667471,
      "grad_norm": 0.3011371564775238,
      "learning_rate": 3.246966028425632e-06,
      "loss": 0.0242,
      "step": 7301
    },
    {
      "epoch": 1.9581657280772324,
      "grad_norm": 0.2544898288661922,
      "learning_rate": 3.2455049631102896e-06,
      "loss": 0.0187,
      "step": 7302
    },
    {
      "epoch": 1.9584338964869938,
      "grad_norm": 0.21456869475937265,
      "learning_rate": 3.244044068625472e-06,
      "loss": 0.0187,
      "step": 7303
    },
    {
      "epoch": 1.9587020648967552,
      "grad_norm": 0.26101091975300716,
      "learning_rate": 3.242583345113421e-06,
      "loss": 0.0209,
      "step": 7304
    },
    {
      "epoch": 1.9589702333065167,
      "grad_norm": 0.2598864629403866,
      "learning_rate": 3.2411227927163636e-06,
      "loss": 0.0202,
      "step": 7305
    },
    {
      "epoch": 1.9592384017162778,
      "grad_norm": 0.2660274408710632,
      "learning_rate": 3.2396624115765095e-06,
      "loss": 0.02,
      "step": 7306
    },
    {
      "epoch": 1.959506570126039,
      "grad_norm": 0.2783187593611475,
      "learning_rate": 3.2382022018360537e-06,
      "loss": 0.0306,
      "step": 7307
    },
    {
      "epoch": 1.9597747385358004,
      "grad_norm": 0.26086595271569946,
      "learning_rate": 3.2367421636371717e-06,
      "loss": 0.0237,
      "step": 7308
    },
    {
      "epoch": 1.9600429069455618,
      "grad_norm": 0.19809928884422928,
      "learning_rate": 3.2352822971220255e-06,
      "loss": 0.0136,
      "step": 7309
    },
    {
      "epoch": 1.9603110753553232,
      "grad_norm": 0.272623021485699,
      "learning_rate": 3.2338226024327545e-06,
      "loss": 0.0284,
      "step": 7310
    },
    {
      "epoch": 1.9605792437650846,
      "grad_norm": 0.19195028280333934,
      "learning_rate": 3.2323630797114892e-06,
      "loss": 0.0163,
      "step": 7311
    },
    {
      "epoch": 1.9608474121748458,
      "grad_norm": 0.2516825566785902,
      "learning_rate": 3.2309037291003363e-06,
      "loss": 0.0191,
      "step": 7312
    },
    {
      "epoch": 1.961115580584607,
      "grad_norm": 0.20860736884877254,
      "learning_rate": 3.229444550741392e-06,
      "loss": 0.0164,
      "step": 7313
    },
    {
      "epoch": 1.9613837489943684,
      "grad_norm": 0.2018496953655276,
      "learning_rate": 3.22798554477673e-06,
      "loss": 0.017,
      "step": 7314
    },
    {
      "epoch": 1.9616519174041298,
      "grad_norm": 0.18254981433131096,
      "learning_rate": 3.2265267113484113e-06,
      "loss": 0.017,
      "step": 7315
    },
    {
      "epoch": 1.9619200858138912,
      "grad_norm": 0.22958193195765997,
      "learning_rate": 3.2250680505984773e-06,
      "loss": 0.0167,
      "step": 7316
    },
    {
      "epoch": 1.9621882542236526,
      "grad_norm": 0.3633120850043984,
      "learning_rate": 3.223609562668956e-06,
      "loss": 0.0234,
      "step": 7317
    },
    {
      "epoch": 1.9624564226334138,
      "grad_norm": 0.254500331645394,
      "learning_rate": 3.2221512477018546e-06,
      "loss": 0.0169,
      "step": 7318
    },
    {
      "epoch": 1.962724591043175,
      "grad_norm": 0.24183995972691102,
      "learning_rate": 3.2206931058391668e-06,
      "loss": 0.0248,
      "step": 7319
    },
    {
      "epoch": 1.9629927594529364,
      "grad_norm": 0.22707204737752154,
      "learning_rate": 3.219235137222867e-06,
      "loss": 0.0163,
      "step": 7320
    },
    {
      "epoch": 1.9632609278626978,
      "grad_norm": 0.2720860932807066,
      "learning_rate": 3.217777341994915e-06,
      "loss": 0.0287,
      "step": 7321
    },
    {
      "epoch": 1.9635290962724592,
      "grad_norm": 0.22002086700853982,
      "learning_rate": 3.2163197202972505e-06,
      "loss": 0.0196,
      "step": 7322
    },
    {
      "epoch": 1.9637972646822206,
      "grad_norm": 0.22239090285014979,
      "learning_rate": 3.214862272271799e-06,
      "loss": 0.0151,
      "step": 7323
    },
    {
      "epoch": 1.9640654330919818,
      "grad_norm": 0.20407577512504013,
      "learning_rate": 3.21340499806047e-06,
      "loss": 0.0185,
      "step": 7324
    },
    {
      "epoch": 1.964333601501743,
      "grad_norm": 0.20539263802490418,
      "learning_rate": 3.211947897805151e-06,
      "loss": 0.0163,
      "step": 7325
    },
    {
      "epoch": 1.9646017699115044,
      "grad_norm": 0.24540685676415408,
      "learning_rate": 3.2104909716477184e-06,
      "loss": 0.0199,
      "step": 7326
    },
    {
      "epoch": 1.9648699383212658,
      "grad_norm": 0.23768470831320426,
      "learning_rate": 3.2090342197300296e-06,
      "loss": 0.0179,
      "step": 7327
    },
    {
      "epoch": 1.9651381067310272,
      "grad_norm": 0.22811968805251903,
      "learning_rate": 3.2075776421939235e-06,
      "loss": 0.0231,
      "step": 7328
    },
    {
      "epoch": 1.9654062751407884,
      "grad_norm": 0.35688940079785403,
      "learning_rate": 3.2061212391812237e-06,
      "loss": 0.0299,
      "step": 7329
    },
    {
      "epoch": 1.9656744435505498,
      "grad_norm": 0.25489483982162836,
      "learning_rate": 3.204665010833736e-06,
      "loss": 0.0188,
      "step": 7330
    },
    {
      "epoch": 1.965942611960311,
      "grad_norm": 0.3011087616456042,
      "learning_rate": 3.2032089572932507e-06,
      "loss": 0.0205,
      "step": 7331
    },
    {
      "epoch": 1.9662107803700724,
      "grad_norm": 0.22885799414865518,
      "learning_rate": 3.2017530787015383e-06,
      "loss": 0.024,
      "step": 7332
    },
    {
      "epoch": 1.9664789487798338,
      "grad_norm": 0.378775383877279,
      "learning_rate": 3.200297375200355e-06,
      "loss": 0.0157,
      "step": 7333
    },
    {
      "epoch": 1.9667471171895952,
      "grad_norm": 0.17746184834781112,
      "learning_rate": 3.1988418469314374e-06,
      "loss": 0.0129,
      "step": 7334
    },
    {
      "epoch": 1.9670152855993563,
      "grad_norm": 0.2993150294588135,
      "learning_rate": 3.1973864940365076e-06,
      "loss": 0.034,
      "step": 7335
    },
    {
      "epoch": 1.9672834540091177,
      "grad_norm": 0.26434196326787734,
      "learning_rate": 3.195931316657268e-06,
      "loss": 0.0196,
      "step": 7336
    },
    {
      "epoch": 1.967551622418879,
      "grad_norm": 0.20504594226240752,
      "learning_rate": 3.1944763149354064e-06,
      "loss": 0.0192,
      "step": 7337
    },
    {
      "epoch": 1.9678197908286403,
      "grad_norm": 0.2653934118817414,
      "learning_rate": 3.1930214890125925e-06,
      "loss": 0.0203,
      "step": 7338
    },
    {
      "epoch": 1.9680879592384017,
      "grad_norm": 0.24862668972346125,
      "learning_rate": 3.1915668390304794e-06,
      "loss": 0.0202,
      "step": 7339
    },
    {
      "epoch": 1.9683561276481631,
      "grad_norm": 0.22975807044828392,
      "learning_rate": 3.190112365130702e-06,
      "loss": 0.019,
      "step": 7340
    },
    {
      "epoch": 1.9686242960579243,
      "grad_norm": 0.18577627817299944,
      "learning_rate": 3.1886580674548784e-06,
      "loss": 0.0165,
      "step": 7341
    },
    {
      "epoch": 1.9688924644676857,
      "grad_norm": 0.3053345260670494,
      "learning_rate": 3.187203946144609e-06,
      "loss": 0.0186,
      "step": 7342
    },
    {
      "epoch": 1.969160632877447,
      "grad_norm": 0.6893799695770465,
      "learning_rate": 3.1857500013414782e-06,
      "loss": 0.0313,
      "step": 7343
    },
    {
      "epoch": 1.9694288012872083,
      "grad_norm": 0.27379994105932004,
      "learning_rate": 3.184296233187054e-06,
      "loss": 0.022,
      "step": 7344
    },
    {
      "epoch": 1.9696969696969697,
      "grad_norm": 0.3314645121203494,
      "learning_rate": 3.182842641822887e-06,
      "loss": 0.0328,
      "step": 7345
    },
    {
      "epoch": 1.9699651381067311,
      "grad_norm": 0.18876612622861563,
      "learning_rate": 3.1813892273905056e-06,
      "loss": 0.0149,
      "step": 7346
    },
    {
      "epoch": 1.9702333065164923,
      "grad_norm": 0.32912056657911964,
      "learning_rate": 3.179935990031425e-06,
      "loss": 0.0257,
      "step": 7347
    },
    {
      "epoch": 1.9705014749262537,
      "grad_norm": 0.21852652622625018,
      "learning_rate": 3.1784829298871467e-06,
      "loss": 0.0186,
      "step": 7348
    },
    {
      "epoch": 1.970769643336015,
      "grad_norm": 0.33535109451713574,
      "learning_rate": 3.1770300470991487e-06,
      "loss": 0.0231,
      "step": 7349
    },
    {
      "epoch": 1.9710378117457763,
      "grad_norm": 0.2788020002932741,
      "learning_rate": 3.1755773418088966e-06,
      "loss": 0.0232,
      "step": 7350
    },
    {
      "epoch": 1.9713059801555377,
      "grad_norm": 0.20622793888609375,
      "learning_rate": 3.1741248141578334e-06,
      "loss": 0.016,
      "step": 7351
    },
    {
      "epoch": 1.9715741485652991,
      "grad_norm": 0.2550309528468716,
      "learning_rate": 3.17267246428739e-06,
      "loss": 0.0166,
      "step": 7352
    },
    {
      "epoch": 1.9718423169750603,
      "grad_norm": 0.2029493381118636,
      "learning_rate": 3.171220292338978e-06,
      "loss": 0.0178,
      "step": 7353
    },
    {
      "epoch": 1.9721104853848217,
      "grad_norm": 0.3143055064180271,
      "learning_rate": 3.169768298453989e-06,
      "loss": 0.0174,
      "step": 7354
    },
    {
      "epoch": 1.9723786537945829,
      "grad_norm": 0.24823623314562687,
      "learning_rate": 3.1683164827738024e-06,
      "loss": 0.0215,
      "step": 7355
    },
    {
      "epoch": 1.9726468222043443,
      "grad_norm": 0.2916915396643583,
      "learning_rate": 3.1668648454397778e-06,
      "loss": 0.0254,
      "step": 7356
    },
    {
      "epoch": 1.9729149906141057,
      "grad_norm": 0.24807135993276522,
      "learning_rate": 3.165413386593256e-06,
      "loss": 0.0206,
      "step": 7357
    },
    {
      "epoch": 1.973183159023867,
      "grad_norm": 0.38789111520190983,
      "learning_rate": 3.1639621063755622e-06,
      "loss": 0.0429,
      "step": 7358
    },
    {
      "epoch": 1.9734513274336283,
      "grad_norm": 0.21587589980216482,
      "learning_rate": 3.162511004928003e-06,
      "loss": 0.015,
      "step": 7359
    },
    {
      "epoch": 1.9737194958433897,
      "grad_norm": 0.21760216320504983,
      "learning_rate": 3.161060082391869e-06,
      "loss": 0.0201,
      "step": 7360
    },
    {
      "epoch": 1.9739876642531509,
      "grad_norm": 0.24406782437877078,
      "learning_rate": 3.1596093389084327e-06,
      "loss": 0.0203,
      "step": 7361
    },
    {
      "epoch": 1.9742558326629123,
      "grad_norm": 0.22483130290143524,
      "learning_rate": 3.158158774618948e-06,
      "loss": 0.0163,
      "step": 7362
    },
    {
      "epoch": 1.9745240010726737,
      "grad_norm": 0.36283461802211503,
      "learning_rate": 3.1567083896646545e-06,
      "loss": 0.0239,
      "step": 7363
    },
    {
      "epoch": 1.974792169482435,
      "grad_norm": 0.27661984566535774,
      "learning_rate": 3.1552581841867695e-06,
      "loss": 0.0271,
      "step": 7364
    },
    {
      "epoch": 1.9750603378921963,
      "grad_norm": 0.21704552032261018,
      "learning_rate": 3.153808158326499e-06,
      "loss": 0.0177,
      "step": 7365
    },
    {
      "epoch": 1.9753285063019577,
      "grad_norm": 0.32349575102684636,
      "learning_rate": 3.1523583122250262e-06,
      "loss": 0.0265,
      "step": 7366
    },
    {
      "epoch": 1.9755966747117188,
      "grad_norm": 0.2065823253197649,
      "learning_rate": 3.1509086460235196e-06,
      "loss": 0.0146,
      "step": 7367
    },
    {
      "epoch": 1.9758648431214803,
      "grad_norm": 0.2576949465385535,
      "learning_rate": 3.149459159863128e-06,
      "loss": 0.023,
      "step": 7368
    },
    {
      "epoch": 1.9761330115312417,
      "grad_norm": 0.23996351615262054,
      "learning_rate": 3.1480098538849858e-06,
      "loss": 0.0261,
      "step": 7369
    },
    {
      "epoch": 1.976401179941003,
      "grad_norm": 0.2528034941796443,
      "learning_rate": 3.146560728230208e-06,
      "loss": 0.0269,
      "step": 7370
    },
    {
      "epoch": 1.9766693483507642,
      "grad_norm": 0.26831565550418635,
      "learning_rate": 3.1451117830398896e-06,
      "loss": 0.0272,
      "step": 7371
    },
    {
      "epoch": 1.9769375167605256,
      "grad_norm": 0.3226853905969348,
      "learning_rate": 3.1436630184551133e-06,
      "loss": 0.0162,
      "step": 7372
    },
    {
      "epoch": 1.9772056851702868,
      "grad_norm": 0.2878301407997294,
      "learning_rate": 3.1422144346169392e-06,
      "loss": 0.0291,
      "step": 7373
    },
    {
      "epoch": 1.9774738535800482,
      "grad_norm": 0.23692783745681056,
      "learning_rate": 3.1407660316664135e-06,
      "loss": 0.02,
      "step": 7374
    },
    {
      "epoch": 1.9777420219898096,
      "grad_norm": 0.26298198501097647,
      "learning_rate": 3.139317809744563e-06,
      "loss": 0.0235,
      "step": 7375
    },
    {
      "epoch": 1.978010190399571,
      "grad_norm": 0.2416055364972395,
      "learning_rate": 3.1378697689923987e-06,
      "loss": 0.0198,
      "step": 7376
    },
    {
      "epoch": 1.9782783588093322,
      "grad_norm": 0.2944780137913131,
      "learning_rate": 3.1364219095509097e-06,
      "loss": 0.0375,
      "step": 7377
    },
    {
      "epoch": 1.9785465272190936,
      "grad_norm": 0.2031424635839962,
      "learning_rate": 3.1349742315610728e-06,
      "loss": 0.0217,
      "step": 7378
    },
    {
      "epoch": 1.9788146956288548,
      "grad_norm": 0.22271753917188494,
      "learning_rate": 3.133526735163843e-06,
      "loss": 0.0205,
      "step": 7379
    },
    {
      "epoch": 1.9790828640386162,
      "grad_norm": 0.2136198449560959,
      "learning_rate": 3.1320794205001593e-06,
      "loss": 0.0171,
      "step": 7380
    },
    {
      "epoch": 1.9793510324483776,
      "grad_norm": 0.282087360678035,
      "learning_rate": 3.1306322877109428e-06,
      "loss": 0.0176,
      "step": 7381
    },
    {
      "epoch": 1.979619200858139,
      "grad_norm": 0.222004588244296,
      "learning_rate": 3.1291853369370994e-06,
      "loss": 0.0131,
      "step": 7382
    },
    {
      "epoch": 1.9798873692679002,
      "grad_norm": 0.3151243301339527,
      "learning_rate": 3.1277385683195117e-06,
      "loss": 0.0201,
      "step": 7383
    },
    {
      "epoch": 1.9801555376776616,
      "grad_norm": 0.4881756423051212,
      "learning_rate": 3.1262919819990487e-06,
      "loss": 0.0227,
      "step": 7384
    },
    {
      "epoch": 1.9804237060874228,
      "grad_norm": 0.22106597372028725,
      "learning_rate": 3.12484557811656e-06,
      "loss": 0.0208,
      "step": 7385
    },
    {
      "epoch": 1.9806918744971842,
      "grad_norm": 0.34861999846441566,
      "learning_rate": 3.1233993568128795e-06,
      "loss": 0.038,
      "step": 7386
    },
    {
      "epoch": 1.9809600429069456,
      "grad_norm": 0.2140444409961593,
      "learning_rate": 3.1219533182288215e-06,
      "loss": 0.0151,
      "step": 7387
    },
    {
      "epoch": 1.981228211316707,
      "grad_norm": 0.23096658649425703,
      "learning_rate": 3.120507462505183e-06,
      "loss": 0.0194,
      "step": 7388
    },
    {
      "epoch": 1.9814963797264682,
      "grad_norm": 0.2571300662907828,
      "learning_rate": 3.1190617897827424e-06,
      "loss": 0.0205,
      "step": 7389
    },
    {
      "epoch": 1.9817645481362296,
      "grad_norm": 0.26626853620922364,
      "learning_rate": 3.117616300202262e-06,
      "loss": 0.0178,
      "step": 7390
    },
    {
      "epoch": 1.9820327165459908,
      "grad_norm": 0.2347314821306523,
      "learning_rate": 3.116170993904484e-06,
      "loss": 0.018,
      "step": 7391
    },
    {
      "epoch": 1.9823008849557522,
      "grad_norm": 0.2665582790794659,
      "learning_rate": 3.1147258710301353e-06,
      "loss": 0.0253,
      "step": 7392
    },
    {
      "epoch": 1.9825690533655136,
      "grad_norm": 0.2235231577765956,
      "learning_rate": 3.1132809317199232e-06,
      "loss": 0.0163,
      "step": 7393
    },
    {
      "epoch": 1.982837221775275,
      "grad_norm": 0.291832044055794,
      "learning_rate": 3.11183617611454e-06,
      "loss": 0.0276,
      "step": 7394
    },
    {
      "epoch": 1.9831053901850362,
      "grad_norm": 0.2565445659398976,
      "learning_rate": 3.110391604354652e-06,
      "loss": 0.0182,
      "step": 7395
    },
    {
      "epoch": 1.9833735585947976,
      "grad_norm": 0.24574814995802347,
      "learning_rate": 3.1089472165809182e-06,
      "loss": 0.022,
      "step": 7396
    },
    {
      "epoch": 1.9836417270045588,
      "grad_norm": 0.21008965169423827,
      "learning_rate": 3.1075030129339726e-06,
      "loss": 0.0194,
      "step": 7397
    },
    {
      "epoch": 1.9839098954143202,
      "grad_norm": 0.27711553277624235,
      "learning_rate": 3.106058993554435e-06,
      "loss": 0.024,
      "step": 7398
    },
    {
      "epoch": 1.9841780638240816,
      "grad_norm": 0.2827563800443053,
      "learning_rate": 3.1046151585829044e-06,
      "loss": 0.019,
      "step": 7399
    },
    {
      "epoch": 1.984446232233843,
      "grad_norm": 0.24667839000918576,
      "learning_rate": 3.103171508159963e-06,
      "loss": 0.0213,
      "step": 7400
    },
    {
      "epoch": 1.9847144006436042,
      "grad_norm": 0.2594698001185999,
      "learning_rate": 3.101728042426177e-06,
      "loss": 0.0143,
      "step": 7401
    },
    {
      "epoch": 1.9849825690533656,
      "grad_norm": 0.3020645267810662,
      "learning_rate": 3.10028476152209e-06,
      "loss": 0.0224,
      "step": 7402
    },
    {
      "epoch": 1.9852507374631267,
      "grad_norm": 0.17948683487701536,
      "learning_rate": 3.0988416655882336e-06,
      "loss": 0.0181,
      "step": 7403
    },
    {
      "epoch": 1.9855189058728882,
      "grad_norm": 0.2313641352568494,
      "learning_rate": 3.0973987547651165e-06,
      "loss": 0.0181,
      "step": 7404
    },
    {
      "epoch": 1.9857870742826496,
      "grad_norm": 0.3050315924544107,
      "learning_rate": 3.0959560291932312e-06,
      "loss": 0.0348,
      "step": 7405
    },
    {
      "epoch": 1.986055242692411,
      "grad_norm": 0.20137844016622877,
      "learning_rate": 3.0945134890130535e-06,
      "loss": 0.0188,
      "step": 7406
    },
    {
      "epoch": 1.9863234111021721,
      "grad_norm": 0.19319960525229268,
      "learning_rate": 3.093071134365037e-06,
      "loss": 0.0188,
      "step": 7407
    },
    {
      "epoch": 1.9865915795119335,
      "grad_norm": 0.28920204216037787,
      "learning_rate": 3.0916289653896213e-06,
      "loss": 0.0228,
      "step": 7408
    },
    {
      "epoch": 1.9868597479216947,
      "grad_norm": 0.4729750825144965,
      "learning_rate": 3.0901869822272273e-06,
      "loss": 0.0249,
      "step": 7409
    },
    {
      "epoch": 1.9871279163314561,
      "grad_norm": 0.2444558524642889,
      "learning_rate": 3.0887451850182558e-06,
      "loss": 0.0204,
      "step": 7410
    },
    {
      "epoch": 1.9873960847412175,
      "grad_norm": 0.3185866682215508,
      "learning_rate": 3.087303573903092e-06,
      "loss": 0.0258,
      "step": 7411
    },
    {
      "epoch": 1.987664253150979,
      "grad_norm": 0.21494902143895483,
      "learning_rate": 3.0858621490220997e-06,
      "loss": 0.0192,
      "step": 7412
    },
    {
      "epoch": 1.9879324215607401,
      "grad_norm": 0.3666635706981367,
      "learning_rate": 3.0844209105156287e-06,
      "loss": 0.0265,
      "step": 7413
    },
    {
      "epoch": 1.9882005899705013,
      "grad_norm": 0.28634963661309604,
      "learning_rate": 3.082979858524008e-06,
      "loss": 0.0381,
      "step": 7414
    },
    {
      "epoch": 1.9884687583802627,
      "grad_norm": 0.1712615015464958,
      "learning_rate": 3.0815389931875487e-06,
      "loss": 0.015,
      "step": 7415
    },
    {
      "epoch": 1.9887369267900241,
      "grad_norm": 0.3530635406178965,
      "learning_rate": 3.0800983146465445e-06,
      "loss": 0.0223,
      "step": 7416
    },
    {
      "epoch": 1.9890050951997855,
      "grad_norm": 0.2845667185918553,
      "learning_rate": 3.0786578230412707e-06,
      "loss": 0.0332,
      "step": 7417
    },
    {
      "epoch": 1.989273263609547,
      "grad_norm": 0.21285679255801424,
      "learning_rate": 3.077217518511984e-06,
      "loss": 0.0178,
      "step": 7418
    },
    {
      "epoch": 1.989541432019308,
      "grad_norm": 0.24307945681723617,
      "learning_rate": 3.075777401198922e-06,
      "loss": 0.0187,
      "step": 7419
    },
    {
      "epoch": 1.9898096004290693,
      "grad_norm": 0.21302391386330244,
      "learning_rate": 3.0743374712423052e-06,
      "loss": 0.012,
      "step": 7420
    },
    {
      "epoch": 1.9900777688388307,
      "grad_norm": 0.1799656239805504,
      "learning_rate": 3.072897728782336e-06,
      "loss": 0.0148,
      "step": 7421
    },
    {
      "epoch": 1.990345937248592,
      "grad_norm": 0.24642616148258245,
      "learning_rate": 3.0714581739591987e-06,
      "loss": 0.019,
      "step": 7422
    },
    {
      "epoch": 1.9906141056583535,
      "grad_norm": 0.21716787213132804,
      "learning_rate": 3.0700188069130592e-06,
      "loss": 0.0163,
      "step": 7423
    },
    {
      "epoch": 1.990882274068115,
      "grad_norm": 0.2968506622592472,
      "learning_rate": 3.0685796277840646e-06,
      "loss": 0.0226,
      "step": 7424
    },
    {
      "epoch": 1.991150442477876,
      "grad_norm": 0.24817357913670118,
      "learning_rate": 3.0671406367123435e-06,
      "loss": 0.0229,
      "step": 7425
    },
    {
      "epoch": 1.9914186108876373,
      "grad_norm": 0.2665521491486088,
      "learning_rate": 3.0657018338380073e-06,
      "loss": 0.0262,
      "step": 7426
    },
    {
      "epoch": 1.9916867792973987,
      "grad_norm": 0.18469737353500937,
      "learning_rate": 3.064263219301148e-06,
      "loss": 0.0197,
      "step": 7427
    },
    {
      "epoch": 1.99195494770716,
      "grad_norm": 0.24652167297969219,
      "learning_rate": 3.06282479324184e-06,
      "loss": 0.019,
      "step": 7428
    },
    {
      "epoch": 1.9922231161169215,
      "grad_norm": 0.24431141321946612,
      "learning_rate": 3.0613865558001375e-06,
      "loss": 0.0219,
      "step": 7429
    },
    {
      "epoch": 1.992491284526683,
      "grad_norm": 0.2665662830426719,
      "learning_rate": 3.0599485071160816e-06,
      "loss": 0.027,
      "step": 7430
    },
    {
      "epoch": 1.992759452936444,
      "grad_norm": 0.27019070076379764,
      "learning_rate": 3.058510647329688e-06,
      "loss": 0.0226,
      "step": 7431
    },
    {
      "epoch": 1.9930276213462053,
      "grad_norm": 0.24214299235672676,
      "learning_rate": 3.057072976580957e-06,
      "loss": 0.0175,
      "step": 7432
    },
    {
      "epoch": 1.9932957897559667,
      "grad_norm": 0.20842572650993418,
      "learning_rate": 3.0556354950098723e-06,
      "loss": 0.0117,
      "step": 7433
    },
    {
      "epoch": 1.993563958165728,
      "grad_norm": 0.19974925738073906,
      "learning_rate": 3.0541982027563966e-06,
      "loss": 0.0186,
      "step": 7434
    },
    {
      "epoch": 1.9938321265754895,
      "grad_norm": 0.2588081528180662,
      "learning_rate": 3.052761099960477e-06,
      "loss": 0.0243,
      "step": 7435
    },
    {
      "epoch": 1.9941002949852509,
      "grad_norm": 0.25642999768080527,
      "learning_rate": 3.0513241867620393e-06,
      "loss": 0.0207,
      "step": 7436
    },
    {
      "epoch": 1.994368463395012,
      "grad_norm": 0.3141689587911263,
      "learning_rate": 3.049887463300991e-06,
      "loss": 0.0239,
      "step": 7437
    },
    {
      "epoch": 1.9946366318047732,
      "grad_norm": 0.18123282776401012,
      "learning_rate": 3.0484509297172238e-06,
      "loss": 0.0148,
      "step": 7438
    },
    {
      "epoch": 1.9949048002145346,
      "grad_norm": 0.2051447647572944,
      "learning_rate": 3.0470145861506067e-06,
      "loss": 0.0113,
      "step": 7439
    },
    {
      "epoch": 1.995172968624296,
      "grad_norm": 0.2906063319378687,
      "learning_rate": 3.0455784327409952e-06,
      "loss": 0.0171,
      "step": 7440
    },
    {
      "epoch": 1.9954411370340575,
      "grad_norm": 0.18450219791355704,
      "learning_rate": 3.044142469628224e-06,
      "loss": 0.0211,
      "step": 7441
    },
    {
      "epoch": 1.9957093054438189,
      "grad_norm": 0.2906210929336202,
      "learning_rate": 3.042706696952109e-06,
      "loss": 0.0241,
      "step": 7442
    },
    {
      "epoch": 1.99597747385358,
      "grad_norm": 0.310680163917282,
      "learning_rate": 3.041271114852443e-06,
      "loss": 0.0241,
      "step": 7443
    },
    {
      "epoch": 1.9962456422633412,
      "grad_norm": 0.2281328746717507,
      "learning_rate": 3.03983572346901e-06,
      "loss": 0.0244,
      "step": 7444
    },
    {
      "epoch": 1.9965138106731026,
      "grad_norm": 0.253398596561827,
      "learning_rate": 3.038400522941568e-06,
      "loss": 0.0271,
      "step": 7445
    },
    {
      "epoch": 1.996781979082864,
      "grad_norm": 0.27319636185335416,
      "learning_rate": 3.036965513409859e-06,
      "loss": 0.0226,
      "step": 7446
    },
    {
      "epoch": 1.9970501474926254,
      "grad_norm": 0.19227077943847015,
      "learning_rate": 3.0355306950136064e-06,
      "loss": 0.0148,
      "step": 7447
    },
    {
      "epoch": 1.9973183159023868,
      "grad_norm": 0.18355091154511627,
      "learning_rate": 3.0340960678925145e-06,
      "loss": 0.0178,
      "step": 7448
    },
    {
      "epoch": 1.997586484312148,
      "grad_norm": 0.23418098390373618,
      "learning_rate": 3.0326616321862697e-06,
      "loss": 0.0214,
      "step": 7449
    },
    {
      "epoch": 1.9978546527219092,
      "grad_norm": 0.199396815002417,
      "learning_rate": 3.0312273880345366e-06,
      "loss": 0.0164,
      "step": 7450
    },
    {
      "epoch": 1.9981228211316706,
      "grad_norm": 0.2767245716782409,
      "learning_rate": 3.029793335576967e-06,
      "loss": 0.0191,
      "step": 7451
    },
    {
      "epoch": 1.998390989541432,
      "grad_norm": 0.2053903300568752,
      "learning_rate": 3.0283594749531897e-06,
      "loss": 0.0162,
      "step": 7452
    },
    {
      "epoch": 1.9986591579511934,
      "grad_norm": 0.38045886851341454,
      "learning_rate": 3.0269258063028164e-06,
      "loss": 0.0265,
      "step": 7453
    },
    {
      "epoch": 1.9989273263609548,
      "grad_norm": 0.20079491219064619,
      "learning_rate": 3.025492329765439e-06,
      "loss": 0.0151,
      "step": 7454
    },
    {
      "epoch": 1.999195494770716,
      "grad_norm": 0.23426590368769132,
      "learning_rate": 3.02405904548063e-06,
      "loss": 0.019,
      "step": 7455
    },
    {
      "epoch": 1.9994636631804772,
      "grad_norm": 0.20517232366711743,
      "learning_rate": 3.0226259535879464e-06,
      "loss": 0.0159,
      "step": 7456
    },
    {
      "epoch": 1.9997318315902386,
      "grad_norm": 0.22793208301656498,
      "learning_rate": 3.0211930542269235e-06,
      "loss": 0.016,
      "step": 7457
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.39354771443243336,
      "learning_rate": 3.0197603475370796e-06,
      "loss": 0.0234,
      "step": 7458
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.02269063889980316,
      "eval_runtime": 292.4648,
      "eval_samples_per_second": 85.894,
      "eval_steps_per_second": 1.344,
      "step": 7458
    },
    {
      "epoch": 2.0002681684097614,
      "grad_norm": 0.23489950279988983,
      "learning_rate": 3.018327833657913e-06,
      "loss": 0.0139,
      "step": 7459
    },
    {
      "epoch": 2.000536336819523,
      "grad_norm": 0.2166443793372242,
      "learning_rate": 3.016895512728903e-06,
      "loss": 0.0233,
      "step": 7460
    },
    {
      "epoch": 2.0008045052292838,
      "grad_norm": 0.23708360077148635,
      "learning_rate": 3.0154633848895133e-06,
      "loss": 0.0226,
      "step": 7461
    },
    {
      "epoch": 2.001072673639045,
      "grad_norm": 0.19538134775717597,
      "learning_rate": 3.0140314502791844e-06,
      "loss": 0.0188,
      "step": 7462
    },
    {
      "epoch": 2.0013408420488066,
      "grad_norm": 0.1601255417953357,
      "learning_rate": 3.0125997090373405e-06,
      "loss": 0.0122,
      "step": 7463
    },
    {
      "epoch": 2.001609010458568,
      "grad_norm": 0.16805485886862115,
      "learning_rate": 3.0111681613033865e-06,
      "loss": 0.0104,
      "step": 7464
    },
    {
      "epoch": 2.0018771788683294,
      "grad_norm": 0.1656221174760972,
      "learning_rate": 3.0097368072167087e-06,
      "loss": 0.0169,
      "step": 7465
    },
    {
      "epoch": 2.002145347278091,
      "grad_norm": 0.2600000150978306,
      "learning_rate": 3.0083056469166747e-06,
      "loss": 0.0209,
      "step": 7466
    },
    {
      "epoch": 2.0024135156878518,
      "grad_norm": 0.19028199662574155,
      "learning_rate": 3.0068746805426318e-06,
      "loss": 0.0114,
      "step": 7467
    },
    {
      "epoch": 2.002681684097613,
      "grad_norm": 0.25343507668890985,
      "learning_rate": 3.0054439082339087e-06,
      "loss": 0.0198,
      "step": 7468
    },
    {
      "epoch": 2.0029498525073746,
      "grad_norm": 0.21261757869788384,
      "learning_rate": 3.004013330129817e-06,
      "loss": 0.0175,
      "step": 7469
    },
    {
      "epoch": 2.003218020917136,
      "grad_norm": 0.1839270582563166,
      "learning_rate": 3.002582946369647e-06,
      "loss": 0.0125,
      "step": 7470
    },
    {
      "epoch": 2.0034861893268974,
      "grad_norm": 0.2037284747646333,
      "learning_rate": 3.0011527570926737e-06,
      "loss": 0.0151,
      "step": 7471
    },
    {
      "epoch": 2.0037543577366588,
      "grad_norm": 0.19902822935756748,
      "learning_rate": 2.99972276243815e-06,
      "loss": 0.0151,
      "step": 7472
    },
    {
      "epoch": 2.0040225261464197,
      "grad_norm": 0.1973807850271639,
      "learning_rate": 2.9982929625453095e-06,
      "loss": 0.0141,
      "step": 7473
    },
    {
      "epoch": 2.004290694556181,
      "grad_norm": 0.42438275880131476,
      "learning_rate": 2.996863357553369e-06,
      "loss": 0.0216,
      "step": 7474
    },
    {
      "epoch": 2.0045588629659425,
      "grad_norm": 0.2076909436157284,
      "learning_rate": 2.995433947601526e-06,
      "loss": 0.0108,
      "step": 7475
    },
    {
      "epoch": 2.004827031375704,
      "grad_norm": 0.31569897887903686,
      "learning_rate": 2.9940047328289567e-06,
      "loss": 0.0187,
      "step": 7476
    },
    {
      "epoch": 2.0050951997854654,
      "grad_norm": 0.26309934735449786,
      "learning_rate": 2.9925757133748195e-06,
      "loss": 0.017,
      "step": 7477
    },
    {
      "epoch": 2.0053633681952268,
      "grad_norm": 0.2644244493175591,
      "learning_rate": 2.9911468893782567e-06,
      "loss": 0.0165,
      "step": 7478
    },
    {
      "epoch": 2.0056315366049877,
      "grad_norm": 0.32883418637146883,
      "learning_rate": 2.9897182609783905e-06,
      "loss": 0.0154,
      "step": 7479
    },
    {
      "epoch": 2.005899705014749,
      "grad_norm": 0.18981880730016132,
      "learning_rate": 2.9882898283143168e-06,
      "loss": 0.0096,
      "step": 7480
    },
    {
      "epoch": 2.0061678734245105,
      "grad_norm": 0.22745313225750138,
      "learning_rate": 2.986861591525122e-06,
      "loss": 0.0133,
      "step": 7481
    },
    {
      "epoch": 2.006436041834272,
      "grad_norm": 0.23484915996811426,
      "learning_rate": 2.9854335507498687e-06,
      "loss": 0.016,
      "step": 7482
    },
    {
      "epoch": 2.0067042102440333,
      "grad_norm": 0.220663042121539,
      "learning_rate": 2.9840057061276033e-06,
      "loss": 0.0142,
      "step": 7483
    },
    {
      "epoch": 2.0069723786537947,
      "grad_norm": 0.16448025834135904,
      "learning_rate": 2.9825780577973493e-06,
      "loss": 0.0097,
      "step": 7484
    },
    {
      "epoch": 2.0072405470635557,
      "grad_norm": 0.16562107674097515,
      "learning_rate": 2.9811506058981143e-06,
      "loss": 0.01,
      "step": 7485
    },
    {
      "epoch": 2.007508715473317,
      "grad_norm": 0.25094263465801914,
      "learning_rate": 2.9797233505688837e-06,
      "loss": 0.0143,
      "step": 7486
    },
    {
      "epoch": 2.0077768838830785,
      "grad_norm": 0.24932291826034217,
      "learning_rate": 2.978296291948627e-06,
      "loss": 0.0294,
      "step": 7487
    },
    {
      "epoch": 2.00804505229284,
      "grad_norm": 0.19548318427479167,
      "learning_rate": 2.9768694301762935e-06,
      "loss": 0.0158,
      "step": 7488
    },
    {
      "epoch": 2.0083132207026013,
      "grad_norm": 0.17589784508019232,
      "learning_rate": 2.975442765390811e-06,
      "loss": 0.0155,
      "step": 7489
    },
    {
      "epoch": 2.0085813891123627,
      "grad_norm": 0.23841810665647767,
      "learning_rate": 2.9740162977310927e-06,
      "loss": 0.0167,
      "step": 7490
    },
    {
      "epoch": 2.0088495575221237,
      "grad_norm": 0.19605529369839467,
      "learning_rate": 2.97259002733603e-06,
      "loss": 0.0143,
      "step": 7491
    },
    {
      "epoch": 2.009117725931885,
      "grad_norm": 0.5871476747400058,
      "learning_rate": 2.9711639543444916e-06,
      "loss": 0.0404,
      "step": 7492
    },
    {
      "epoch": 2.0093858943416465,
      "grad_norm": 0.26888065896842267,
      "learning_rate": 2.969738078895333e-06,
      "loss": 0.0184,
      "step": 7493
    },
    {
      "epoch": 2.009654062751408,
      "grad_norm": 0.2823101081848726,
      "learning_rate": 2.9683124011273866e-06,
      "loss": 0.0206,
      "step": 7494
    },
    {
      "epoch": 2.0099222311611693,
      "grad_norm": 0.28155577329191805,
      "learning_rate": 2.9668869211794687e-06,
      "loss": 0.0181,
      "step": 7495
    },
    {
      "epoch": 2.0101903995709307,
      "grad_norm": 0.48348232181364204,
      "learning_rate": 2.9654616391903733e-06,
      "loss": 0.017,
      "step": 7496
    },
    {
      "epoch": 2.0104585679806917,
      "grad_norm": 0.25450549495080865,
      "learning_rate": 2.964036555298876e-06,
      "loss": 0.0167,
      "step": 7497
    },
    {
      "epoch": 2.010726736390453,
      "grad_norm": 0.20343532503522083,
      "learning_rate": 2.9626116696437333e-06,
      "loss": 0.0127,
      "step": 7498
    },
    {
      "epoch": 2.0109949048002145,
      "grad_norm": 0.19794852314342581,
      "learning_rate": 2.961186982363684e-06,
      "loss": 0.0151,
      "step": 7499
    },
    {
      "epoch": 2.011263073209976,
      "grad_norm": 0.2279525975982331,
      "learning_rate": 2.9597624935974456e-06,
      "loss": 0.0112,
      "step": 7500
    },
    {
      "epoch": 2.0115312416197373,
      "grad_norm": 0.22218260609266904,
      "learning_rate": 2.958338203483716e-06,
      "loss": 0.0133,
      "step": 7501
    },
    {
      "epoch": 2.0117994100294987,
      "grad_norm": 0.23456234024098258,
      "learning_rate": 2.9569141121611755e-06,
      "loss": 0.0112,
      "step": 7502
    },
    {
      "epoch": 2.0120675784392597,
      "grad_norm": 0.2313246901210944,
      "learning_rate": 2.9554902197684843e-06,
      "loss": 0.0125,
      "step": 7503
    },
    {
      "epoch": 2.012335746849021,
      "grad_norm": 0.23036932680523303,
      "learning_rate": 2.954066526444281e-06,
      "loss": 0.0148,
      "step": 7504
    },
    {
      "epoch": 2.0126039152587825,
      "grad_norm": 0.2679572720914199,
      "learning_rate": 2.9526430323271895e-06,
      "loss": 0.0135,
      "step": 7505
    },
    {
      "epoch": 2.012872083668544,
      "grad_norm": 0.24741511811374528,
      "learning_rate": 2.951219737555809e-06,
      "loss": 0.0145,
      "step": 7506
    },
    {
      "epoch": 2.0131402520783053,
      "grad_norm": 0.4171189661729088,
      "learning_rate": 2.949796642268724e-06,
      "loss": 0.027,
      "step": 7507
    },
    {
      "epoch": 2.0134084204880667,
      "grad_norm": 0.1830619943929899,
      "learning_rate": 2.9483737466044947e-06,
      "loss": 0.0086,
      "step": 7508
    },
    {
      "epoch": 2.0136765888978276,
      "grad_norm": 0.22385108941513912,
      "learning_rate": 2.9469510507016685e-06,
      "loss": 0.014,
      "step": 7509
    },
    {
      "epoch": 2.013944757307589,
      "grad_norm": 0.21351214165674365,
      "learning_rate": 2.945528554698768e-06,
      "loss": 0.0111,
      "step": 7510
    },
    {
      "epoch": 2.0142129257173504,
      "grad_norm": 0.2433055974705479,
      "learning_rate": 2.944106258734297e-06,
      "loss": 0.0149,
      "step": 7511
    },
    {
      "epoch": 2.014481094127112,
      "grad_norm": 0.387752912025201,
      "learning_rate": 2.9426841629467416e-06,
      "loss": 0.0166,
      "step": 7512
    },
    {
      "epoch": 2.0147492625368733,
      "grad_norm": 0.20330569283308908,
      "learning_rate": 2.9412622674745673e-06,
      "loss": 0.0089,
      "step": 7513
    },
    {
      "epoch": 2.0150174309466347,
      "grad_norm": 0.25700264546417606,
      "learning_rate": 2.939840572456221e-06,
      "loss": 0.0158,
      "step": 7514
    },
    {
      "epoch": 2.0152855993563956,
      "grad_norm": 0.25914163364973664,
      "learning_rate": 2.938419078030128e-06,
      "loss": 0.0135,
      "step": 7515
    },
    {
      "epoch": 2.015553767766157,
      "grad_norm": 0.2063836713738416,
      "learning_rate": 2.9369977843346953e-06,
      "loss": 0.0137,
      "step": 7516
    },
    {
      "epoch": 2.0158219361759184,
      "grad_norm": 0.21921724638019818,
      "learning_rate": 2.935576691508312e-06,
      "loss": 0.0145,
      "step": 7517
    },
    {
      "epoch": 2.01609010458568,
      "grad_norm": 0.24915627274791488,
      "learning_rate": 2.9341557996893434e-06,
      "loss": 0.0196,
      "step": 7518
    },
    {
      "epoch": 2.0163582729954412,
      "grad_norm": 0.2083609775697057,
      "learning_rate": 2.9327351090161416e-06,
      "loss": 0.0149,
      "step": 7519
    },
    {
      "epoch": 2.0166264414052026,
      "grad_norm": 0.26257055507353083,
      "learning_rate": 2.931314619627033e-06,
      "loss": 0.0142,
      "step": 7520
    },
    {
      "epoch": 2.0168946098149636,
      "grad_norm": 0.1936169381201401,
      "learning_rate": 2.929894331660328e-06,
      "loss": 0.0103,
      "step": 7521
    },
    {
      "epoch": 2.017162778224725,
      "grad_norm": 0.24293578218180717,
      "learning_rate": 2.9284742452543158e-06,
      "loss": 0.0148,
      "step": 7522
    },
    {
      "epoch": 2.0174309466344864,
      "grad_norm": 0.6438504353893504,
      "learning_rate": 2.9270543605472666e-06,
      "loss": 0.0175,
      "step": 7523
    },
    {
      "epoch": 2.017699115044248,
      "grad_norm": 0.22089208552804374,
      "learning_rate": 2.9256346776774314e-06,
      "loss": 0.017,
      "step": 7524
    },
    {
      "epoch": 2.017967283454009,
      "grad_norm": 0.2349678535373348,
      "learning_rate": 2.9242151967830383e-06,
      "loss": 0.0143,
      "step": 7525
    },
    {
      "epoch": 2.0182354518637706,
      "grad_norm": 0.21829784135254585,
      "learning_rate": 2.9227959180023022e-06,
      "loss": 0.0138,
      "step": 7526
    },
    {
      "epoch": 2.0185036202735316,
      "grad_norm": 0.22976061677915935,
      "learning_rate": 2.9213768414734146e-06,
      "loss": 0.0127,
      "step": 7527
    },
    {
      "epoch": 2.018771788683293,
      "grad_norm": 0.29345166398674855,
      "learning_rate": 2.9199579673345425e-06,
      "loss": 0.0179,
      "step": 7528
    },
    {
      "epoch": 2.0190399570930544,
      "grad_norm": 0.21773320080179898,
      "learning_rate": 2.91853929572384e-06,
      "loss": 0.0174,
      "step": 7529
    },
    {
      "epoch": 2.019308125502816,
      "grad_norm": 0.22800539812649076,
      "learning_rate": 2.9171208267794426e-06,
      "loss": 0.0127,
      "step": 7530
    },
    {
      "epoch": 2.019576293912577,
      "grad_norm": 0.17651883533777846,
      "learning_rate": 2.9157025606394585e-06,
      "loss": 0.0125,
      "step": 7531
    },
    {
      "epoch": 2.0198444623223386,
      "grad_norm": 0.2341419831339138,
      "learning_rate": 2.9142844974419816e-06,
      "loss": 0.0119,
      "step": 7532
    },
    {
      "epoch": 2.0201126307320996,
      "grad_norm": 0.27717449494580126,
      "learning_rate": 2.912866637325088e-06,
      "loss": 0.0178,
      "step": 7533
    },
    {
      "epoch": 2.020380799141861,
      "grad_norm": 0.2600731613202258,
      "learning_rate": 2.911448980426827e-06,
      "loss": 0.0131,
      "step": 7534
    },
    {
      "epoch": 2.0206489675516224,
      "grad_norm": 0.311333670750166,
      "learning_rate": 2.9100315268852353e-06,
      "loss": 0.0185,
      "step": 7535
    },
    {
      "epoch": 2.020917135961384,
      "grad_norm": 0.47447332089608946,
      "learning_rate": 2.9086142768383234e-06,
      "loss": 0.0149,
      "step": 7536
    },
    {
      "epoch": 2.021185304371145,
      "grad_norm": 0.3262803253519475,
      "learning_rate": 2.9071972304240893e-06,
      "loss": 0.0182,
      "step": 7537
    },
    {
      "epoch": 2.0214534727809066,
      "grad_norm": 0.2638230073706627,
      "learning_rate": 2.9057803877805014e-06,
      "loss": 0.0132,
      "step": 7538
    },
    {
      "epoch": 2.0217216411906676,
      "grad_norm": 0.28829189656336746,
      "learning_rate": 2.90436374904552e-06,
      "loss": 0.0208,
      "step": 7539
    },
    {
      "epoch": 2.021989809600429,
      "grad_norm": 0.1812612058895704,
      "learning_rate": 2.902947314357076e-06,
      "loss": 0.0086,
      "step": 7540
    },
    {
      "epoch": 2.0222579780101904,
      "grad_norm": 0.255219537459475,
      "learning_rate": 2.901531083853083e-06,
      "loss": 0.0115,
      "step": 7541
    },
    {
      "epoch": 2.0225261464199518,
      "grad_norm": 0.29334351375240375,
      "learning_rate": 2.9001150576714365e-06,
      "loss": 0.0203,
      "step": 7542
    },
    {
      "epoch": 2.022794314829713,
      "grad_norm": 0.2524550255023345,
      "learning_rate": 2.8986992359500145e-06,
      "loss": 0.0167,
      "step": 7543
    },
    {
      "epoch": 2.0230624832394746,
      "grad_norm": 0.25868206463526977,
      "learning_rate": 2.8972836188266663e-06,
      "loss": 0.0142,
      "step": 7544
    },
    {
      "epoch": 2.0233306516492355,
      "grad_norm": 0.25746762839820775,
      "learning_rate": 2.8958682064392316e-06,
      "loss": 0.0162,
      "step": 7545
    },
    {
      "epoch": 2.023598820058997,
      "grad_norm": 0.2501647148560753,
      "learning_rate": 2.894452998925522e-06,
      "loss": 0.0118,
      "step": 7546
    },
    {
      "epoch": 2.0238669884687583,
      "grad_norm": 0.29233433862903013,
      "learning_rate": 2.893037996423335e-06,
      "loss": 0.0148,
      "step": 7547
    },
    {
      "epoch": 2.0241351568785197,
      "grad_norm": 0.20507037645838974,
      "learning_rate": 2.8916231990704426e-06,
      "loss": 0.011,
      "step": 7548
    },
    {
      "epoch": 2.024403325288281,
      "grad_norm": 0.29959654755109144,
      "learning_rate": 2.8902086070046022e-06,
      "loss": 0.0169,
      "step": 7549
    },
    {
      "epoch": 2.0246714936980426,
      "grad_norm": 0.20979613095539784,
      "learning_rate": 2.88879422036355e-06,
      "loss": 0.0133,
      "step": 7550
    },
    {
      "epoch": 2.0249396621078035,
      "grad_norm": 0.24080328660394865,
      "learning_rate": 2.887380039284999e-06,
      "loss": 0.0161,
      "step": 7551
    },
    {
      "epoch": 2.025207830517565,
      "grad_norm": 0.25259137629101686,
      "learning_rate": 2.885966063906645e-06,
      "loss": 0.0152,
      "step": 7552
    },
    {
      "epoch": 2.0254759989273263,
      "grad_norm": 0.20275738222937442,
      "learning_rate": 2.8845522943661607e-06,
      "loss": 0.0092,
      "step": 7553
    },
    {
      "epoch": 2.0257441673370877,
      "grad_norm": 0.41217218894728186,
      "learning_rate": 2.8831387308012026e-06,
      "loss": 0.0147,
      "step": 7554
    },
    {
      "epoch": 2.026012335746849,
      "grad_norm": 0.530517225164426,
      "learning_rate": 2.881725373349409e-06,
      "loss": 0.0103,
      "step": 7555
    },
    {
      "epoch": 2.0262805041566105,
      "grad_norm": 0.21636164293780694,
      "learning_rate": 2.8803122221483894e-06,
      "loss": 0.0149,
      "step": 7556
    },
    {
      "epoch": 2.0265486725663715,
      "grad_norm": 0.19866528391209431,
      "learning_rate": 2.8788992773357434e-06,
      "loss": 0.0126,
      "step": 7557
    },
    {
      "epoch": 2.026816840976133,
      "grad_norm": 0.25466899694722617,
      "learning_rate": 2.877486539049041e-06,
      "loss": 0.0132,
      "step": 7558
    },
    {
      "epoch": 2.0270850093858943,
      "grad_norm": 0.3411201873121345,
      "learning_rate": 2.876074007425839e-06,
      "loss": 0.0238,
      "step": 7559
    },
    {
      "epoch": 2.0273531777956557,
      "grad_norm": 0.30859737679804833,
      "learning_rate": 2.8746616826036743e-06,
      "loss": 0.02,
      "step": 7560
    },
    {
      "epoch": 2.027621346205417,
      "grad_norm": 0.19501590285559692,
      "learning_rate": 2.873249564720057e-06,
      "loss": 0.013,
      "step": 7561
    },
    {
      "epoch": 2.0278895146151785,
      "grad_norm": 0.1990772040455571,
      "learning_rate": 2.8718376539124847e-06,
      "loss": 0.0099,
      "step": 7562
    },
    {
      "epoch": 2.0281576830249395,
      "grad_norm": 0.22436812306168474,
      "learning_rate": 2.8704259503184306e-06,
      "loss": 0.0157,
      "step": 7563
    },
    {
      "epoch": 2.028425851434701,
      "grad_norm": 0.19783852265454183,
      "learning_rate": 2.869014454075345e-06,
      "loss": 0.0116,
      "step": 7564
    },
    {
      "epoch": 2.0286940198444623,
      "grad_norm": 0.33222482334949044,
      "learning_rate": 2.867603165320667e-06,
      "loss": 0.0189,
      "step": 7565
    },
    {
      "epoch": 2.0289621882542237,
      "grad_norm": 0.28303302603205105,
      "learning_rate": 2.866192084191805e-06,
      "loss": 0.0184,
      "step": 7566
    },
    {
      "epoch": 2.029230356663985,
      "grad_norm": 0.3262104097521295,
      "learning_rate": 2.864781210826157e-06,
      "loss": 0.0155,
      "step": 7567
    },
    {
      "epoch": 2.0294985250737465,
      "grad_norm": 0.2378418028714297,
      "learning_rate": 2.863370545361091e-06,
      "loss": 0.0142,
      "step": 7568
    },
    {
      "epoch": 2.0297666934835075,
      "grad_norm": 0.28313730795896697,
      "learning_rate": 2.861960087933965e-06,
      "loss": 0.0115,
      "step": 7569
    },
    {
      "epoch": 2.030034861893269,
      "grad_norm": 0.44450501048029917,
      "learning_rate": 2.860549838682106e-06,
      "loss": 0.0193,
      "step": 7570
    },
    {
      "epoch": 2.0303030303030303,
      "grad_norm": 0.3046844779196827,
      "learning_rate": 2.8591397977428302e-06,
      "loss": 0.0143,
      "step": 7571
    },
    {
      "epoch": 2.0305711987127917,
      "grad_norm": 0.2293520704442382,
      "learning_rate": 2.8577299652534296e-06,
      "loss": 0.0128,
      "step": 7572
    },
    {
      "epoch": 2.030839367122553,
      "grad_norm": 0.2959090792850215,
      "learning_rate": 2.856320341351173e-06,
      "loss": 0.0203,
      "step": 7573
    },
    {
      "epoch": 2.0311075355323145,
      "grad_norm": 0.26044329440030056,
      "learning_rate": 2.8549109261733157e-06,
      "loss": 0.017,
      "step": 7574
    },
    {
      "epoch": 2.0313757039420755,
      "grad_norm": 0.20129799051204625,
      "learning_rate": 2.853501719857086e-06,
      "loss": 0.0126,
      "step": 7575
    },
    {
      "epoch": 2.031643872351837,
      "grad_norm": 0.20048314134540815,
      "learning_rate": 2.852092722539693e-06,
      "loss": 0.0111,
      "step": 7576
    },
    {
      "epoch": 2.0319120407615983,
      "grad_norm": 0.2376770429565458,
      "learning_rate": 2.8506839343583305e-06,
      "loss": 0.0142,
      "step": 7577
    },
    {
      "epoch": 2.0321802091713597,
      "grad_norm": 0.32710284887905805,
      "learning_rate": 2.8492753554501654e-06,
      "loss": 0.0129,
      "step": 7578
    },
    {
      "epoch": 2.032448377581121,
      "grad_norm": 0.22974977356813775,
      "learning_rate": 2.847866985952351e-06,
      "loss": 0.011,
      "step": 7579
    },
    {
      "epoch": 2.0327165459908825,
      "grad_norm": 0.27761784279976204,
      "learning_rate": 2.8464588260020127e-06,
      "loss": 0.0184,
      "step": 7580
    },
    {
      "epoch": 2.0329847144006434,
      "grad_norm": 0.3002205915429074,
      "learning_rate": 2.8450508757362604e-06,
      "loss": 0.0207,
      "step": 7581
    },
    {
      "epoch": 2.033252882810405,
      "grad_norm": 0.2364849171722325,
      "learning_rate": 2.8436431352921858e-06,
      "loss": 0.0168,
      "step": 7582
    },
    {
      "epoch": 2.0335210512201662,
      "grad_norm": 0.18787496974897674,
      "learning_rate": 2.842235604806851e-06,
      "loss": 0.0117,
      "step": 7583
    },
    {
      "epoch": 2.0337892196299276,
      "grad_norm": 0.19239178388055117,
      "learning_rate": 2.8408282844173095e-06,
      "loss": 0.0113,
      "step": 7584
    },
    {
      "epoch": 2.034057388039689,
      "grad_norm": 0.18366943066791078,
      "learning_rate": 2.839421174260583e-06,
      "loss": 0.0095,
      "step": 7585
    },
    {
      "epoch": 2.0343255564494505,
      "grad_norm": 0.45739077596746874,
      "learning_rate": 2.8380142744736825e-06,
      "loss": 0.0141,
      "step": 7586
    },
    {
      "epoch": 2.0345937248592114,
      "grad_norm": 0.1678249315102992,
      "learning_rate": 2.8366075851935927e-06,
      "loss": 0.01,
      "step": 7587
    },
    {
      "epoch": 2.034861893268973,
      "grad_norm": 0.3544836929280356,
      "learning_rate": 2.8352011065572765e-06,
      "loss": 0.0216,
      "step": 7588
    },
    {
      "epoch": 2.0351300616787342,
      "grad_norm": 0.2493553050579759,
      "learning_rate": 2.833794838701683e-06,
      "loss": 0.0157,
      "step": 7589
    },
    {
      "epoch": 2.0353982300884956,
      "grad_norm": 0.34303654641597986,
      "learning_rate": 2.832388781763733e-06,
      "loss": 0.0199,
      "step": 7590
    },
    {
      "epoch": 2.035666398498257,
      "grad_norm": 0.21021688702140712,
      "learning_rate": 2.830982935880332e-06,
      "loss": 0.0131,
      "step": 7591
    },
    {
      "epoch": 2.0359345669080184,
      "grad_norm": 0.2813391094914667,
      "learning_rate": 2.829577301188366e-06,
      "loss": 0.0222,
      "step": 7592
    },
    {
      "epoch": 2.0362027353177794,
      "grad_norm": 0.2761304039721216,
      "learning_rate": 2.8281718778246935e-06,
      "loss": 0.0168,
      "step": 7593
    },
    {
      "epoch": 2.036470903727541,
      "grad_norm": 0.24069181603053547,
      "learning_rate": 2.8267666659261618e-06,
      "loss": 0.0139,
      "step": 7594
    },
    {
      "epoch": 2.036739072137302,
      "grad_norm": 0.6661830368658052,
      "learning_rate": 2.825361665629588e-06,
      "loss": 0.0195,
      "step": 7595
    },
    {
      "epoch": 2.0370072405470636,
      "grad_norm": 0.2277700657493533,
      "learning_rate": 2.8239568770717773e-06,
      "loss": 0.015,
      "step": 7596
    },
    {
      "epoch": 2.037275408956825,
      "grad_norm": 0.34759315231499377,
      "learning_rate": 2.822552300389506e-06,
      "loss": 0.0187,
      "step": 7597
    },
    {
      "epoch": 2.0375435773665864,
      "grad_norm": 0.23356602970661425,
      "learning_rate": 2.8211479357195358e-06,
      "loss": 0.0127,
      "step": 7598
    },
    {
      "epoch": 2.0378117457763474,
      "grad_norm": 0.25408002371108773,
      "learning_rate": 2.8197437831986085e-06,
      "loss": 0.0152,
      "step": 7599
    },
    {
      "epoch": 2.038079914186109,
      "grad_norm": 0.31036162616530666,
      "learning_rate": 2.818339842963441e-06,
      "loss": 0.0163,
      "step": 7600
    },
    {
      "epoch": 2.03834808259587,
      "grad_norm": 0.19402082880136062,
      "learning_rate": 2.816936115150729e-06,
      "loss": 0.0109,
      "step": 7601
    },
    {
      "epoch": 2.0386162510056316,
      "grad_norm": 0.25002141296901886,
      "learning_rate": 2.815532599897153e-06,
      "loss": 0.0148,
      "step": 7602
    },
    {
      "epoch": 2.038884419415393,
      "grad_norm": 0.21040791446249388,
      "learning_rate": 2.8141292973393665e-06,
      "loss": 0.011,
      "step": 7603
    },
    {
      "epoch": 2.0391525878251544,
      "grad_norm": 0.20722636571663264,
      "learning_rate": 2.8127262076140083e-06,
      "loss": 0.0178,
      "step": 7604
    },
    {
      "epoch": 2.0394207562349154,
      "grad_norm": 0.2634016195371481,
      "learning_rate": 2.8113233308576905e-06,
      "loss": 0.0176,
      "step": 7605
    },
    {
      "epoch": 2.0396889246446768,
      "grad_norm": 0.19717885553385983,
      "learning_rate": 2.809920667207011e-06,
      "loss": 0.0117,
      "step": 7606
    },
    {
      "epoch": 2.039957093054438,
      "grad_norm": 0.22116492212406716,
      "learning_rate": 2.80851821679854e-06,
      "loss": 0.0126,
      "step": 7607
    },
    {
      "epoch": 2.0402252614641996,
      "grad_norm": 0.23852568685772516,
      "learning_rate": 2.8071159797688317e-06,
      "loss": 0.0128,
      "step": 7608
    },
    {
      "epoch": 2.040493429873961,
      "grad_norm": 0.21021542830938886,
      "learning_rate": 2.8057139562544206e-06,
      "loss": 0.0133,
      "step": 7609
    },
    {
      "epoch": 2.0407615982837224,
      "grad_norm": 0.2883124516518787,
      "learning_rate": 2.804312146391814e-06,
      "loss": 0.0188,
      "step": 7610
    },
    {
      "epoch": 2.0410297666934833,
      "grad_norm": 0.3689620261501219,
      "learning_rate": 2.802910550317506e-06,
      "loss": 0.0341,
      "step": 7611
    },
    {
      "epoch": 2.0412979351032448,
      "grad_norm": 0.2067491822100232,
      "learning_rate": 2.801509168167964e-06,
      "loss": 0.0137,
      "step": 7612
    },
    {
      "epoch": 2.041566103513006,
      "grad_norm": 0.2905114627154889,
      "learning_rate": 2.8001080000796367e-06,
      "loss": 0.0164,
      "step": 7613
    },
    {
      "epoch": 2.0418342719227676,
      "grad_norm": 0.21047354864477175,
      "learning_rate": 2.798707046188954e-06,
      "loss": 0.0122,
      "step": 7614
    },
    {
      "epoch": 2.042102440332529,
      "grad_norm": 0.2917454901723074,
      "learning_rate": 2.7973063066323204e-06,
      "loss": 0.0157,
      "step": 7615
    },
    {
      "epoch": 2.0423706087422904,
      "grad_norm": 0.4873833906904073,
      "learning_rate": 2.795905781546125e-06,
      "loss": 0.0178,
      "step": 7616
    },
    {
      "epoch": 2.0426387771520513,
      "grad_norm": 0.31929872031208717,
      "learning_rate": 2.79450547106673e-06,
      "loss": 0.0193,
      "step": 7617
    },
    {
      "epoch": 2.0429069455618127,
      "grad_norm": 0.21727485611875952,
      "learning_rate": 2.7931053753304823e-06,
      "loss": 0.011,
      "step": 7618
    },
    {
      "epoch": 2.043175113971574,
      "grad_norm": 0.194203062363349,
      "learning_rate": 2.7917054944737067e-06,
      "loss": 0.0095,
      "step": 7619
    },
    {
      "epoch": 2.0434432823813355,
      "grad_norm": 0.3394330226034791,
      "learning_rate": 2.7903058286327024e-06,
      "loss": 0.0151,
      "step": 7620
    },
    {
      "epoch": 2.043711450791097,
      "grad_norm": 0.2620041953314781,
      "learning_rate": 2.788906377943754e-06,
      "loss": 0.0141,
      "step": 7621
    },
    {
      "epoch": 2.0439796192008584,
      "grad_norm": 0.21341633109770938,
      "learning_rate": 2.78750714254312e-06,
      "loss": 0.0118,
      "step": 7622
    },
    {
      "epoch": 2.0442477876106193,
      "grad_norm": 0.2529716269629954,
      "learning_rate": 2.786108122567044e-06,
      "loss": 0.015,
      "step": 7623
    },
    {
      "epoch": 2.0445159560203807,
      "grad_norm": 0.25797681493688296,
      "learning_rate": 2.7847093181517416e-06,
      "loss": 0.0156,
      "step": 7624
    },
    {
      "epoch": 2.044784124430142,
      "grad_norm": 0.2288560546611802,
      "learning_rate": 2.7833107294334093e-06,
      "loss": 0.0154,
      "step": 7625
    },
    {
      "epoch": 2.0450522928399035,
      "grad_norm": 0.2591306875180103,
      "learning_rate": 2.781912356548229e-06,
      "loss": 0.012,
      "step": 7626
    },
    {
      "epoch": 2.045320461249665,
      "grad_norm": 0.3645268065715245,
      "learning_rate": 2.78051419963235e-06,
      "loss": 0.0122,
      "step": 7627
    },
    {
      "epoch": 2.0455886296594263,
      "grad_norm": 0.39408770236896157,
      "learning_rate": 2.779116258821912e-06,
      "loss": 0.0191,
      "step": 7628
    },
    {
      "epoch": 2.0458567980691873,
      "grad_norm": 0.39490895944363624,
      "learning_rate": 2.77771853425303e-06,
      "loss": 0.0205,
      "step": 7629
    },
    {
      "epoch": 2.0461249664789487,
      "grad_norm": 0.320728686999571,
      "learning_rate": 2.7763210260617913e-06,
      "loss": 0.0176,
      "step": 7630
    },
    {
      "epoch": 2.04639313488871,
      "grad_norm": 0.2222840622875859,
      "learning_rate": 2.774923734384273e-06,
      "loss": 0.0204,
      "step": 7631
    },
    {
      "epoch": 2.0466613032984715,
      "grad_norm": 0.27130286422010474,
      "learning_rate": 2.7735266593565213e-06,
      "loss": 0.0167,
      "step": 7632
    },
    {
      "epoch": 2.046929471708233,
      "grad_norm": 0.29069233423556323,
      "learning_rate": 2.7721298011145694e-06,
      "loss": 0.0121,
      "step": 7633
    },
    {
      "epoch": 2.047197640117994,
      "grad_norm": 0.30051265335161315,
      "learning_rate": 2.770733159794422e-06,
      "loss": 0.0196,
      "step": 7634
    },
    {
      "epoch": 2.0474658085277553,
      "grad_norm": 0.656495921034612,
      "learning_rate": 2.769336735532068e-06,
      "loss": 0.0228,
      "step": 7635
    },
    {
      "epoch": 2.0477339769375167,
      "grad_norm": 0.29592878171203374,
      "learning_rate": 2.7679405284634776e-06,
      "loss": 0.0147,
      "step": 7636
    },
    {
      "epoch": 2.048002145347278,
      "grad_norm": 0.21991749633550534,
      "learning_rate": 2.766544538724588e-06,
      "loss": 0.0121,
      "step": 7637
    },
    {
      "epoch": 2.0482703137570395,
      "grad_norm": 0.29130025046941943,
      "learning_rate": 2.7651487664513276e-06,
      "loss": 0.021,
      "step": 7638
    },
    {
      "epoch": 2.048538482166801,
      "grad_norm": 0.3385503778050744,
      "learning_rate": 2.7637532117795997e-06,
      "loss": 0.0177,
      "step": 7639
    },
    {
      "epoch": 2.0488066505765623,
      "grad_norm": 0.2012070288436613,
      "learning_rate": 2.762357874845283e-06,
      "loss": 0.012,
      "step": 7640
    },
    {
      "epoch": 2.0490748189863233,
      "grad_norm": 0.2794097486466313,
      "learning_rate": 2.760962755784241e-06,
      "loss": 0.0133,
      "step": 7641
    },
    {
      "epoch": 2.0493429873960847,
      "grad_norm": 0.2128953119863023,
      "learning_rate": 2.7595678547323086e-06,
      "loss": 0.012,
      "step": 7642
    },
    {
      "epoch": 2.049611155805846,
      "grad_norm": 0.2769894902868293,
      "learning_rate": 2.758173171825308e-06,
      "loss": 0.0198,
      "step": 7643
    },
    {
      "epoch": 2.0498793242156075,
      "grad_norm": 0.2266302090848311,
      "learning_rate": 2.756778707199032e-06,
      "loss": 0.0133,
      "step": 7644
    },
    {
      "epoch": 2.050147492625369,
      "grad_norm": 0.21090134476753206,
      "learning_rate": 2.7553844609892577e-06,
      "loss": 0.0112,
      "step": 7645
    },
    {
      "epoch": 2.05041566103513,
      "grad_norm": 0.29409961820948194,
      "learning_rate": 2.7539904333317414e-06,
      "loss": 0.0238,
      "step": 7646
    },
    {
      "epoch": 2.0506838294448912,
      "grad_norm": 0.1719398691378149,
      "learning_rate": 2.7525966243622105e-06,
      "loss": 0.0085,
      "step": 7647
    },
    {
      "epoch": 2.0509519978546527,
      "grad_norm": 0.2694756083312958,
      "learning_rate": 2.751203034216384e-06,
      "loss": 0.02,
      "step": 7648
    },
    {
      "epoch": 2.051220166264414,
      "grad_norm": 0.21639886610137676,
      "learning_rate": 2.7498096630299442e-06,
      "loss": 0.0178,
      "step": 7649
    },
    {
      "epoch": 2.0514883346741755,
      "grad_norm": 0.25489676965159835,
      "learning_rate": 2.748416510938563e-06,
      "loss": 0.0128,
      "step": 7650
    },
    {
      "epoch": 2.051756503083937,
      "grad_norm": 0.2179017758461054,
      "learning_rate": 2.7470235780778905e-06,
      "loss": 0.0107,
      "step": 7651
    },
    {
      "epoch": 2.052024671493698,
      "grad_norm": 0.2214242766125523,
      "learning_rate": 2.7456308645835483e-06,
      "loss": 0.0169,
      "step": 7652
    },
    {
      "epoch": 2.0522928399034592,
      "grad_norm": 0.21097792498119794,
      "learning_rate": 2.7442383705911455e-06,
      "loss": 0.0133,
      "step": 7653
    },
    {
      "epoch": 2.0525610083132206,
      "grad_norm": 0.22408887807779684,
      "learning_rate": 2.7428460962362618e-06,
      "loss": 0.0133,
      "step": 7654
    },
    {
      "epoch": 2.052829176722982,
      "grad_norm": 0.30138300679382246,
      "learning_rate": 2.7414540416544612e-06,
      "loss": 0.0206,
      "step": 7655
    },
    {
      "epoch": 2.0530973451327434,
      "grad_norm": 0.22421951874922283,
      "learning_rate": 2.7400622069812866e-06,
      "loss": 0.0232,
      "step": 7656
    },
    {
      "epoch": 2.053365513542505,
      "grad_norm": 0.2119899403753158,
      "learning_rate": 2.7386705923522515e-06,
      "loss": 0.0161,
      "step": 7657
    },
    {
      "epoch": 2.053633681952266,
      "grad_norm": 0.2262296328539605,
      "learning_rate": 2.7372791979028607e-06,
      "loss": 0.015,
      "step": 7658
    },
    {
      "epoch": 2.053901850362027,
      "grad_norm": 0.31485860418057443,
      "learning_rate": 2.7358880237685844e-06,
      "loss": 0.0179,
      "step": 7659
    },
    {
      "epoch": 2.0541700187717886,
      "grad_norm": 0.252466256203146,
      "learning_rate": 2.7344970700848816e-06,
      "loss": 0.0152,
      "step": 7660
    },
    {
      "epoch": 2.05443818718155,
      "grad_norm": 0.1868073987475624,
      "learning_rate": 2.7331063369871845e-06,
      "loss": 0.0102,
      "step": 7661
    },
    {
      "epoch": 2.0547063555913114,
      "grad_norm": 0.2619938090491766,
      "learning_rate": 2.7317158246109037e-06,
      "loss": 0.0254,
      "step": 7662
    },
    {
      "epoch": 2.054974524001073,
      "grad_norm": 0.31282357805329125,
      "learning_rate": 2.7303255330914322e-06,
      "loss": 0.0203,
      "step": 7663
    },
    {
      "epoch": 2.055242692410834,
      "grad_norm": 0.18330107006077012,
      "learning_rate": 2.728935462564136e-06,
      "loss": 0.007,
      "step": 7664
    },
    {
      "epoch": 2.055510860820595,
      "grad_norm": 0.26481260107731464,
      "learning_rate": 2.7275456131643662e-06,
      "loss": 0.0141,
      "step": 7665
    },
    {
      "epoch": 2.0557790292303566,
      "grad_norm": 0.23886668223133986,
      "learning_rate": 2.7261559850274444e-06,
      "loss": 0.0122,
      "step": 7666
    },
    {
      "epoch": 2.056047197640118,
      "grad_norm": 0.2709176131999788,
      "learning_rate": 2.724766578288678e-06,
      "loss": 0.0138,
      "step": 7667
    },
    {
      "epoch": 2.0563153660498794,
      "grad_norm": 0.19414494778516486,
      "learning_rate": 2.7233773930833507e-06,
      "loss": 0.0104,
      "step": 7668
    },
    {
      "epoch": 2.056583534459641,
      "grad_norm": 0.28984598341842055,
      "learning_rate": 2.72198842954672e-06,
      "loss": 0.0236,
      "step": 7669
    },
    {
      "epoch": 2.0568517028694018,
      "grad_norm": 0.2601846943539223,
      "learning_rate": 2.7205996878140305e-06,
      "loss": 0.016,
      "step": 7670
    },
    {
      "epoch": 2.057119871279163,
      "grad_norm": 0.7737512536214434,
      "learning_rate": 2.7192111680204957e-06,
      "loss": 0.0156,
      "step": 7671
    },
    {
      "epoch": 2.0573880396889246,
      "grad_norm": 0.3452347217588337,
      "learning_rate": 2.7178228703013165e-06,
      "loss": 0.0109,
      "step": 7672
    },
    {
      "epoch": 2.057656208098686,
      "grad_norm": 0.2894236150007278,
      "learning_rate": 2.7164347947916647e-06,
      "loss": 0.0179,
      "step": 7673
    },
    {
      "epoch": 2.0579243765084474,
      "grad_norm": 0.2723769176509527,
      "learning_rate": 2.7150469416266923e-06,
      "loss": 0.017,
      "step": 7674
    },
    {
      "epoch": 2.058192544918209,
      "grad_norm": 0.9211103277351875,
      "learning_rate": 2.7136593109415342e-06,
      "loss": 0.0263,
      "step": 7675
    },
    {
      "epoch": 2.0584607133279698,
      "grad_norm": 0.29048002427244957,
      "learning_rate": 2.712271902871298e-06,
      "loss": 0.0197,
      "step": 7676
    },
    {
      "epoch": 2.058728881737731,
      "grad_norm": 0.3654829854584317,
      "learning_rate": 2.710884717551072e-06,
      "loss": 0.0113,
      "step": 7677
    },
    {
      "epoch": 2.0589970501474926,
      "grad_norm": 0.18818697340074578,
      "learning_rate": 2.7094977551159263e-06,
      "loss": 0.0118,
      "step": 7678
    },
    {
      "epoch": 2.059265218557254,
      "grad_norm": 0.32981962789332275,
      "learning_rate": 2.7081110157009005e-06,
      "loss": 0.0257,
      "step": 7679
    },
    {
      "epoch": 2.0595333869670154,
      "grad_norm": 0.30327205015097103,
      "learning_rate": 2.7067244994410224e-06,
      "loss": 0.0196,
      "step": 7680
    },
    {
      "epoch": 2.059801555376777,
      "grad_norm": 0.30993515510548025,
      "learning_rate": 2.705338206471289e-06,
      "loss": 0.017,
      "step": 7681
    },
    {
      "epoch": 2.0600697237865377,
      "grad_norm": 0.2639664835435476,
      "learning_rate": 2.703952136926684e-06,
      "loss": 0.0128,
      "step": 7682
    },
    {
      "epoch": 2.060337892196299,
      "grad_norm": 0.24405637754774448,
      "learning_rate": 2.7025662909421625e-06,
      "loss": 0.0148,
      "step": 7683
    },
    {
      "epoch": 2.0606060606060606,
      "grad_norm": 0.2474789652189972,
      "learning_rate": 2.7011806686526633e-06,
      "loss": 0.0126,
      "step": 7684
    },
    {
      "epoch": 2.060874229015822,
      "grad_norm": 0.280570317037517,
      "learning_rate": 2.6997952701930985e-06,
      "loss": 0.0169,
      "step": 7685
    },
    {
      "epoch": 2.0611423974255834,
      "grad_norm": 0.30008611115585604,
      "learning_rate": 2.6984100956983606e-06,
      "loss": 0.0137,
      "step": 7686
    },
    {
      "epoch": 2.0614105658353448,
      "grad_norm": 0.2377929929992578,
      "learning_rate": 2.69702514530332e-06,
      "loss": 0.0132,
      "step": 7687
    },
    {
      "epoch": 2.0616787342451057,
      "grad_norm": 0.40781943512293817,
      "learning_rate": 2.6956404191428287e-06,
      "loss": 0.0139,
      "step": 7688
    },
    {
      "epoch": 2.061946902654867,
      "grad_norm": 0.2248374944241652,
      "learning_rate": 2.6942559173517104e-06,
      "loss": 0.012,
      "step": 7689
    },
    {
      "epoch": 2.0622150710646285,
      "grad_norm": 0.22354030588739984,
      "learning_rate": 2.6928716400647736e-06,
      "loss": 0.0118,
      "step": 7690
    },
    {
      "epoch": 2.06248323947439,
      "grad_norm": 0.2739315495654821,
      "learning_rate": 2.6914875874167978e-06,
      "loss": 0.0177,
      "step": 7691
    },
    {
      "epoch": 2.0627514078841513,
      "grad_norm": 0.26515409843676874,
      "learning_rate": 2.6901037595425484e-06,
      "loss": 0.0176,
      "step": 7692
    },
    {
      "epoch": 2.0630195762939127,
      "grad_norm": 0.282707264927694,
      "learning_rate": 2.6887201565767617e-06,
      "loss": 0.013,
      "step": 7693
    },
    {
      "epoch": 2.0632877447036737,
      "grad_norm": 0.260192818733073,
      "learning_rate": 2.687336778654157e-06,
      "loss": 0.0165,
      "step": 7694
    },
    {
      "epoch": 2.063555913113435,
      "grad_norm": 0.20039845935710288,
      "learning_rate": 2.685953625909432e-06,
      "loss": 0.0139,
      "step": 7695
    },
    {
      "epoch": 2.0638240815231965,
      "grad_norm": 0.27477559579936767,
      "learning_rate": 2.684570698477259e-06,
      "loss": 0.0165,
      "step": 7696
    },
    {
      "epoch": 2.064092249932958,
      "grad_norm": 0.19862559956921383,
      "learning_rate": 2.6831879964922876e-06,
      "loss": 0.0109,
      "step": 7697
    },
    {
      "epoch": 2.0643604183427193,
      "grad_norm": 0.3618500579923991,
      "learning_rate": 2.6818055200891515e-06,
      "loss": 0.0109,
      "step": 7698
    },
    {
      "epoch": 2.0646285867524807,
      "grad_norm": 0.26668207173788555,
      "learning_rate": 2.6804232694024557e-06,
      "loss": 0.015,
      "step": 7699
    },
    {
      "epoch": 2.0648967551622417,
      "grad_norm": 0.23372684348939288,
      "learning_rate": 2.67904124456679e-06,
      "loss": 0.0107,
      "step": 7700
    },
    {
      "epoch": 2.065164923572003,
      "grad_norm": 0.19300226494686745,
      "learning_rate": 2.677659445716714e-06,
      "loss": 0.0078,
      "step": 7701
    },
    {
      "epoch": 2.0654330919817645,
      "grad_norm": 0.17021524042968345,
      "learning_rate": 2.6762778729867738e-06,
      "loss": 0.0122,
      "step": 7702
    },
    {
      "epoch": 2.065701260391526,
      "grad_norm": 0.43643968464307215,
      "learning_rate": 2.6748965265114862e-06,
      "loss": 0.0166,
      "step": 7703
    },
    {
      "epoch": 2.0659694288012873,
      "grad_norm": 0.494705108252885,
      "learning_rate": 2.673515406425351e-06,
      "loss": 0.0166,
      "step": 7704
    },
    {
      "epoch": 2.0662375972110487,
      "grad_norm": 0.2141935557451624,
      "learning_rate": 2.672134512862845e-06,
      "loss": 0.0141,
      "step": 7705
    },
    {
      "epoch": 2.0665057656208097,
      "grad_norm": 0.36310639989559573,
      "learning_rate": 2.6707538459584204e-06,
      "loss": 0.0259,
      "step": 7706
    },
    {
      "epoch": 2.066773934030571,
      "grad_norm": 0.24215238697875252,
      "learning_rate": 2.6693734058465105e-06,
      "loss": 0.0174,
      "step": 7707
    },
    {
      "epoch": 2.0670421024403325,
      "grad_norm": 0.21232332047147068,
      "learning_rate": 2.6679931926615255e-06,
      "loss": 0.012,
      "step": 7708
    },
    {
      "epoch": 2.067310270850094,
      "grad_norm": 0.23199012496554883,
      "learning_rate": 2.6666132065378493e-06,
      "loss": 0.0167,
      "step": 7709
    },
    {
      "epoch": 2.0675784392598553,
      "grad_norm": 0.1840605866346855,
      "learning_rate": 2.6652334476098517e-06,
      "loss": 0.0113,
      "step": 7710
    },
    {
      "epoch": 2.0678466076696167,
      "grad_norm": 0.33176599349519953,
      "learning_rate": 2.663853916011874e-06,
      "loss": 0.0133,
      "step": 7711
    },
    {
      "epoch": 2.0681147760793777,
      "grad_norm": 0.21128316165982455,
      "learning_rate": 2.662474611878239e-06,
      "loss": 0.0117,
      "step": 7712
    },
    {
      "epoch": 2.068382944489139,
      "grad_norm": 0.2357375298379638,
      "learning_rate": 2.6610955353432434e-06,
      "loss": 0.0137,
      "step": 7713
    },
    {
      "epoch": 2.0686511128989005,
      "grad_norm": 0.2154175164718772,
      "learning_rate": 2.6597166865411663e-06,
      "loss": 0.0155,
      "step": 7714
    },
    {
      "epoch": 2.068919281308662,
      "grad_norm": 0.23370564422429574,
      "learning_rate": 2.6583380656062633e-06,
      "loss": 0.014,
      "step": 7715
    },
    {
      "epoch": 2.0691874497184233,
      "grad_norm": 0.2846005711439057,
      "learning_rate": 2.6569596726727644e-06,
      "loss": 0.016,
      "step": 7716
    },
    {
      "epoch": 2.0694556181281847,
      "grad_norm": 0.2309099498552967,
      "learning_rate": 2.655581507874884e-06,
      "loss": 0.0117,
      "step": 7717
    },
    {
      "epoch": 2.0697237865379456,
      "grad_norm": 0.2481727782034555,
      "learning_rate": 2.6542035713468044e-06,
      "loss": 0.0168,
      "step": 7718
    },
    {
      "epoch": 2.069991954947707,
      "grad_norm": 0.21572005422211615,
      "learning_rate": 2.652825863222698e-06,
      "loss": 0.0119,
      "step": 7719
    },
    {
      "epoch": 2.0702601233574685,
      "grad_norm": 0.4252085788635779,
      "learning_rate": 2.6514483836367032e-06,
      "loss": 0.0162,
      "step": 7720
    },
    {
      "epoch": 2.07052829176723,
      "grad_norm": 0.25702842832019757,
      "learning_rate": 2.6500711327229466e-06,
      "loss": 0.0124,
      "step": 7721
    },
    {
      "epoch": 2.0707964601769913,
      "grad_norm": 0.2369580040043926,
      "learning_rate": 2.648694110615525e-06,
      "loss": 0.0158,
      "step": 7722
    },
    {
      "epoch": 2.0710646285867527,
      "grad_norm": 0.25925912573718557,
      "learning_rate": 2.647317317448513e-06,
      "loss": 0.0142,
      "step": 7723
    },
    {
      "epoch": 2.0713327969965136,
      "grad_norm": 0.26507247157478286,
      "learning_rate": 2.645940753355967e-06,
      "loss": 0.0156,
      "step": 7724
    },
    {
      "epoch": 2.071600965406275,
      "grad_norm": 0.28349726846616613,
      "learning_rate": 2.6445644184719227e-06,
      "loss": 0.0168,
      "step": 7725
    },
    {
      "epoch": 2.0718691338160364,
      "grad_norm": 0.24756852377775093,
      "learning_rate": 2.6431883129303847e-06,
      "loss": 0.0155,
      "step": 7726
    },
    {
      "epoch": 2.072137302225798,
      "grad_norm": 0.22453952784582623,
      "learning_rate": 2.6418124368653463e-06,
      "loss": 0.018,
      "step": 7727
    },
    {
      "epoch": 2.0724054706355592,
      "grad_norm": 0.3895212167507296,
      "learning_rate": 2.6404367904107674e-06,
      "loss": 0.0206,
      "step": 7728
    },
    {
      "epoch": 2.0726736390453206,
      "grad_norm": 0.20973319491954764,
      "learning_rate": 2.6390613737005964e-06,
      "loss": 0.0101,
      "step": 7729
    },
    {
      "epoch": 2.0729418074550816,
      "grad_norm": 0.2724216918795566,
      "learning_rate": 2.6376861868687486e-06,
      "loss": 0.0182,
      "step": 7730
    },
    {
      "epoch": 2.073209975864843,
      "grad_norm": 0.241944237983217,
      "learning_rate": 2.636311230049125e-06,
      "loss": 0.0142,
      "step": 7731
    },
    {
      "epoch": 2.0734781442746044,
      "grad_norm": 0.2454953485204207,
      "learning_rate": 2.6349365033756036e-06,
      "loss": 0.0159,
      "step": 7732
    },
    {
      "epoch": 2.073746312684366,
      "grad_norm": 0.30023736463301626,
      "learning_rate": 2.6335620069820357e-06,
      "loss": 0.0141,
      "step": 7733
    },
    {
      "epoch": 2.0740144810941272,
      "grad_norm": 0.22403473353155562,
      "learning_rate": 2.632187741002252e-06,
      "loss": 0.0108,
      "step": 7734
    },
    {
      "epoch": 2.0742826495038886,
      "grad_norm": 0.2822306552643744,
      "learning_rate": 2.6308137055700607e-06,
      "loss": 0.0153,
      "step": 7735
    },
    {
      "epoch": 2.0745508179136496,
      "grad_norm": 0.2082721746204847,
      "learning_rate": 2.6294399008192482e-06,
      "loss": 0.0143,
      "step": 7736
    },
    {
      "epoch": 2.074818986323411,
      "grad_norm": 0.23018017849928601,
      "learning_rate": 2.6280663268835806e-06,
      "loss": 0.0136,
      "step": 7737
    },
    {
      "epoch": 2.0750871547331724,
      "grad_norm": 0.2164380704003625,
      "learning_rate": 2.626692983896796e-06,
      "loss": 0.0159,
      "step": 7738
    },
    {
      "epoch": 2.075355323142934,
      "grad_norm": 0.24478738378930645,
      "learning_rate": 2.6253198719926153e-06,
      "loss": 0.019,
      "step": 7739
    },
    {
      "epoch": 2.075623491552695,
      "grad_norm": 0.22315854515723202,
      "learning_rate": 2.623946991304733e-06,
      "loss": 0.0243,
      "step": 7740
    },
    {
      "epoch": 2.0758916599624566,
      "grad_norm": 0.2656852748846713,
      "learning_rate": 2.622574341966824e-06,
      "loss": 0.0224,
      "step": 7741
    },
    {
      "epoch": 2.0761598283722176,
      "grad_norm": 0.3456167092166858,
      "learning_rate": 2.621201924112541e-06,
      "loss": 0.0208,
      "step": 7742
    },
    {
      "epoch": 2.076427996781979,
      "grad_norm": 0.20223844046148842,
      "learning_rate": 2.619829737875509e-06,
      "loss": 0.0141,
      "step": 7743
    },
    {
      "epoch": 2.0766961651917404,
      "grad_norm": 0.2576686887612501,
      "learning_rate": 2.6184577833893385e-06,
      "loss": 0.015,
      "step": 7744
    },
    {
      "epoch": 2.076964333601502,
      "grad_norm": 0.885705167326344,
      "learning_rate": 2.6170860607876103e-06,
      "loss": 0.0327,
      "step": 7745
    },
    {
      "epoch": 2.077232502011263,
      "grad_norm": 0.27918864192851117,
      "learning_rate": 2.6157145702038846e-06,
      "loss": 0.0192,
      "step": 7746
    },
    {
      "epoch": 2.0775006704210246,
      "grad_norm": 0.21790265963339858,
      "learning_rate": 2.6143433117717023e-06,
      "loss": 0.0108,
      "step": 7747
    },
    {
      "epoch": 2.0777688388307856,
      "grad_norm": 0.20373973622903138,
      "learning_rate": 2.6129722856245766e-06,
      "loss": 0.0114,
      "step": 7748
    },
    {
      "epoch": 2.078037007240547,
      "grad_norm": 0.2104928684702517,
      "learning_rate": 2.611601491896003e-06,
      "loss": 0.0179,
      "step": 7749
    },
    {
      "epoch": 2.0783051756503084,
      "grad_norm": 0.21137726954524697,
      "learning_rate": 2.6102309307194496e-06,
      "loss": 0.0119,
      "step": 7750
    },
    {
      "epoch": 2.0785733440600698,
      "grad_norm": 0.2918565734931733,
      "learning_rate": 2.6088606022283657e-06,
      "loss": 0.0162,
      "step": 7751
    },
    {
      "epoch": 2.078841512469831,
      "grad_norm": 0.24273231733085882,
      "learning_rate": 2.6074905065561797e-06,
      "loss": 0.0109,
      "step": 7752
    },
    {
      "epoch": 2.0791096808795926,
      "grad_norm": 0.23991044131939565,
      "learning_rate": 2.6061206438362874e-06,
      "loss": 0.0177,
      "step": 7753
    },
    {
      "epoch": 2.0793778492893535,
      "grad_norm": 0.250030478586736,
      "learning_rate": 2.604751014202075e-06,
      "loss": 0.0106,
      "step": 7754
    },
    {
      "epoch": 2.079646017699115,
      "grad_norm": 0.23984327350975468,
      "learning_rate": 2.6033816177868954e-06,
      "loss": 0.0112,
      "step": 7755
    },
    {
      "epoch": 2.0799141861088764,
      "grad_norm": 0.31751966435235784,
      "learning_rate": 2.6020124547240864e-06,
      "loss": 0.0141,
      "step": 7756
    },
    {
      "epoch": 2.0801823545186378,
      "grad_norm": 0.2576445124129311,
      "learning_rate": 2.6006435251469585e-06,
      "loss": 0.0154,
      "step": 7757
    },
    {
      "epoch": 2.080450522928399,
      "grad_norm": 0.24406682761798773,
      "learning_rate": 2.599274829188798e-06,
      "loss": 0.0141,
      "step": 7758
    },
    {
      "epoch": 2.0807186913381606,
      "grad_norm": 0.2435202725609098,
      "learning_rate": 2.5979063669828764e-06,
      "loss": 0.0157,
      "step": 7759
    },
    {
      "epoch": 2.0809868597479215,
      "grad_norm": 0.2598945965659765,
      "learning_rate": 2.596538138662431e-06,
      "loss": 0.0156,
      "step": 7760
    },
    {
      "epoch": 2.081255028157683,
      "grad_norm": 0.24044787656469926,
      "learning_rate": 2.5951701443606894e-06,
      "loss": 0.0151,
      "step": 7761
    },
    {
      "epoch": 2.0815231965674443,
      "grad_norm": 0.2493977258919132,
      "learning_rate": 2.5938023842108432e-06,
      "loss": 0.0128,
      "step": 7762
    },
    {
      "epoch": 2.0817913649772057,
      "grad_norm": 0.23491541652871945,
      "learning_rate": 2.5924348583460714e-06,
      "loss": 0.0118,
      "step": 7763
    },
    {
      "epoch": 2.082059533386967,
      "grad_norm": 0.3087070291815123,
      "learning_rate": 2.5910675668995265e-06,
      "loss": 0.0188,
      "step": 7764
    },
    {
      "epoch": 2.0823277017967285,
      "grad_norm": 0.33557035765790544,
      "learning_rate": 2.5897005100043354e-06,
      "loss": 0.0173,
      "step": 7765
    },
    {
      "epoch": 2.0825958702064895,
      "grad_norm": 0.18503835081425637,
      "learning_rate": 2.5883336877936082e-06,
      "loss": 0.0126,
      "step": 7766
    },
    {
      "epoch": 2.082864038616251,
      "grad_norm": 0.20954249333395397,
      "learning_rate": 2.5869671004004256e-06,
      "loss": 0.0136,
      "step": 7767
    },
    {
      "epoch": 2.0831322070260123,
      "grad_norm": 0.2704965951378242,
      "learning_rate": 2.585600747957849e-06,
      "loss": 0.0176,
      "step": 7768
    },
    {
      "epoch": 2.0834003754357737,
      "grad_norm": 0.41926793949021407,
      "learning_rate": 2.5842346305989224e-06,
      "loss": 0.0149,
      "step": 7769
    },
    {
      "epoch": 2.083668543845535,
      "grad_norm": 0.2375903577275654,
      "learning_rate": 2.5828687484566516e-06,
      "loss": 0.0159,
      "step": 7770
    },
    {
      "epoch": 2.0839367122552965,
      "grad_norm": 0.19387852836069325,
      "learning_rate": 2.5815031016640356e-06,
      "loss": 0.011,
      "step": 7771
    },
    {
      "epoch": 2.0842048806650575,
      "grad_norm": 0.2839735062556969,
      "learning_rate": 2.5801376903540397e-06,
      "loss": 0.015,
      "step": 7772
    },
    {
      "epoch": 2.084473049074819,
      "grad_norm": 0.2775165273328617,
      "learning_rate": 2.578772514659612e-06,
      "loss": 0.0157,
      "step": 7773
    },
    {
      "epoch": 2.0847412174845803,
      "grad_norm": 0.27001353630556,
      "learning_rate": 2.5774075747136784e-06,
      "loss": 0.0203,
      "step": 7774
    },
    {
      "epoch": 2.0850093858943417,
      "grad_norm": 0.23540319952856875,
      "learning_rate": 2.576042870649136e-06,
      "loss": 0.0144,
      "step": 7775
    },
    {
      "epoch": 2.085277554304103,
      "grad_norm": 0.2499355909973123,
      "learning_rate": 2.574678402598865e-06,
      "loss": 0.0147,
      "step": 7776
    },
    {
      "epoch": 2.0855457227138645,
      "grad_norm": 0.22177443038973987,
      "learning_rate": 2.5733141706957164e-06,
      "loss": 0.0195,
      "step": 7777
    },
    {
      "epoch": 2.0858138911236255,
      "grad_norm": 0.24356244243561812,
      "learning_rate": 2.5719501750725275e-06,
      "loss": 0.0097,
      "step": 7778
    },
    {
      "epoch": 2.086082059533387,
      "grad_norm": 0.18332421961323714,
      "learning_rate": 2.5705864158621008e-06,
      "loss": 0.0103,
      "step": 7779
    },
    {
      "epoch": 2.0863502279431483,
      "grad_norm": 0.19608620605403743,
      "learning_rate": 2.569222893197225e-06,
      "loss": 0.011,
      "step": 7780
    },
    {
      "epoch": 2.0866183963529097,
      "grad_norm": 0.22955382461034374,
      "learning_rate": 2.567859607210667e-06,
      "loss": 0.0155,
      "step": 7781
    },
    {
      "epoch": 2.086886564762671,
      "grad_norm": 0.2637592287841738,
      "learning_rate": 2.5664965580351577e-06,
      "loss": 0.0139,
      "step": 7782
    },
    {
      "epoch": 2.0871547331724325,
      "grad_norm": 0.3404513739504032,
      "learning_rate": 2.5651337458034177e-06,
      "loss": 0.0243,
      "step": 7783
    },
    {
      "epoch": 2.0874229015821935,
      "grad_norm": 0.24505780048582096,
      "learning_rate": 2.5637711706481417e-06,
      "loss": 0.0173,
      "step": 7784
    },
    {
      "epoch": 2.087691069991955,
      "grad_norm": 0.2653853423336899,
      "learning_rate": 2.5624088327019973e-06,
      "loss": 0.0178,
      "step": 7785
    },
    {
      "epoch": 2.0879592384017163,
      "grad_norm": 0.28839193322153894,
      "learning_rate": 2.561046732097635e-06,
      "loss": 0.0205,
      "step": 7786
    },
    {
      "epoch": 2.0882274068114777,
      "grad_norm": 0.21391527893694753,
      "learning_rate": 2.559684868967676e-06,
      "loss": 0.0115,
      "step": 7787
    },
    {
      "epoch": 2.088495575221239,
      "grad_norm": 0.3067244003732284,
      "learning_rate": 2.5583232434447235e-06,
      "loss": 0.0233,
      "step": 7788
    },
    {
      "epoch": 2.0887637436310005,
      "grad_norm": 0.19904461998871822,
      "learning_rate": 2.5569618556613524e-06,
      "loss": 0.0128,
      "step": 7789
    },
    {
      "epoch": 2.0890319120407614,
      "grad_norm": 0.23003269012907868,
      "learning_rate": 2.5556007057501193e-06,
      "loss": 0.014,
      "step": 7790
    },
    {
      "epoch": 2.089300080450523,
      "grad_norm": 0.20781771652649791,
      "learning_rate": 2.5542397938435574e-06,
      "loss": 0.0108,
      "step": 7791
    },
    {
      "epoch": 2.0895682488602842,
      "grad_norm": 0.19135770209049782,
      "learning_rate": 2.552879120074172e-06,
      "loss": 0.0091,
      "step": 7792
    },
    {
      "epoch": 2.0898364172700457,
      "grad_norm": 0.20109474538050728,
      "learning_rate": 2.551518684574451e-06,
      "loss": 0.0147,
      "step": 7793
    },
    {
      "epoch": 2.090104585679807,
      "grad_norm": 0.29519919017980856,
      "learning_rate": 2.550158487476856e-06,
      "loss": 0.0203,
      "step": 7794
    },
    {
      "epoch": 2.0903727540895685,
      "grad_norm": 0.20840732632833808,
      "learning_rate": 2.548798528913822e-06,
      "loss": 0.0129,
      "step": 7795
    },
    {
      "epoch": 2.0906409224993294,
      "grad_norm": 0.26036703547348683,
      "learning_rate": 2.5474388090177703e-06,
      "loss": 0.017,
      "step": 7796
    },
    {
      "epoch": 2.090909090909091,
      "grad_norm": 0.4551663720834099,
      "learning_rate": 2.5460793279210875e-06,
      "loss": 0.0184,
      "step": 7797
    },
    {
      "epoch": 2.0911772593188522,
      "grad_norm": 0.23990134021228154,
      "learning_rate": 2.5447200857561483e-06,
      "loss": 0.0142,
      "step": 7798
    },
    {
      "epoch": 2.0914454277286136,
      "grad_norm": 0.2780961309644883,
      "learning_rate": 2.5433610826552928e-06,
      "loss": 0.0159,
      "step": 7799
    },
    {
      "epoch": 2.091713596138375,
      "grad_norm": 0.41872528793108194,
      "learning_rate": 2.542002318750847e-06,
      "loss": 0.018,
      "step": 7800
    },
    {
      "epoch": 2.0919817645481364,
      "grad_norm": 0.23145321134317562,
      "learning_rate": 2.5406437941751115e-06,
      "loss": 0.014,
      "step": 7801
    },
    {
      "epoch": 2.0922499329578974,
      "grad_norm": 0.4889226539412063,
      "learning_rate": 2.5392855090603585e-06,
      "loss": 0.0174,
      "step": 7802
    },
    {
      "epoch": 2.092518101367659,
      "grad_norm": 0.20524264001478548,
      "learning_rate": 2.537927463538844e-06,
      "loss": 0.0147,
      "step": 7803
    },
    {
      "epoch": 2.09278626977742,
      "grad_norm": 0.2479107714634576,
      "learning_rate": 2.536569657742794e-06,
      "loss": 0.0177,
      "step": 7804
    },
    {
      "epoch": 2.0930544381871816,
      "grad_norm": 0.22561911625782316,
      "learning_rate": 2.5352120918044177e-06,
      "loss": 0.0131,
      "step": 7805
    },
    {
      "epoch": 2.093322606596943,
      "grad_norm": 0.20841409737777744,
      "learning_rate": 2.5338547658558964e-06,
      "loss": 0.0133,
      "step": 7806
    },
    {
      "epoch": 2.093590775006704,
      "grad_norm": 0.28581054212687634,
      "learning_rate": 2.5324976800293878e-06,
      "loss": 0.0138,
      "step": 7807
    },
    {
      "epoch": 2.0938589434164654,
      "grad_norm": 0.25722424408873795,
      "learning_rate": 2.5311408344570296e-06,
      "loss": 0.0154,
      "step": 7808
    },
    {
      "epoch": 2.094127111826227,
      "grad_norm": 0.1909763448951231,
      "learning_rate": 2.529784229270933e-06,
      "loss": 0.012,
      "step": 7809
    },
    {
      "epoch": 2.094395280235988,
      "grad_norm": 0.2764486631129896,
      "learning_rate": 2.5284278646031866e-06,
      "loss": 0.0121,
      "step": 7810
    },
    {
      "epoch": 2.0946634486457496,
      "grad_norm": 0.2563672268585981,
      "learning_rate": 2.5270717405858595e-06,
      "loss": 0.0113,
      "step": 7811
    },
    {
      "epoch": 2.094931617055511,
      "grad_norm": 0.3380069394499706,
      "learning_rate": 2.5257158573509897e-06,
      "loss": 0.0164,
      "step": 7812
    },
    {
      "epoch": 2.0951997854652724,
      "grad_norm": 0.27771938916306876,
      "learning_rate": 2.5243602150306e-06,
      "loss": 0.0208,
      "step": 7813
    },
    {
      "epoch": 2.0954679538750334,
      "grad_norm": 0.21363875538458896,
      "learning_rate": 2.523004813756681e-06,
      "loss": 0.015,
      "step": 7814
    },
    {
      "epoch": 2.0957361222847948,
      "grad_norm": 0.25004011044349933,
      "learning_rate": 2.521649653661209e-06,
      "loss": 0.0169,
      "step": 7815
    },
    {
      "epoch": 2.096004290694556,
      "grad_norm": 0.26408537920125097,
      "learning_rate": 2.5202947348761285e-06,
      "loss": 0.0154,
      "step": 7816
    },
    {
      "epoch": 2.0962724591043176,
      "grad_norm": 0.19232667758573357,
      "learning_rate": 2.5189400575333683e-06,
      "loss": 0.0112,
      "step": 7817
    },
    {
      "epoch": 2.096540627514079,
      "grad_norm": 0.2784143252662543,
      "learning_rate": 2.5175856217648276e-06,
      "loss": 0.0133,
      "step": 7818
    },
    {
      "epoch": 2.09680879592384,
      "grad_norm": 0.30414013804970524,
      "learning_rate": 2.516231427702382e-06,
      "loss": 0.0148,
      "step": 7819
    },
    {
      "epoch": 2.0970769643336014,
      "grad_norm": 0.2594960721784442,
      "learning_rate": 2.5148774754778883e-06,
      "loss": 0.0121,
      "step": 7820
    },
    {
      "epoch": 2.0973451327433628,
      "grad_norm": 0.2412482390136821,
      "learning_rate": 2.5135237652231786e-06,
      "loss": 0.0141,
      "step": 7821
    },
    {
      "epoch": 2.097613301153124,
      "grad_norm": 0.29111093189466325,
      "learning_rate": 2.512170297070057e-06,
      "loss": 0.0187,
      "step": 7822
    },
    {
      "epoch": 2.0978814695628856,
      "grad_norm": 0.24866882541305857,
      "learning_rate": 2.5108170711503098e-06,
      "loss": 0.0196,
      "step": 7823
    },
    {
      "epoch": 2.098149637972647,
      "grad_norm": 0.21974691579932207,
      "learning_rate": 2.509464087595694e-06,
      "loss": 0.0129,
      "step": 7824
    },
    {
      "epoch": 2.0984178063824084,
      "grad_norm": 0.29530791030069214,
      "learning_rate": 2.5081113465379494e-06,
      "loss": 0.017,
      "step": 7825
    },
    {
      "epoch": 2.0986859747921693,
      "grad_norm": 0.2932522976070235,
      "learning_rate": 2.506758848108786e-06,
      "loss": 0.0158,
      "step": 7826
    },
    {
      "epoch": 2.0989541432019307,
      "grad_norm": 0.21374569472312432,
      "learning_rate": 2.5054065924398934e-06,
      "loss": 0.0117,
      "step": 7827
    },
    {
      "epoch": 2.099222311611692,
      "grad_norm": 0.2687163209847595,
      "learning_rate": 2.50405457966294e-06,
      "loss": 0.015,
      "step": 7828
    },
    {
      "epoch": 2.0994904800214536,
      "grad_norm": 0.222392673636387,
      "learning_rate": 2.5027028099095657e-06,
      "loss": 0.0123,
      "step": 7829
    },
    {
      "epoch": 2.099758648431215,
      "grad_norm": 0.238935249020314,
      "learning_rate": 2.5013512833113882e-06,
      "loss": 0.0152,
      "step": 7830
    },
    {
      "epoch": 2.100026816840976,
      "grad_norm": 0.3125364967092152,
      "learning_rate": 2.5000000000000015e-06,
      "loss": 0.0166,
      "step": 7831
    },
    {
      "epoch": 2.1002949852507373,
      "grad_norm": 0.3790112076625781,
      "learning_rate": 2.4986489601069763e-06,
      "loss": 0.02,
      "step": 7832
    },
    {
      "epoch": 2.1005631536604987,
      "grad_norm": 0.20425784330227806,
      "learning_rate": 2.4972981637638636e-06,
      "loss": 0.0108,
      "step": 7833
    },
    {
      "epoch": 2.10083132207026,
      "grad_norm": 0.2963836894867957,
      "learning_rate": 2.495947611102182e-06,
      "loss": 0.0139,
      "step": 7834
    },
    {
      "epoch": 2.1010994904800215,
      "grad_norm": 0.1973590109006194,
      "learning_rate": 2.494597302253435e-06,
      "loss": 0.0143,
      "step": 7835
    },
    {
      "epoch": 2.101367658889783,
      "grad_norm": 0.2516538588973656,
      "learning_rate": 2.4932472373490956e-06,
      "loss": 0.0167,
      "step": 7836
    },
    {
      "epoch": 2.101635827299544,
      "grad_norm": 0.19969133623237156,
      "learning_rate": 2.4918974165206167e-06,
      "loss": 0.0093,
      "step": 7837
    },
    {
      "epoch": 2.1019039957093053,
      "grad_norm": 0.32134704318520835,
      "learning_rate": 2.4905478398994297e-06,
      "loss": 0.0241,
      "step": 7838
    },
    {
      "epoch": 2.1021721641190667,
      "grad_norm": 1.138671433871642,
      "learning_rate": 2.4891985076169356e-06,
      "loss": 0.0146,
      "step": 7839
    },
    {
      "epoch": 2.102440332528828,
      "grad_norm": 0.2859464090871284,
      "learning_rate": 2.487849419804518e-06,
      "loss": 0.0153,
      "step": 7840
    },
    {
      "epoch": 2.1027085009385895,
      "grad_norm": 0.2745663932216737,
      "learning_rate": 2.4865005765935315e-06,
      "loss": 0.014,
      "step": 7841
    },
    {
      "epoch": 2.102976669348351,
      "grad_norm": 0.22531938428063855,
      "learning_rate": 2.485151978115312e-06,
      "loss": 0.0106,
      "step": 7842
    },
    {
      "epoch": 2.103244837758112,
      "grad_norm": 0.29499935625440626,
      "learning_rate": 2.483803624501168e-06,
      "loss": 0.0132,
      "step": 7843
    },
    {
      "epoch": 2.1035130061678733,
      "grad_norm": 0.23111056666336333,
      "learning_rate": 2.4824555158823826e-06,
      "loss": 0.0127,
      "step": 7844
    },
    {
      "epoch": 2.1037811745776347,
      "grad_norm": 0.447364076529288,
      "learning_rate": 2.481107652390222e-06,
      "loss": 0.0177,
      "step": 7845
    },
    {
      "epoch": 2.104049342987396,
      "grad_norm": 0.2742038452840805,
      "learning_rate": 2.4797600341559193e-06,
      "loss": 0.014,
      "step": 7846
    },
    {
      "epoch": 2.1043175113971575,
      "grad_norm": 0.24256478834088863,
      "learning_rate": 2.478412661310693e-06,
      "loss": 0.0188,
      "step": 7847
    },
    {
      "epoch": 2.104585679806919,
      "grad_norm": 0.3775046509980866,
      "learning_rate": 2.4770655339857293e-06,
      "loss": 0.027,
      "step": 7848
    },
    {
      "epoch": 2.10485384821668,
      "grad_norm": 0.19176612309853314,
      "learning_rate": 2.475718652312196e-06,
      "loss": 0.0178,
      "step": 7849
    },
    {
      "epoch": 2.1051220166264413,
      "grad_norm": 0.19988670578350395,
      "learning_rate": 2.4743720164212376e-06,
      "loss": 0.0099,
      "step": 7850
    },
    {
      "epoch": 2.1053901850362027,
      "grad_norm": 0.2910376339324989,
      "learning_rate": 2.473025626443969e-06,
      "loss": 0.0147,
      "step": 7851
    },
    {
      "epoch": 2.105658353445964,
      "grad_norm": 0.2557734102761934,
      "learning_rate": 2.4716794825114884e-06,
      "loss": 0.0177,
      "step": 7852
    },
    {
      "epoch": 2.1059265218557255,
      "grad_norm": 0.4139985571789397,
      "learning_rate": 2.4703335847548615e-06,
      "loss": 0.0163,
      "step": 7853
    },
    {
      "epoch": 2.106194690265487,
      "grad_norm": 0.21542528515995693,
      "learning_rate": 2.468987933305139e-06,
      "loss": 0.013,
      "step": 7854
    },
    {
      "epoch": 2.106462858675248,
      "grad_norm": 0.2672350685822383,
      "learning_rate": 2.4676425282933423e-06,
      "loss": 0.014,
      "step": 7855
    },
    {
      "epoch": 2.1067310270850093,
      "grad_norm": 0.3782202885623261,
      "learning_rate": 2.4662973698504678e-06,
      "loss": 0.0194,
      "step": 7856
    },
    {
      "epoch": 2.1069991954947707,
      "grad_norm": 0.20681462266289063,
      "learning_rate": 2.464952458107493e-06,
      "loss": 0.0114,
      "step": 7857
    },
    {
      "epoch": 2.107267363904532,
      "grad_norm": 0.2659457043971076,
      "learning_rate": 2.4636077931953646e-06,
      "loss": 0.0155,
      "step": 7858
    },
    {
      "epoch": 2.1075355323142935,
      "grad_norm": 0.24802988596543613,
      "learning_rate": 2.462263375245012e-06,
      "loss": 0.0156,
      "step": 7859
    },
    {
      "epoch": 2.107803700724055,
      "grad_norm": 0.22603785920846087,
      "learning_rate": 2.4609192043873387e-06,
      "loss": 0.0149,
      "step": 7860
    },
    {
      "epoch": 2.108071869133816,
      "grad_norm": 0.20330812952321403,
      "learning_rate": 2.4595752807532192e-06,
      "loss": 0.0102,
      "step": 7861
    },
    {
      "epoch": 2.1083400375435772,
      "grad_norm": 0.2908241200696346,
      "learning_rate": 2.458231604473512e-06,
      "loss": 0.0195,
      "step": 7862
    },
    {
      "epoch": 2.1086082059533386,
      "grad_norm": 0.22060566708912435,
      "learning_rate": 2.4568881756790436e-06,
      "loss": 0.0143,
      "step": 7863
    },
    {
      "epoch": 2.1088763743631,
      "grad_norm": 0.2576106355817076,
      "learning_rate": 2.455544994500621e-06,
      "loss": 0.0153,
      "step": 7864
    },
    {
      "epoch": 2.1091445427728615,
      "grad_norm": 0.19877036564384692,
      "learning_rate": 2.4542020610690293e-06,
      "loss": 0.0096,
      "step": 7865
    },
    {
      "epoch": 2.109412711182623,
      "grad_norm": 0.36707721080367367,
      "learning_rate": 2.4528593755150235e-06,
      "loss": 0.0164,
      "step": 7866
    },
    {
      "epoch": 2.109680879592384,
      "grad_norm": 0.20635686105251694,
      "learning_rate": 2.4515169379693387e-06,
      "loss": 0.0097,
      "step": 7867
    },
    {
      "epoch": 2.1099490480021452,
      "grad_norm": 0.2171341227359243,
      "learning_rate": 2.4501747485626808e-06,
      "loss": 0.0141,
      "step": 7868
    },
    {
      "epoch": 2.1102172164119066,
      "grad_norm": 0.24348098058460246,
      "learning_rate": 2.4488328074257385e-06,
      "loss": 0.0175,
      "step": 7869
    },
    {
      "epoch": 2.110485384821668,
      "grad_norm": 0.25065814847701606,
      "learning_rate": 2.447491114689174e-06,
      "loss": 0.0147,
      "step": 7870
    },
    {
      "epoch": 2.1107535532314294,
      "grad_norm": 0.2540897462396622,
      "learning_rate": 2.4461496704836212e-06,
      "loss": 0.0135,
      "step": 7871
    },
    {
      "epoch": 2.111021721641191,
      "grad_norm": 0.29617889437252837,
      "learning_rate": 2.4448084749396966e-06,
      "loss": 0.0158,
      "step": 7872
    },
    {
      "epoch": 2.111289890050952,
      "grad_norm": 0.2407685159018953,
      "learning_rate": 2.443467528187985e-06,
      "loss": 0.011,
      "step": 7873
    },
    {
      "epoch": 2.111558058460713,
      "grad_norm": 0.23089395471355048,
      "learning_rate": 2.4421268303590543e-06,
      "loss": 0.0145,
      "step": 7874
    },
    {
      "epoch": 2.1118262268704746,
      "grad_norm": 0.23309342069539835,
      "learning_rate": 2.4407863815834414e-06,
      "loss": 0.0188,
      "step": 7875
    },
    {
      "epoch": 2.112094395280236,
      "grad_norm": 0.3329747006629458,
      "learning_rate": 2.4394461819916643e-06,
      "loss": 0.0212,
      "step": 7876
    },
    {
      "epoch": 2.1123625636899974,
      "grad_norm": 0.2214556082068569,
      "learning_rate": 2.4381062317142155e-06,
      "loss": 0.0119,
      "step": 7877
    },
    {
      "epoch": 2.112630732099759,
      "grad_norm": 0.1957792309757757,
      "learning_rate": 2.436766530881562e-06,
      "loss": 0.0103,
      "step": 7878
    },
    {
      "epoch": 2.11289890050952,
      "grad_norm": 0.2780324052074412,
      "learning_rate": 2.4354270796241435e-06,
      "loss": 0.012,
      "step": 7879
    },
    {
      "epoch": 2.113167068919281,
      "grad_norm": 0.23415876658599571,
      "learning_rate": 2.434087878072384e-06,
      "loss": 0.0149,
      "step": 7880
    },
    {
      "epoch": 2.1134352373290426,
      "grad_norm": 0.2830567756037706,
      "learning_rate": 2.4327489263566735e-06,
      "loss": 0.0183,
      "step": 7881
    },
    {
      "epoch": 2.113703405738804,
      "grad_norm": 0.2626538161238261,
      "learning_rate": 2.4314102246073857e-06,
      "loss": 0.0139,
      "step": 7882
    },
    {
      "epoch": 2.1139715741485654,
      "grad_norm": 0.24968245809550474,
      "learning_rate": 2.4300717729548627e-06,
      "loss": 0.0125,
      "step": 7883
    },
    {
      "epoch": 2.114239742558327,
      "grad_norm": 0.27107280847669807,
      "learning_rate": 2.428733571529431e-06,
      "loss": 0.0147,
      "step": 7884
    },
    {
      "epoch": 2.1145079109680878,
      "grad_norm": 0.29881299769198505,
      "learning_rate": 2.427395620461382e-06,
      "loss": 0.017,
      "step": 7885
    },
    {
      "epoch": 2.114776079377849,
      "grad_norm": 0.2792013123844687,
      "learning_rate": 2.4260579198809912e-06,
      "loss": 0.0158,
      "step": 7886
    },
    {
      "epoch": 2.1150442477876106,
      "grad_norm": 0.25540705407203057,
      "learning_rate": 2.42472046991851e-06,
      "loss": 0.0164,
      "step": 7887
    },
    {
      "epoch": 2.115312416197372,
      "grad_norm": 0.25699479520029667,
      "learning_rate": 2.4233832707041567e-06,
      "loss": 0.0197,
      "step": 7888
    },
    {
      "epoch": 2.1155805846071334,
      "grad_norm": 0.45805036653003944,
      "learning_rate": 2.4220463223681357e-06,
      "loss": 0.0285,
      "step": 7889
    },
    {
      "epoch": 2.115848753016895,
      "grad_norm": 0.20084173007950878,
      "learning_rate": 2.42070962504062e-06,
      "loss": 0.011,
      "step": 7890
    },
    {
      "epoch": 2.1161169214266558,
      "grad_norm": 0.2367780411226924,
      "learning_rate": 2.419373178851758e-06,
      "loss": 0.014,
      "step": 7891
    },
    {
      "epoch": 2.116385089836417,
      "grad_norm": 0.2179794871607164,
      "learning_rate": 2.4180369839316803e-06,
      "loss": 0.0154,
      "step": 7892
    },
    {
      "epoch": 2.1166532582461786,
      "grad_norm": 0.18526240429238738,
      "learning_rate": 2.4167010404104845e-06,
      "loss": 0.0144,
      "step": 7893
    },
    {
      "epoch": 2.11692142665594,
      "grad_norm": 0.2225525873483801,
      "learning_rate": 2.415365348418252e-06,
      "loss": 0.0103,
      "step": 7894
    },
    {
      "epoch": 2.1171895950657014,
      "grad_norm": 0.2237395291312261,
      "learning_rate": 2.4140299080850317e-06,
      "loss": 0.0174,
      "step": 7895
    },
    {
      "epoch": 2.1174577634754628,
      "grad_norm": 0.20714817735279512,
      "learning_rate": 2.4126947195408528e-06,
      "loss": 0.0108,
      "step": 7896
    },
    {
      "epoch": 2.1177259318852237,
      "grad_norm": 0.21782107388237565,
      "learning_rate": 2.4113597829157226e-06,
      "loss": 0.0112,
      "step": 7897
    },
    {
      "epoch": 2.117994100294985,
      "grad_norm": 0.23074007462769786,
      "learning_rate": 2.4100250983396157e-06,
      "loss": 0.0127,
      "step": 7898
    },
    {
      "epoch": 2.1182622687047465,
      "grad_norm": 0.27758486052621467,
      "learning_rate": 2.4086906659424904e-06,
      "loss": 0.0089,
      "step": 7899
    },
    {
      "epoch": 2.118530437114508,
      "grad_norm": 0.3587687430402201,
      "learning_rate": 2.407356485854273e-06,
      "loss": 0.0125,
      "step": 7900
    },
    {
      "epoch": 2.1187986055242694,
      "grad_norm": 0.2513200682101335,
      "learning_rate": 2.4060225582048734e-06,
      "loss": 0.0154,
      "step": 7901
    },
    {
      "epoch": 2.1190667739340308,
      "grad_norm": 0.27459538461571376,
      "learning_rate": 2.4046888831241704e-06,
      "loss": 0.0201,
      "step": 7902
    },
    {
      "epoch": 2.1193349423437917,
      "grad_norm": 0.18576660592158561,
      "learning_rate": 2.4033554607420186e-06,
      "loss": 0.0097,
      "step": 7903
    },
    {
      "epoch": 2.119603110753553,
      "grad_norm": 0.2996524996532301,
      "learning_rate": 2.4020222911882536e-06,
      "loss": 0.0148,
      "step": 7904
    },
    {
      "epoch": 2.1198712791633145,
      "grad_norm": 0.222826924939333,
      "learning_rate": 2.4006893745926786e-06,
      "loss": 0.0113,
      "step": 7905
    },
    {
      "epoch": 2.120139447573076,
      "grad_norm": 0.20557119178983438,
      "learning_rate": 2.399356711085078e-06,
      "loss": 0.0127,
      "step": 7906
    },
    {
      "epoch": 2.1204076159828373,
      "grad_norm": 0.24452158468071297,
      "learning_rate": 2.398024300795212e-06,
      "loss": 0.015,
      "step": 7907
    },
    {
      "epoch": 2.1206757843925987,
      "grad_norm": 0.32798380747471023,
      "learning_rate": 2.3966921438528095e-06,
      "loss": 0.0179,
      "step": 7908
    },
    {
      "epoch": 2.1209439528023597,
      "grad_norm": 0.2233215998080106,
      "learning_rate": 2.395360240387584e-06,
      "loss": 0.0127,
      "step": 7909
    },
    {
      "epoch": 2.121212121212121,
      "grad_norm": 0.28499317142519504,
      "learning_rate": 2.394028590529214e-06,
      "loss": 0.0156,
      "step": 7910
    },
    {
      "epoch": 2.1214802896218825,
      "grad_norm": 0.21155333048068273,
      "learning_rate": 2.392697194407363e-06,
      "loss": 0.0088,
      "step": 7911
    },
    {
      "epoch": 2.121748458031644,
      "grad_norm": 0.19665835875003398,
      "learning_rate": 2.391366052151663e-06,
      "loss": 0.0112,
      "step": 7912
    },
    {
      "epoch": 2.1220166264414053,
      "grad_norm": 0.17652852110713374,
      "learning_rate": 2.3900351638917243e-06,
      "loss": 0.008,
      "step": 7913
    },
    {
      "epoch": 2.1222847948511667,
      "grad_norm": 0.24171013199667543,
      "learning_rate": 2.388704529757136e-06,
      "loss": 0.012,
      "step": 7914
    },
    {
      "epoch": 2.1225529632609277,
      "grad_norm": 0.29826887274469777,
      "learning_rate": 2.387374149877451e-06,
      "loss": 0.0152,
      "step": 7915
    },
    {
      "epoch": 2.122821131670689,
      "grad_norm": 0.3003719743003063,
      "learning_rate": 2.3860440243822085e-06,
      "loss": 0.0174,
      "step": 7916
    },
    {
      "epoch": 2.1230893000804505,
      "grad_norm": 0.2832756450107226,
      "learning_rate": 2.384714153400922e-06,
      "loss": 0.019,
      "step": 7917
    },
    {
      "epoch": 2.123357468490212,
      "grad_norm": 0.30385719804939426,
      "learning_rate": 2.3833845370630727e-06,
      "loss": 0.0197,
      "step": 7918
    },
    {
      "epoch": 2.1236256368999733,
      "grad_norm": 0.2564785470875464,
      "learning_rate": 2.382055175498126e-06,
      "loss": 0.0175,
      "step": 7919
    },
    {
      "epoch": 2.1238938053097347,
      "grad_norm": 0.27910507181896177,
      "learning_rate": 2.380726068835514e-06,
      "loss": 0.0164,
      "step": 7920
    },
    {
      "epoch": 2.1241619737194957,
      "grad_norm": 0.44421601596509386,
      "learning_rate": 2.379397217204653e-06,
      "loss": 0.0133,
      "step": 7921
    },
    {
      "epoch": 2.124430142129257,
      "grad_norm": 0.1992449718667111,
      "learning_rate": 2.378068620734926e-06,
      "loss": 0.0104,
      "step": 7922
    },
    {
      "epoch": 2.1246983105390185,
      "grad_norm": 0.25798960941600607,
      "learning_rate": 2.3767402795556953e-06,
      "loss": 0.0111,
      "step": 7923
    },
    {
      "epoch": 2.12496647894878,
      "grad_norm": 0.3322222649260662,
      "learning_rate": 2.375412193796301e-06,
      "loss": 0.0142,
      "step": 7924
    },
    {
      "epoch": 2.1252346473585413,
      "grad_norm": 0.33046065382713263,
      "learning_rate": 2.3740843635860515e-06,
      "loss": 0.0148,
      "step": 7925
    },
    {
      "epoch": 2.1255028157683027,
      "grad_norm": 0.2904105108956896,
      "learning_rate": 2.3727567890542396e-06,
      "loss": 0.018,
      "step": 7926
    },
    {
      "epoch": 2.1257709841780636,
      "grad_norm": 0.3750946399741476,
      "learning_rate": 2.3714294703301203e-06,
      "loss": 0.015,
      "step": 7927
    },
    {
      "epoch": 2.126039152587825,
      "grad_norm": 0.22133139515784714,
      "learning_rate": 2.3701024075429353e-06,
      "loss": 0.0136,
      "step": 7928
    },
    {
      "epoch": 2.1263073209975865,
      "grad_norm": 0.2962988690057557,
      "learning_rate": 2.3687756008218978e-06,
      "loss": 0.0176,
      "step": 7929
    },
    {
      "epoch": 2.126575489407348,
      "grad_norm": 0.23685636651246786,
      "learning_rate": 2.367449050296193e-06,
      "loss": 0.0148,
      "step": 7930
    },
    {
      "epoch": 2.1268436578171093,
      "grad_norm": 0.2779211132296686,
      "learning_rate": 2.3661227560949855e-06,
      "loss": 0.0193,
      "step": 7931
    },
    {
      "epoch": 2.1271118262268707,
      "grad_norm": 0.28342744533710484,
      "learning_rate": 2.3647967183474118e-06,
      "loss": 0.0194,
      "step": 7932
    },
    {
      "epoch": 2.1273799946366316,
      "grad_norm": 0.31928777118279994,
      "learning_rate": 2.363470937182585e-06,
      "loss": 0.0117,
      "step": 7933
    },
    {
      "epoch": 2.127648163046393,
      "grad_norm": 0.28861081928734145,
      "learning_rate": 2.362145412729595e-06,
      "loss": 0.0129,
      "step": 7934
    },
    {
      "epoch": 2.1279163314561544,
      "grad_norm": 0.20055476340124134,
      "learning_rate": 2.3608201451175004e-06,
      "loss": 0.0091,
      "step": 7935
    },
    {
      "epoch": 2.128184499865916,
      "grad_norm": 0.16959510091002783,
      "learning_rate": 2.3594951344753443e-06,
      "loss": 0.008,
      "step": 7936
    },
    {
      "epoch": 2.1284526682756773,
      "grad_norm": 0.32126133917708066,
      "learning_rate": 2.3581703809321338e-06,
      "loss": 0.0175,
      "step": 7937
    },
    {
      "epoch": 2.1287208366854387,
      "grad_norm": 0.4955649652298568,
      "learning_rate": 2.356845884616862e-06,
      "loss": 0.0217,
      "step": 7938
    },
    {
      "epoch": 2.1289890050951996,
      "grad_norm": 0.5623919626805461,
      "learning_rate": 2.3555216456584883e-06,
      "loss": 0.0334,
      "step": 7939
    },
    {
      "epoch": 2.129257173504961,
      "grad_norm": 0.2709677742092666,
      "learning_rate": 2.3541976641859494e-06,
      "loss": 0.0162,
      "step": 7940
    },
    {
      "epoch": 2.1295253419147224,
      "grad_norm": 0.30091657535775834,
      "learning_rate": 2.3528739403281616e-06,
      "loss": 0.0172,
      "step": 7941
    },
    {
      "epoch": 2.129793510324484,
      "grad_norm": 0.3937953634404663,
      "learning_rate": 2.3515504742140077e-06,
      "loss": 0.0209,
      "step": 7942
    },
    {
      "epoch": 2.1300616787342452,
      "grad_norm": 0.2676133334986401,
      "learning_rate": 2.3502272659723554e-06,
      "loss": 0.0142,
      "step": 7943
    },
    {
      "epoch": 2.1303298471440066,
      "grad_norm": 0.2446688247433252,
      "learning_rate": 2.348904315732036e-06,
      "loss": 0.0103,
      "step": 7944
    },
    {
      "epoch": 2.1305980155537676,
      "grad_norm": 0.351231810117543,
      "learning_rate": 2.3475816236218663e-06,
      "loss": 0.0201,
      "step": 7945
    },
    {
      "epoch": 2.130866183963529,
      "grad_norm": 0.32394632780185834,
      "learning_rate": 2.3462591897706327e-06,
      "loss": 0.0155,
      "step": 7946
    },
    {
      "epoch": 2.1311343523732904,
      "grad_norm": 0.2429997740011695,
      "learning_rate": 2.3449370143070948e-06,
      "loss": 0.0095,
      "step": 7947
    },
    {
      "epoch": 2.131402520783052,
      "grad_norm": 0.25634774073929834,
      "learning_rate": 2.343615097359993e-06,
      "loss": 0.0145,
      "step": 7948
    },
    {
      "epoch": 2.131670689192813,
      "grad_norm": 0.2167753492891867,
      "learning_rate": 2.3422934390580345e-06,
      "loss": 0.0109,
      "step": 7949
    },
    {
      "epoch": 2.1319388576025746,
      "grad_norm": 0.3045759647870103,
      "learning_rate": 2.340972039529909e-06,
      "loss": 0.0173,
      "step": 7950
    },
    {
      "epoch": 2.1322070260123356,
      "grad_norm": 0.5835128277045868,
      "learning_rate": 2.339650898904277e-06,
      "loss": 0.0188,
      "step": 7951
    },
    {
      "epoch": 2.132475194422097,
      "grad_norm": 0.4367412297104279,
      "learning_rate": 2.3383300173097723e-06,
      "loss": 0.0235,
      "step": 7952
    },
    {
      "epoch": 2.1327433628318584,
      "grad_norm": 0.25357733879432376,
      "learning_rate": 2.337009394875009e-06,
      "loss": 0.0162,
      "step": 7953
    },
    {
      "epoch": 2.13301153124162,
      "grad_norm": 0.22250726293856296,
      "learning_rate": 2.3356890317285687e-06,
      "loss": 0.0129,
      "step": 7954
    },
    {
      "epoch": 2.133279699651381,
      "grad_norm": 0.3782824619425818,
      "learning_rate": 2.334368927999013e-06,
      "loss": 0.0194,
      "step": 7955
    },
    {
      "epoch": 2.1335478680611426,
      "grad_norm": 0.3218885402518952,
      "learning_rate": 2.3330490838148806e-06,
      "loss": 0.0148,
      "step": 7956
    },
    {
      "epoch": 2.1338160364709036,
      "grad_norm": 0.2479602933161318,
      "learning_rate": 2.3317294993046765e-06,
      "loss": 0.0129,
      "step": 7957
    },
    {
      "epoch": 2.134084204880665,
      "grad_norm": 0.22774505833059555,
      "learning_rate": 2.330410174596888e-06,
      "loss": 0.0103,
      "step": 7958
    },
    {
      "epoch": 2.1343523732904264,
      "grad_norm": 0.3026791635196307,
      "learning_rate": 2.329091109819972e-06,
      "loss": 0.0291,
      "step": 7959
    },
    {
      "epoch": 2.134620541700188,
      "grad_norm": 0.648516341782001,
      "learning_rate": 2.327772305102365e-06,
      "loss": 0.0197,
      "step": 7960
    },
    {
      "epoch": 2.134888710109949,
      "grad_norm": 0.3180973921918388,
      "learning_rate": 2.3264537605724724e-06,
      "loss": 0.0109,
      "step": 7961
    },
    {
      "epoch": 2.1351568785197106,
      "grad_norm": 0.19322295606814613,
      "learning_rate": 2.3251354763586786e-06,
      "loss": 0.0105,
      "step": 7962
    },
    {
      "epoch": 2.1354250469294715,
      "grad_norm": 0.5242891891270779,
      "learning_rate": 2.323817452589346e-06,
      "loss": 0.0172,
      "step": 7963
    },
    {
      "epoch": 2.135693215339233,
      "grad_norm": 0.2889423331338352,
      "learning_rate": 2.322499689392799e-06,
      "loss": 0.0149,
      "step": 7964
    },
    {
      "epoch": 2.1359613837489944,
      "grad_norm": 0.27990676759463123,
      "learning_rate": 2.3211821868973487e-06,
      "loss": 0.0138,
      "step": 7965
    },
    {
      "epoch": 2.1362295521587558,
      "grad_norm": 0.30288348465110804,
      "learning_rate": 2.3198649452312783e-06,
      "loss": 0.0179,
      "step": 7966
    },
    {
      "epoch": 2.136497720568517,
      "grad_norm": 0.16490793946641785,
      "learning_rate": 2.318547964522841e-06,
      "loss": 0.0116,
      "step": 7967
    },
    {
      "epoch": 2.136765888978278,
      "grad_norm": 0.21618514503702513,
      "learning_rate": 2.317231244900271e-06,
      "loss": 0.014,
      "step": 7968
    },
    {
      "epoch": 2.1370340573880395,
      "grad_norm": 0.2245955543113481,
      "learning_rate": 2.31591478649177e-06,
      "loss": 0.0118,
      "step": 7969
    },
    {
      "epoch": 2.137302225797801,
      "grad_norm": 0.20190763900114359,
      "learning_rate": 2.3145985894255228e-06,
      "loss": 0.0116,
      "step": 7970
    },
    {
      "epoch": 2.1375703942075623,
      "grad_norm": 0.24431367920414004,
      "learning_rate": 2.313282653829679e-06,
      "loss": 0.0088,
      "step": 7971
    },
    {
      "epoch": 2.1378385626173237,
      "grad_norm": 0.225336851625098,
      "learning_rate": 2.31196697983237e-06,
      "loss": 0.0162,
      "step": 7972
    },
    {
      "epoch": 2.138106731027085,
      "grad_norm": 0.24235146202362276,
      "learning_rate": 2.3106515675617014e-06,
      "loss": 0.0148,
      "step": 7973
    },
    {
      "epoch": 2.1383748994368466,
      "grad_norm": 0.2595781778976322,
      "learning_rate": 2.3093364171457476e-06,
      "loss": 0.014,
      "step": 7974
    },
    {
      "epoch": 2.1386430678466075,
      "grad_norm": 0.20727620067291733,
      "learning_rate": 2.3080215287125655e-06,
      "loss": 0.0129,
      "step": 7975
    },
    {
      "epoch": 2.138911236256369,
      "grad_norm": 0.5119276331929072,
      "learning_rate": 2.30670690239018e-06,
      "loss": 0.0163,
      "step": 7976
    },
    {
      "epoch": 2.1391794046661303,
      "grad_norm": 0.2443581168711156,
      "learning_rate": 2.305392538306591e-06,
      "loss": 0.0142,
      "step": 7977
    },
    {
      "epoch": 2.1394475730758917,
      "grad_norm": 0.20444266188957452,
      "learning_rate": 2.3040784365897783e-06,
      "loss": 0.0133,
      "step": 7978
    },
    {
      "epoch": 2.139715741485653,
      "grad_norm": 0.3315969333253959,
      "learning_rate": 2.3027645973676894e-06,
      "loss": 0.0221,
      "step": 7979
    },
    {
      "epoch": 2.139983909895414,
      "grad_norm": 0.24608245859256378,
      "learning_rate": 2.301451020768252e-06,
      "loss": 0.0135,
      "step": 7980
    },
    {
      "epoch": 2.1402520783051755,
      "grad_norm": 0.21084460797230892,
      "learning_rate": 2.300137706919362e-06,
      "loss": 0.0146,
      "step": 7981
    },
    {
      "epoch": 2.140520246714937,
      "grad_norm": 0.39647172003924136,
      "learning_rate": 2.298824655948896e-06,
      "loss": 0.0199,
      "step": 7982
    },
    {
      "epoch": 2.1407884151246983,
      "grad_norm": 0.23181260270623663,
      "learning_rate": 2.297511867984703e-06,
      "loss": 0.0123,
      "step": 7983
    },
    {
      "epoch": 2.1410565835344597,
      "grad_norm": 0.2510461195579083,
      "learning_rate": 2.296199343154603e-06,
      "loss": 0.0134,
      "step": 7984
    },
    {
      "epoch": 2.141324751944221,
      "grad_norm": 0.255312553623265,
      "learning_rate": 2.2948870815863965e-06,
      "loss": 0.0176,
      "step": 7985
    },
    {
      "epoch": 2.1415929203539825,
      "grad_norm": 0.3179942836296037,
      "learning_rate": 2.293575083407851e-06,
      "loss": 0.0244,
      "step": 7986
    },
    {
      "epoch": 2.1418610887637435,
      "grad_norm": 0.2797568873496874,
      "learning_rate": 2.292263348746716e-06,
      "loss": 0.0211,
      "step": 7987
    },
    {
      "epoch": 2.142129257173505,
      "grad_norm": 0.23380182424407198,
      "learning_rate": 2.29095187773071e-06,
      "loss": 0.016,
      "step": 7988
    },
    {
      "epoch": 2.1423974255832663,
      "grad_norm": 0.22688390934712127,
      "learning_rate": 2.289640670487525e-06,
      "loss": 0.0129,
      "step": 7989
    },
    {
      "epoch": 2.1426655939930277,
      "grad_norm": 0.29219539544329115,
      "learning_rate": 2.288329727144834e-06,
      "loss": 0.0241,
      "step": 7990
    },
    {
      "epoch": 2.142933762402789,
      "grad_norm": 0.3458336592294186,
      "learning_rate": 2.2870190478302766e-06,
      "loss": 0.0188,
      "step": 7991
    },
    {
      "epoch": 2.14320193081255,
      "grad_norm": 0.34207450175899656,
      "learning_rate": 2.2857086326714718e-06,
      "loss": 0.0154,
      "step": 7992
    },
    {
      "epoch": 2.1434700992223115,
      "grad_norm": 0.2231714773757065,
      "learning_rate": 2.2843984817960123e-06,
      "loss": 0.0145,
      "step": 7993
    },
    {
      "epoch": 2.143738267632073,
      "grad_norm": 0.27455225666629635,
      "learning_rate": 2.2830885953314614e-06,
      "loss": 0.0165,
      "step": 7994
    },
    {
      "epoch": 2.1440064360418343,
      "grad_norm": 0.23481039828751385,
      "learning_rate": 2.2817789734053626e-06,
      "loss": 0.0128,
      "step": 7995
    },
    {
      "epoch": 2.1442746044515957,
      "grad_norm": 0.2691035594236558,
      "learning_rate": 2.2804696161452267e-06,
      "loss": 0.0129,
      "step": 7996
    },
    {
      "epoch": 2.144542772861357,
      "grad_norm": 0.24946058396114074,
      "learning_rate": 2.279160523678546e-06,
      "loss": 0.0166,
      "step": 7997
    },
    {
      "epoch": 2.1448109412711185,
      "grad_norm": 0.23980744038236382,
      "learning_rate": 2.277851696132779e-06,
      "loss": 0.0111,
      "step": 7998
    },
    {
      "epoch": 2.1450791096808794,
      "grad_norm": 0.2706526655985602,
      "learning_rate": 2.2765431336353673e-06,
      "loss": 0.0157,
      "step": 7999
    },
    {
      "epoch": 2.145347278090641,
      "grad_norm": 0.23442760280382924,
      "learning_rate": 2.27523483631372e-06,
      "loss": 0.0133,
      "step": 8000
    },
    {
      "epoch": 2.1456154465004023,
      "grad_norm": 0.24755454079045205,
      "learning_rate": 2.273926804295221e-06,
      "loss": 0.0122,
      "step": 8001
    },
    {
      "epoch": 2.1458836149101637,
      "grad_norm": 0.346717729662209,
      "learning_rate": 2.272619037707231e-06,
      "loss": 0.016,
      "step": 8002
    },
    {
      "epoch": 2.146151783319925,
      "grad_norm": 0.20501631531767148,
      "learning_rate": 2.2713115366770865e-06,
      "loss": 0.0125,
      "step": 8003
    },
    {
      "epoch": 2.146419951729686,
      "grad_norm": 0.23392242412954342,
      "learning_rate": 2.2700043013320912e-06,
      "loss": 0.0155,
      "step": 8004
    },
    {
      "epoch": 2.1466881201394474,
      "grad_norm": 0.26199723562550337,
      "learning_rate": 2.268697331799531e-06,
      "loss": 0.0184,
      "step": 8005
    },
    {
      "epoch": 2.146956288549209,
      "grad_norm": 0.29942856671827195,
      "learning_rate": 2.2673906282066582e-06,
      "loss": 0.0149,
      "step": 8006
    },
    {
      "epoch": 2.1472244569589702,
      "grad_norm": 0.1804427056572445,
      "learning_rate": 2.266084190680707e-06,
      "loss": 0.0115,
      "step": 8007
    },
    {
      "epoch": 2.1474926253687316,
      "grad_norm": 0.20302925191029045,
      "learning_rate": 2.264778019348878e-06,
      "loss": 0.0119,
      "step": 8008
    },
    {
      "epoch": 2.147760793778493,
      "grad_norm": 0.2621412986424994,
      "learning_rate": 2.263472114338351e-06,
      "loss": 0.0134,
      "step": 8009
    },
    {
      "epoch": 2.1480289621882545,
      "grad_norm": 0.21007018715899442,
      "learning_rate": 2.262166475776281e-06,
      "loss": 0.0133,
      "step": 8010
    },
    {
      "epoch": 2.1482971305980154,
      "grad_norm": 0.2674813162364021,
      "learning_rate": 2.260861103789792e-06,
      "loss": 0.0103,
      "step": 8011
    },
    {
      "epoch": 2.148565299007777,
      "grad_norm": 0.20911402696491696,
      "learning_rate": 2.2595559985059834e-06,
      "loss": 0.0136,
      "step": 8012
    },
    {
      "epoch": 2.1488334674175382,
      "grad_norm": 0.14423736370677825,
      "learning_rate": 2.2582511600519336e-06,
      "loss": 0.008,
      "step": 8013
    },
    {
      "epoch": 2.1491016358272996,
      "grad_norm": 0.2528889306427548,
      "learning_rate": 2.256946588554687e-06,
      "loss": 0.0152,
      "step": 8014
    },
    {
      "epoch": 2.149369804237061,
      "grad_norm": 0.3306481744455614,
      "learning_rate": 2.25564228414127e-06,
      "loss": 0.0161,
      "step": 8015
    },
    {
      "epoch": 2.149637972646822,
      "grad_norm": 0.2721051172761682,
      "learning_rate": 2.2543382469386766e-06,
      "loss": 0.0163,
      "step": 8016
    },
    {
      "epoch": 2.1499061410565834,
      "grad_norm": 0.2645572583329774,
      "learning_rate": 2.25303447707388e-06,
      "loss": 0.0148,
      "step": 8017
    },
    {
      "epoch": 2.150174309466345,
      "grad_norm": 0.21385219535839486,
      "learning_rate": 2.2517309746738214e-06,
      "loss": 0.0153,
      "step": 8018
    },
    {
      "epoch": 2.150442477876106,
      "grad_norm": 0.2792873192621192,
      "learning_rate": 2.250427739865421e-06,
      "loss": 0.015,
      "step": 8019
    },
    {
      "epoch": 2.1507106462858676,
      "grad_norm": 0.25895326669984053,
      "learning_rate": 2.2491247727755737e-06,
      "loss": 0.0143,
      "step": 8020
    },
    {
      "epoch": 2.150978814695629,
      "grad_norm": 0.25932999380094524,
      "learning_rate": 2.2478220735311424e-06,
      "loss": 0.013,
      "step": 8021
    },
    {
      "epoch": 2.1512469831053904,
      "grad_norm": 0.2593156958550851,
      "learning_rate": 2.24651964225897e-06,
      "loss": 0.0174,
      "step": 8022
    },
    {
      "epoch": 2.1515151515151514,
      "grad_norm": 0.2604807961204911,
      "learning_rate": 2.2452174790858706e-06,
      "loss": 0.0164,
      "step": 8023
    },
    {
      "epoch": 2.151783319924913,
      "grad_norm": 0.2493312177837043,
      "learning_rate": 2.243915584138629e-06,
      "loss": 0.0184,
      "step": 8024
    },
    {
      "epoch": 2.152051488334674,
      "grad_norm": 0.21246497462815678,
      "learning_rate": 2.2426139575440116e-06,
      "loss": 0.0082,
      "step": 8025
    },
    {
      "epoch": 2.1523196567444356,
      "grad_norm": 0.2339212278506901,
      "learning_rate": 2.2413125994287504e-06,
      "loss": 0.0144,
      "step": 8026
    },
    {
      "epoch": 2.152587825154197,
      "grad_norm": 0.17725549221880427,
      "learning_rate": 2.2400115099195596e-06,
      "loss": 0.0087,
      "step": 8027
    },
    {
      "epoch": 2.152855993563958,
      "grad_norm": 0.22399268231120895,
      "learning_rate": 2.238710689143118e-06,
      "loss": 0.0093,
      "step": 8028
    },
    {
      "epoch": 2.1531241619737194,
      "grad_norm": 0.3159453661187273,
      "learning_rate": 2.237410137226086e-06,
      "loss": 0.0153,
      "step": 8029
    },
    {
      "epoch": 2.1533923303834808,
      "grad_norm": 0.2663617673880538,
      "learning_rate": 2.236109854295096e-06,
      "loss": 0.0187,
      "step": 8030
    },
    {
      "epoch": 2.153660498793242,
      "grad_norm": 0.30266677446731355,
      "learning_rate": 2.23480984047675e-06,
      "loss": 0.0142,
      "step": 8031
    },
    {
      "epoch": 2.1539286672030036,
      "grad_norm": 0.2333768598900261,
      "learning_rate": 2.23351009589763e-06,
      "loss": 0.0142,
      "step": 8032
    },
    {
      "epoch": 2.154196835612765,
      "grad_norm": 0.271393479351756,
      "learning_rate": 2.232210620684285e-06,
      "loss": 0.0136,
      "step": 8033
    },
    {
      "epoch": 2.154465004022526,
      "grad_norm": 0.2277806590665096,
      "learning_rate": 2.2309114149632457e-06,
      "loss": 0.0127,
      "step": 8034
    },
    {
      "epoch": 2.1547331724322873,
      "grad_norm": 0.2914395079232499,
      "learning_rate": 2.22961247886101e-06,
      "loss": 0.0248,
      "step": 8035
    },
    {
      "epoch": 2.1550013408420488,
      "grad_norm": 0.3188716537596488,
      "learning_rate": 2.22831381250405e-06,
      "loss": 0.0172,
      "step": 8036
    },
    {
      "epoch": 2.15526950925181,
      "grad_norm": 0.30413745488996374,
      "learning_rate": 2.2270154160188173e-06,
      "loss": 0.0184,
      "step": 8037
    },
    {
      "epoch": 2.1555376776615716,
      "grad_norm": 0.2674757566824522,
      "learning_rate": 2.2257172895317297e-06,
      "loss": 0.0144,
      "step": 8038
    },
    {
      "epoch": 2.155805846071333,
      "grad_norm": 0.279516692796293,
      "learning_rate": 2.224419433169186e-06,
      "loss": 0.0137,
      "step": 8039
    },
    {
      "epoch": 2.156074014481094,
      "grad_norm": 0.292076871831746,
      "learning_rate": 2.223121847057551e-06,
      "loss": 0.0139,
      "step": 8040
    },
    {
      "epoch": 2.1563421828908553,
      "grad_norm": 0.27152074566572293,
      "learning_rate": 2.2218245313231695e-06,
      "loss": 0.0154,
      "step": 8041
    },
    {
      "epoch": 2.1566103513006167,
      "grad_norm": 0.2611595643577145,
      "learning_rate": 2.2205274860923593e-06,
      "loss": 0.0144,
      "step": 8042
    },
    {
      "epoch": 2.156878519710378,
      "grad_norm": 0.36880324313185925,
      "learning_rate": 2.219230711491406e-06,
      "loss": 0.013,
      "step": 8043
    },
    {
      "epoch": 2.1571466881201395,
      "grad_norm": 0.24430947261024766,
      "learning_rate": 2.217934207646578e-06,
      "loss": 0.0127,
      "step": 8044
    },
    {
      "epoch": 2.157414856529901,
      "grad_norm": 0.24704968214928016,
      "learning_rate": 2.2166379746841083e-06,
      "loss": 0.0169,
      "step": 8045
    },
    {
      "epoch": 2.157683024939662,
      "grad_norm": 0.46221212444762366,
      "learning_rate": 2.215342012730209e-06,
      "loss": 0.0193,
      "step": 8046
    },
    {
      "epoch": 2.1579511933494233,
      "grad_norm": 0.3173248004275007,
      "learning_rate": 2.214046321911068e-06,
      "loss": 0.0198,
      "step": 8047
    },
    {
      "epoch": 2.1582193617591847,
      "grad_norm": 0.2107823871214829,
      "learning_rate": 2.212750902352837e-06,
      "loss": 0.0112,
      "step": 8048
    },
    {
      "epoch": 2.158487530168946,
      "grad_norm": 0.24004260338399516,
      "learning_rate": 2.2114557541816523e-06,
      "loss": 0.018,
      "step": 8049
    },
    {
      "epoch": 2.1587556985787075,
      "grad_norm": 0.23036593168895655,
      "learning_rate": 2.2101608775236153e-06,
      "loss": 0.0147,
      "step": 8050
    },
    {
      "epoch": 2.159023866988469,
      "grad_norm": 0.2747771137780415,
      "learning_rate": 2.2088662725048073e-06,
      "loss": 0.0261,
      "step": 8051
    },
    {
      "epoch": 2.15929203539823,
      "grad_norm": 0.3259847910062458,
      "learning_rate": 2.207571939251281e-06,
      "loss": 0.0171,
      "step": 8052
    },
    {
      "epoch": 2.1595602038079913,
      "grad_norm": 0.24175376801123527,
      "learning_rate": 2.2062778778890602e-06,
      "loss": 0.019,
      "step": 8053
    },
    {
      "epoch": 2.1598283722177527,
      "grad_norm": 0.24278409968790635,
      "learning_rate": 2.204984088544147e-06,
      "loss": 0.014,
      "step": 8054
    },
    {
      "epoch": 2.160096540627514,
      "grad_norm": 0.19992885679243982,
      "learning_rate": 2.2036905713425104e-06,
      "loss": 0.0115,
      "step": 8055
    },
    {
      "epoch": 2.1603647090372755,
      "grad_norm": 0.18310941914294204,
      "learning_rate": 2.2023973264101004e-06,
      "loss": 0.009,
      "step": 8056
    },
    {
      "epoch": 2.160632877447037,
      "grad_norm": 0.22197581903653207,
      "learning_rate": 2.2011043538728344e-06,
      "loss": 0.0131,
      "step": 8057
    },
    {
      "epoch": 2.160901045856798,
      "grad_norm": 0.16658008590452722,
      "learning_rate": 2.1998116538566056e-06,
      "loss": 0.0099,
      "step": 8058
    },
    {
      "epoch": 2.1611692142665593,
      "grad_norm": 0.3532977027193072,
      "learning_rate": 2.1985192264872856e-06,
      "loss": 0.0163,
      "step": 8059
    },
    {
      "epoch": 2.1614373826763207,
      "grad_norm": 0.21228090810477732,
      "learning_rate": 2.197227071890707e-06,
      "loss": 0.0114,
      "step": 8060
    },
    {
      "epoch": 2.161705551086082,
      "grad_norm": 0.18768895141799835,
      "learning_rate": 2.1959351901926873e-06,
      "loss": 0.0077,
      "step": 8061
    },
    {
      "epoch": 2.1619737194958435,
      "grad_norm": 0.19056166787115558,
      "learning_rate": 2.1946435815190153e-06,
      "loss": 0.0091,
      "step": 8062
    },
    {
      "epoch": 2.162241887905605,
      "grad_norm": 0.2591324893431378,
      "learning_rate": 2.1933522459954483e-06,
      "loss": 0.0116,
      "step": 8063
    },
    {
      "epoch": 2.162510056315366,
      "grad_norm": 0.2882499909428108,
      "learning_rate": 2.192061183747723e-06,
      "loss": 0.0179,
      "step": 8064
    },
    {
      "epoch": 2.1627782247251273,
      "grad_norm": 0.24574804368498632,
      "learning_rate": 2.190770394901543e-06,
      "loss": 0.017,
      "step": 8065
    },
    {
      "epoch": 2.1630463931348887,
      "grad_norm": 0.23303605949339473,
      "learning_rate": 2.189479879582594e-06,
      "loss": 0.0153,
      "step": 8066
    },
    {
      "epoch": 2.16331456154465,
      "grad_norm": 0.3211862668699343,
      "learning_rate": 2.1881896379165253e-06,
      "loss": 0.0123,
      "step": 8067
    },
    {
      "epoch": 2.1635827299544115,
      "grad_norm": 0.4703572603519183,
      "learning_rate": 2.1868996700289656e-06,
      "loss": 0.0162,
      "step": 8068
    },
    {
      "epoch": 2.163850898364173,
      "grad_norm": 0.254921749153243,
      "learning_rate": 2.185609976045519e-06,
      "loss": 0.0145,
      "step": 8069
    },
    {
      "epoch": 2.164119066773934,
      "grad_norm": 0.22625737318032926,
      "learning_rate": 2.1843205560917544e-06,
      "loss": 0.0116,
      "step": 8070
    },
    {
      "epoch": 2.1643872351836952,
      "grad_norm": 0.2526296210606797,
      "learning_rate": 2.1830314102932245e-06,
      "loss": 0.0142,
      "step": 8071
    },
    {
      "epoch": 2.1646554035934567,
      "grad_norm": 0.23607886916562654,
      "learning_rate": 2.181742538775447e-06,
      "loss": 0.0134,
      "step": 8072
    },
    {
      "epoch": 2.164923572003218,
      "grad_norm": 0.25202856715482796,
      "learning_rate": 2.1804539416639144e-06,
      "loss": 0.0149,
      "step": 8073
    },
    {
      "epoch": 2.1651917404129795,
      "grad_norm": 0.20994674657045104,
      "learning_rate": 2.179165619084097e-06,
      "loss": 0.0122,
      "step": 8074
    },
    {
      "epoch": 2.165459908822741,
      "grad_norm": 0.269923987967757,
      "learning_rate": 2.177877571161433e-06,
      "loss": 0.0125,
      "step": 8075
    },
    {
      "epoch": 2.165728077232502,
      "grad_norm": 0.2191229608400655,
      "learning_rate": 2.1765897980213386e-06,
      "loss": 0.0083,
      "step": 8076
    },
    {
      "epoch": 2.1659962456422632,
      "grad_norm": 0.2612854820787141,
      "learning_rate": 2.175302299789198e-06,
      "loss": 0.0283,
      "step": 8077
    },
    {
      "epoch": 2.1662644140520246,
      "grad_norm": 0.2165316065735465,
      "learning_rate": 2.174015076590373e-06,
      "loss": 0.0103,
      "step": 8078
    },
    {
      "epoch": 2.166532582461786,
      "grad_norm": 0.15655966155581802,
      "learning_rate": 2.172728128550199e-06,
      "loss": 0.0078,
      "step": 8079
    },
    {
      "epoch": 2.1668007508715474,
      "grad_norm": 0.3866620479020673,
      "learning_rate": 2.171441455793979e-06,
      "loss": 0.0118,
      "step": 8080
    },
    {
      "epoch": 2.167068919281309,
      "grad_norm": 0.2909887317256784,
      "learning_rate": 2.1701550584469965e-06,
      "loss": 0.0131,
      "step": 8081
    },
    {
      "epoch": 2.16733708769107,
      "grad_norm": 0.27286103094111597,
      "learning_rate": 2.168868936634501e-06,
      "loss": 0.0139,
      "step": 8082
    },
    {
      "epoch": 2.167605256100831,
      "grad_norm": 0.2843428855868537,
      "learning_rate": 2.1675830904817225e-06,
      "loss": 0.021,
      "step": 8083
    },
    {
      "epoch": 2.1678734245105926,
      "grad_norm": 0.22651028098052212,
      "learning_rate": 2.166297520113859e-06,
      "loss": 0.0142,
      "step": 8084
    },
    {
      "epoch": 2.168141592920354,
      "grad_norm": 0.22142482329587104,
      "learning_rate": 2.16501222565608e-06,
      "loss": 0.0099,
      "step": 8085
    },
    {
      "epoch": 2.1684097613301154,
      "grad_norm": 0.582591718091082,
      "learning_rate": 2.1637272072335366e-06,
      "loss": 0.0111,
      "step": 8086
    },
    {
      "epoch": 2.168677929739877,
      "grad_norm": 0.24074231755190728,
      "learning_rate": 2.162442464971342e-06,
      "loss": 0.0119,
      "step": 8087
    },
    {
      "epoch": 2.168946098149638,
      "grad_norm": 0.31821855727512416,
      "learning_rate": 2.161157998994592e-06,
      "loss": 0.0174,
      "step": 8088
    },
    {
      "epoch": 2.169214266559399,
      "grad_norm": 0.2907407892989868,
      "learning_rate": 2.1598738094283522e-06,
      "loss": 0.0162,
      "step": 8089
    },
    {
      "epoch": 2.1694824349691606,
      "grad_norm": 0.35207504529931855,
      "learning_rate": 2.158589896397657e-06,
      "loss": 0.0117,
      "step": 8090
    },
    {
      "epoch": 2.169750603378922,
      "grad_norm": 0.20976647789842587,
      "learning_rate": 2.1573062600275217e-06,
      "loss": 0.011,
      "step": 8091
    },
    {
      "epoch": 2.1700187717886834,
      "grad_norm": 0.24300854195882127,
      "learning_rate": 2.1560229004429273e-06,
      "loss": 0.0156,
      "step": 8092
    },
    {
      "epoch": 2.170286940198445,
      "grad_norm": 0.2050031651045972,
      "learning_rate": 2.154739817768834e-06,
      "loss": 0.0138,
      "step": 8093
    },
    {
      "epoch": 2.1705551086082058,
      "grad_norm": 0.24676933215122548,
      "learning_rate": 2.1534570121301683e-06,
      "loss": 0.0133,
      "step": 8094
    },
    {
      "epoch": 2.170823277017967,
      "grad_norm": 0.2226896173331888,
      "learning_rate": 2.1521744836518365e-06,
      "loss": 0.0088,
      "step": 8095
    },
    {
      "epoch": 2.1710914454277286,
      "grad_norm": 0.29122906277162347,
      "learning_rate": 2.1508922324587177e-06,
      "loss": 0.0197,
      "step": 8096
    },
    {
      "epoch": 2.17135961383749,
      "grad_norm": 0.22331543555203767,
      "learning_rate": 2.149610258675655e-06,
      "loss": 0.0149,
      "step": 8097
    },
    {
      "epoch": 2.1716277822472514,
      "grad_norm": 0.2484241043775842,
      "learning_rate": 2.148328562427473e-06,
      "loss": 0.0139,
      "step": 8098
    },
    {
      "epoch": 2.171895950657013,
      "grad_norm": 0.29761822904626356,
      "learning_rate": 2.14704714383897e-06,
      "loss": 0.0194,
      "step": 8099
    },
    {
      "epoch": 2.1721641190667738,
      "grad_norm": 0.29306204871437325,
      "learning_rate": 2.1457660030349103e-06,
      "loss": 0.0216,
      "step": 8100
    },
    {
      "epoch": 2.172432287476535,
      "grad_norm": 0.21978483611076352,
      "learning_rate": 2.144485140140039e-06,
      "loss": 0.0106,
      "step": 8101
    },
    {
      "epoch": 2.1727004558862966,
      "grad_norm": 0.22035098992198182,
      "learning_rate": 2.1432045552790666e-06,
      "loss": 0.0154,
      "step": 8102
    },
    {
      "epoch": 2.172968624296058,
      "grad_norm": 0.3790008117227063,
      "learning_rate": 2.1419242485766834e-06,
      "loss": 0.0276,
      "step": 8103
    },
    {
      "epoch": 2.1732367927058194,
      "grad_norm": 0.24058145481387644,
      "learning_rate": 2.1406442201575466e-06,
      "loss": 0.0147,
      "step": 8104
    },
    {
      "epoch": 2.173504961115581,
      "grad_norm": 0.19604162265155758,
      "learning_rate": 2.1393644701462905e-06,
      "loss": 0.0141,
      "step": 8105
    },
    {
      "epoch": 2.1737731295253417,
      "grad_norm": 0.3256522277797495,
      "learning_rate": 2.1380849986675233e-06,
      "loss": 0.0221,
      "step": 8106
    },
    {
      "epoch": 2.174041297935103,
      "grad_norm": 0.2363901596272482,
      "learning_rate": 2.136805805845819e-06,
      "loss": 0.0164,
      "step": 8107
    },
    {
      "epoch": 2.1743094663448645,
      "grad_norm": 0.23017513138042064,
      "learning_rate": 2.135526891805734e-06,
      "loss": 0.0133,
      "step": 8108
    },
    {
      "epoch": 2.174577634754626,
      "grad_norm": 0.20859311870996733,
      "learning_rate": 2.134248256671791e-06,
      "loss": 0.0098,
      "step": 8109
    },
    {
      "epoch": 2.1748458031643874,
      "grad_norm": 0.22378126404571747,
      "learning_rate": 2.132969900568484e-06,
      "loss": 0.0106,
      "step": 8110
    },
    {
      "epoch": 2.1751139715741488,
      "grad_norm": 0.24085432496567885,
      "learning_rate": 2.131691823620288e-06,
      "loss": 0.0135,
      "step": 8111
    },
    {
      "epoch": 2.1753821399839097,
      "grad_norm": 0.25274668792987276,
      "learning_rate": 2.1304140259516427e-06,
      "loss": 0.0158,
      "step": 8112
    },
    {
      "epoch": 2.175650308393671,
      "grad_norm": 0.34328649584052745,
      "learning_rate": 2.129136507686967e-06,
      "loss": 0.0134,
      "step": 8113
    },
    {
      "epoch": 2.1759184768034325,
      "grad_norm": 0.23421274126571387,
      "learning_rate": 2.127859268950645e-06,
      "loss": 0.0139,
      "step": 8114
    },
    {
      "epoch": 2.176186645213194,
      "grad_norm": 0.20872766785013253,
      "learning_rate": 2.126582309867041e-06,
      "loss": 0.0127,
      "step": 8115
    },
    {
      "epoch": 2.1764548136229553,
      "grad_norm": 0.22089091539277672,
      "learning_rate": 2.1253056305604903e-06,
      "loss": 0.0121,
      "step": 8116
    },
    {
      "epoch": 2.1767229820327167,
      "grad_norm": 0.2025494617496406,
      "learning_rate": 2.124029231155297e-06,
      "loss": 0.0114,
      "step": 8117
    },
    {
      "epoch": 2.1769911504424777,
      "grad_norm": 0.2348462942934537,
      "learning_rate": 2.122753111775743e-06,
      "loss": 0.0127,
      "step": 8118
    },
    {
      "epoch": 2.177259318852239,
      "grad_norm": 0.31586737831969236,
      "learning_rate": 2.1214772725460776e-06,
      "loss": 0.0217,
      "step": 8119
    },
    {
      "epoch": 2.1775274872620005,
      "grad_norm": 0.27694363575770603,
      "learning_rate": 2.12020171359053e-06,
      "loss": 0.0169,
      "step": 8120
    },
    {
      "epoch": 2.177795655671762,
      "grad_norm": 0.2528695585686696,
      "learning_rate": 2.118926435033296e-06,
      "loss": 0.0147,
      "step": 8121
    },
    {
      "epoch": 2.1780638240815233,
      "grad_norm": 0.21992149851146608,
      "learning_rate": 2.117651436998544e-06,
      "loss": 0.0137,
      "step": 8122
    },
    {
      "epoch": 2.1783319924912847,
      "grad_norm": 0.7633392282861222,
      "learning_rate": 2.1163767196104208e-06,
      "loss": 0.0204,
      "step": 8123
    },
    {
      "epoch": 2.1786001609010457,
      "grad_norm": 0.23054099712098766,
      "learning_rate": 2.115102282993039e-06,
      "loss": 0.0091,
      "step": 8124
    },
    {
      "epoch": 2.178868329310807,
      "grad_norm": 0.24569170686263575,
      "learning_rate": 2.1138281272704886e-06,
      "loss": 0.0142,
      "step": 8125
    },
    {
      "epoch": 2.1791364977205685,
      "grad_norm": 0.213678830825019,
      "learning_rate": 2.1125542525668323e-06,
      "loss": 0.011,
      "step": 8126
    },
    {
      "epoch": 2.17940466613033,
      "grad_norm": 0.41499181336326546,
      "learning_rate": 2.1112806590061006e-06,
      "loss": 0.0136,
      "step": 8127
    },
    {
      "epoch": 2.1796728345400913,
      "grad_norm": 0.22999885824234215,
      "learning_rate": 2.1100073467123038e-06,
      "loss": 0.0153,
      "step": 8128
    },
    {
      "epoch": 2.1799410029498527,
      "grad_norm": 0.26293442359681457,
      "learning_rate": 2.108734315809417e-06,
      "loss": 0.0131,
      "step": 8129
    },
    {
      "epoch": 2.1802091713596137,
      "grad_norm": 0.23219623234530773,
      "learning_rate": 2.1074615664213954e-06,
      "loss": 0.0133,
      "step": 8130
    },
    {
      "epoch": 2.180477339769375,
      "grad_norm": 0.26524067834774817,
      "learning_rate": 2.10618909867216e-06,
      "loss": 0.0096,
      "step": 8131
    },
    {
      "epoch": 2.1807455081791365,
      "grad_norm": 0.19115672063442035,
      "learning_rate": 2.104916912685611e-06,
      "loss": 0.0084,
      "step": 8132
    },
    {
      "epoch": 2.181013676588898,
      "grad_norm": 0.20204860780242606,
      "learning_rate": 2.103645008585615e-06,
      "loss": 0.0086,
      "step": 8133
    },
    {
      "epoch": 2.1812818449986593,
      "grad_norm": 0.36618569586391453,
      "learning_rate": 2.102373386496014e-06,
      "loss": 0.0253,
      "step": 8134
    },
    {
      "epoch": 2.1815500134084207,
      "grad_norm": 0.1988264697649137,
      "learning_rate": 2.1011020465406252e-06,
      "loss": 0.0089,
      "step": 8135
    },
    {
      "epoch": 2.1818181818181817,
      "grad_norm": 0.28269542428198746,
      "learning_rate": 2.099830988843231e-06,
      "loss": 0.0198,
      "step": 8136
    },
    {
      "epoch": 2.182086350227943,
      "grad_norm": 0.2660115767027406,
      "learning_rate": 2.0985602135275945e-06,
      "loss": 0.0136,
      "step": 8137
    },
    {
      "epoch": 2.1823545186377045,
      "grad_norm": 0.2717872721489681,
      "learning_rate": 2.097289720717448e-06,
      "loss": 0.0112,
      "step": 8138
    },
    {
      "epoch": 2.182622687047466,
      "grad_norm": 0.1925056333761443,
      "learning_rate": 2.0960195105364935e-06,
      "loss": 0.0098,
      "step": 8139
    },
    {
      "epoch": 2.1828908554572273,
      "grad_norm": 0.31810596385654494,
      "learning_rate": 2.094749583108411e-06,
      "loss": 0.0232,
      "step": 8140
    },
    {
      "epoch": 2.1831590238669882,
      "grad_norm": 0.24952955307448485,
      "learning_rate": 2.0934799385568465e-06,
      "loss": 0.015,
      "step": 8141
    },
    {
      "epoch": 2.1834271922767496,
      "grad_norm": 0.2425271445499934,
      "learning_rate": 2.0922105770054233e-06,
      "loss": 0.0164,
      "step": 8142
    },
    {
      "epoch": 2.183695360686511,
      "grad_norm": 0.32548553308322087,
      "learning_rate": 2.090941498577739e-06,
      "loss": 0.0148,
      "step": 8143
    },
    {
      "epoch": 2.1839635290962724,
      "grad_norm": 0.2211522211170737,
      "learning_rate": 2.089672703397357e-06,
      "loss": 0.014,
      "step": 8144
    },
    {
      "epoch": 2.184231697506034,
      "grad_norm": 0.22841761011608766,
      "learning_rate": 2.0884041915878166e-06,
      "loss": 0.0149,
      "step": 8145
    },
    {
      "epoch": 2.1844998659157953,
      "grad_norm": 0.23570868723918062,
      "learning_rate": 2.0871359632726286e-06,
      "loss": 0.0118,
      "step": 8146
    },
    {
      "epoch": 2.1847680343255567,
      "grad_norm": 0.26331158046353786,
      "learning_rate": 2.085868018575278e-06,
      "loss": 0.0165,
      "step": 8147
    },
    {
      "epoch": 2.1850362027353176,
      "grad_norm": 0.20156220052446783,
      "learning_rate": 2.084600357619224e-06,
      "loss": 0.0126,
      "step": 8148
    },
    {
      "epoch": 2.185304371145079,
      "grad_norm": 0.23608420576128006,
      "learning_rate": 2.0833329805278906e-06,
      "loss": 0.0165,
      "step": 8149
    },
    {
      "epoch": 2.1855725395548404,
      "grad_norm": 0.32920183448389395,
      "learning_rate": 2.0820658874246836e-06,
      "loss": 0.0149,
      "step": 8150
    },
    {
      "epoch": 2.185840707964602,
      "grad_norm": 0.26123006141187294,
      "learning_rate": 2.080799078432972e-06,
      "loss": 0.0151,
      "step": 8151
    },
    {
      "epoch": 2.1861088763743632,
      "grad_norm": 0.2734987516414429,
      "learning_rate": 2.079532553676106e-06,
      "loss": 0.0178,
      "step": 8152
    },
    {
      "epoch": 2.186377044784124,
      "grad_norm": 0.20487937864806954,
      "learning_rate": 2.0782663132773996e-06,
      "loss": 0.0107,
      "step": 8153
    },
    {
      "epoch": 2.1866452131938856,
      "grad_norm": 0.23746548338986337,
      "learning_rate": 2.077000357360145e-06,
      "loss": 0.0192,
      "step": 8154
    },
    {
      "epoch": 2.186913381603647,
      "grad_norm": 0.2048027419257131,
      "learning_rate": 2.0757346860476068e-06,
      "loss": 0.011,
      "step": 8155
    },
    {
      "epoch": 2.1871815500134084,
      "grad_norm": 0.2603045271583022,
      "learning_rate": 2.074469299463019e-06,
      "loss": 0.0154,
      "step": 8156
    },
    {
      "epoch": 2.18744971842317,
      "grad_norm": 0.46841210225369795,
      "learning_rate": 2.0732041977295857e-06,
      "loss": 0.0216,
      "step": 8157
    },
    {
      "epoch": 2.1877178868329312,
      "grad_norm": 0.23584240544035354,
      "learning_rate": 2.071939380970492e-06,
      "loss": 0.0177,
      "step": 8158
    },
    {
      "epoch": 2.1879860552426926,
      "grad_norm": 0.17859593871673468,
      "learning_rate": 2.0706748493088846e-06,
      "loss": 0.0102,
      "step": 8159
    },
    {
      "epoch": 2.1882542236524536,
      "grad_norm": 0.21592963910514304,
      "learning_rate": 2.0694106028678917e-06,
      "loss": 0.0137,
      "step": 8160
    },
    {
      "epoch": 2.188522392062215,
      "grad_norm": 0.43396943060621457,
      "learning_rate": 2.0681466417706063e-06,
      "loss": 0.0221,
      "step": 8161
    },
    {
      "epoch": 2.1887905604719764,
      "grad_norm": 0.2513472478430232,
      "learning_rate": 2.0668829661401e-06,
      "loss": 0.0165,
      "step": 8162
    },
    {
      "epoch": 2.189058728881738,
      "grad_norm": 0.27404586715450113,
      "learning_rate": 2.0656195760994104e-06,
      "loss": 0.0145,
      "step": 8163
    },
    {
      "epoch": 2.189326897291499,
      "grad_norm": 0.3061694564566703,
      "learning_rate": 2.064356471771552e-06,
      "loss": 0.0297,
      "step": 8164
    },
    {
      "epoch": 2.18959506570126,
      "grad_norm": 0.29982641693707146,
      "learning_rate": 2.0630936532795115e-06,
      "loss": 0.0169,
      "step": 8165
    },
    {
      "epoch": 2.1898632341110216,
      "grad_norm": 0.25865829064076146,
      "learning_rate": 2.0618311207462434e-06,
      "loss": 0.0135,
      "step": 8166
    },
    {
      "epoch": 2.190131402520783,
      "grad_norm": 0.22751474621199932,
      "learning_rate": 2.06056887429468e-06,
      "loss": 0.013,
      "step": 8167
    },
    {
      "epoch": 2.1903995709305444,
      "grad_norm": 0.2813881018832031,
      "learning_rate": 2.0593069140477208e-06,
      "loss": 0.0142,
      "step": 8168
    },
    {
      "epoch": 2.190667739340306,
      "grad_norm": 0.2420377068392333,
      "learning_rate": 2.0580452401282385e-06,
      "loss": 0.0151,
      "step": 8169
    },
    {
      "epoch": 2.190935907750067,
      "grad_norm": 0.21646894112325993,
      "learning_rate": 2.056783852659082e-06,
      "loss": 0.011,
      "step": 8170
    },
    {
      "epoch": 2.1912040761598286,
      "grad_norm": 0.37148941040044847,
      "learning_rate": 2.0555227517630667e-06,
      "loss": 0.0208,
      "step": 8171
    },
    {
      "epoch": 2.1914722445695896,
      "grad_norm": 0.34676088293175944,
      "learning_rate": 2.0542619375629845e-06,
      "loss": 0.0181,
      "step": 8172
    },
    {
      "epoch": 2.191740412979351,
      "grad_norm": 0.19305063787602153,
      "learning_rate": 2.0530014101815953e-06,
      "loss": 0.011,
      "step": 8173
    },
    {
      "epoch": 2.1920085813891124,
      "grad_norm": 0.2598414045646956,
      "learning_rate": 2.0517411697416344e-06,
      "loss": 0.0164,
      "step": 8174
    },
    {
      "epoch": 2.1922767497988738,
      "grad_norm": 0.2321866106729343,
      "learning_rate": 2.0504812163658104e-06,
      "loss": 0.0107,
      "step": 8175
    },
    {
      "epoch": 2.192544918208635,
      "grad_norm": 0.2170104636866185,
      "learning_rate": 2.0492215501767976e-06,
      "loss": 0.0141,
      "step": 8176
    },
    {
      "epoch": 2.192813086618396,
      "grad_norm": 0.3740561090199242,
      "learning_rate": 2.0479621712972503e-06,
      "loss": 0.0101,
      "step": 8177
    },
    {
      "epoch": 2.1930812550281575,
      "grad_norm": 0.5604265433835057,
      "learning_rate": 2.046703079849787e-06,
      "loss": 0.0309,
      "step": 8178
    },
    {
      "epoch": 2.193349423437919,
      "grad_norm": 0.2056957541594268,
      "learning_rate": 2.045444275957006e-06,
      "loss": 0.0128,
      "step": 8179
    },
    {
      "epoch": 2.1936175918476803,
      "grad_norm": 0.3069211144298621,
      "learning_rate": 2.044185759741471e-06,
      "loss": 0.0119,
      "step": 8180
    },
    {
      "epoch": 2.1938857602574418,
      "grad_norm": 0.26603444018427486,
      "learning_rate": 2.0429275313257197e-06,
      "loss": 0.0171,
      "step": 8181
    },
    {
      "epoch": 2.194153928667203,
      "grad_norm": 0.2556508232537799,
      "learning_rate": 2.0416695908322652e-06,
      "loss": 0.014,
      "step": 8182
    },
    {
      "epoch": 2.1944220970769646,
      "grad_norm": 0.3488252778296563,
      "learning_rate": 2.040411938383587e-06,
      "loss": 0.0202,
      "step": 8183
    },
    {
      "epoch": 2.1946902654867255,
      "grad_norm": 0.2972577514889578,
      "learning_rate": 2.039154574102141e-06,
      "loss": 0.0163,
      "step": 8184
    },
    {
      "epoch": 2.194958433896487,
      "grad_norm": 0.21881010199856296,
      "learning_rate": 2.0378974981103545e-06,
      "loss": 0.0117,
      "step": 8185
    },
    {
      "epoch": 2.1952266023062483,
      "grad_norm": 0.27077214245101255,
      "learning_rate": 2.036640710530623e-06,
      "loss": 0.0194,
      "step": 8186
    },
    {
      "epoch": 2.1954947707160097,
      "grad_norm": 0.1998819404367838,
      "learning_rate": 2.0353842114853194e-06,
      "loss": 0.01,
      "step": 8187
    },
    {
      "epoch": 2.195762939125771,
      "grad_norm": 0.22260520974842166,
      "learning_rate": 2.0341280010967824e-06,
      "loss": 0.0141,
      "step": 8188
    },
    {
      "epoch": 2.196031107535532,
      "grad_norm": 0.2853771752389982,
      "learning_rate": 2.0328720794873295e-06,
      "loss": 0.0115,
      "step": 8189
    },
    {
      "epoch": 2.1962992759452935,
      "grad_norm": 0.1785541231902277,
      "learning_rate": 2.031616446779243e-06,
      "loss": 0.0098,
      "step": 8190
    },
    {
      "epoch": 2.196567444355055,
      "grad_norm": 0.2972388292447769,
      "learning_rate": 2.0303611030947823e-06,
      "loss": 0.0141,
      "step": 8191
    },
    {
      "epoch": 2.1968356127648163,
      "grad_norm": 0.37714232502248174,
      "learning_rate": 2.0291060485561785e-06,
      "loss": 0.016,
      "step": 8192
    },
    {
      "epoch": 2.1971037811745777,
      "grad_norm": 0.19681231608108196,
      "learning_rate": 2.027851283285631e-06,
      "loss": 0.0114,
      "step": 8193
    },
    {
      "epoch": 2.197371949584339,
      "grad_norm": 0.2280910180781128,
      "learning_rate": 2.026596807405312e-06,
      "loss": 0.0145,
      "step": 8194
    },
    {
      "epoch": 2.1976401179941005,
      "grad_norm": 0.2715471987553624,
      "learning_rate": 2.025342621037369e-06,
      "loss": 0.0161,
      "step": 8195
    },
    {
      "epoch": 2.1979082864038615,
      "grad_norm": 0.2412949872358672,
      "learning_rate": 2.0240887243039154e-06,
      "loss": 0.0196,
      "step": 8196
    },
    {
      "epoch": 2.198176454813623,
      "grad_norm": 0.22691132220098353,
      "learning_rate": 2.0228351173270434e-06,
      "loss": 0.0136,
      "step": 8197
    },
    {
      "epoch": 2.1984446232233843,
      "grad_norm": 0.36719817122546394,
      "learning_rate": 2.0215818002288106e-06,
      "loss": 0.0136,
      "step": 8198
    },
    {
      "epoch": 2.1987127916331457,
      "grad_norm": 0.3220336399465674,
      "learning_rate": 2.020328773131252e-06,
      "loss": 0.016,
      "step": 8199
    },
    {
      "epoch": 2.198980960042907,
      "grad_norm": 0.2644067356268142,
      "learning_rate": 2.0190760361563675e-06,
      "loss": 0.0144,
      "step": 8200
    },
    {
      "epoch": 2.199249128452668,
      "grad_norm": 0.2469330683723476,
      "learning_rate": 2.0178235894261362e-06,
      "loss": 0.014,
      "step": 8201
    },
    {
      "epoch": 2.1995172968624295,
      "grad_norm": 0.2519095389504245,
      "learning_rate": 2.016571433062506e-06,
      "loss": 0.0156,
      "step": 8202
    },
    {
      "epoch": 2.199785465272191,
      "grad_norm": 0.3208467273914004,
      "learning_rate": 2.015319567187392e-06,
      "loss": 0.0187,
      "step": 8203
    },
    {
      "epoch": 2.2000536336819523,
      "grad_norm": 0.19666452044294888,
      "learning_rate": 2.0140679919226895e-06,
      "loss": 0.0121,
      "step": 8204
    },
    {
      "epoch": 2.2003218020917137,
      "grad_norm": 0.24004880719370658,
      "learning_rate": 2.01281670739026e-06,
      "loss": 0.0146,
      "step": 8205
    },
    {
      "epoch": 2.200589970501475,
      "grad_norm": 0.2228543806889746,
      "learning_rate": 2.011565713711934e-06,
      "loss": 0.0106,
      "step": 8206
    },
    {
      "epoch": 2.200858138911236,
      "grad_norm": 0.25683697306301495,
      "learning_rate": 2.010315011009523e-06,
      "loss": 0.0123,
      "step": 8207
    },
    {
      "epoch": 2.2011263073209975,
      "grad_norm": 0.29328489642980393,
      "learning_rate": 2.0090645994047996e-06,
      "loss": 0.0121,
      "step": 8208
    },
    {
      "epoch": 2.201394475730759,
      "grad_norm": 0.352048233486218,
      "learning_rate": 2.0078144790195175e-06,
      "loss": 0.0206,
      "step": 8209
    },
    {
      "epoch": 2.2016626441405203,
      "grad_norm": 0.22909367432900038,
      "learning_rate": 2.006564649975393e-06,
      "loss": 0.0109,
      "step": 8210
    },
    {
      "epoch": 2.2019308125502817,
      "grad_norm": 0.30855616002058456,
      "learning_rate": 2.005315112394122e-06,
      "loss": 0.017,
      "step": 8211
    },
    {
      "epoch": 2.202198980960043,
      "grad_norm": 0.19029501986501723,
      "learning_rate": 2.0040658663973695e-06,
      "loss": 0.0099,
      "step": 8212
    },
    {
      "epoch": 2.202467149369804,
      "grad_norm": 0.26692658344886994,
      "learning_rate": 2.0028169121067677e-06,
      "loss": 0.0128,
      "step": 8213
    },
    {
      "epoch": 2.2027353177795654,
      "grad_norm": 0.2360485003742221,
      "learning_rate": 2.001568249643927e-06,
      "loss": 0.0116,
      "step": 8214
    },
    {
      "epoch": 2.203003486189327,
      "grad_norm": 0.2737590515103898,
      "learning_rate": 2.0003198791304235e-06,
      "loss": 0.0112,
      "step": 8215
    },
    {
      "epoch": 2.2032716545990882,
      "grad_norm": 0.24667496606758174,
      "learning_rate": 1.9990718006878113e-06,
      "loss": 0.0112,
      "step": 8216
    },
    {
      "epoch": 2.2035398230088497,
      "grad_norm": 0.917697910845092,
      "learning_rate": 1.997824014437611e-06,
      "loss": 0.0244,
      "step": 8217
    },
    {
      "epoch": 2.203807991418611,
      "grad_norm": 0.23291443638213938,
      "learning_rate": 1.996576520501313e-06,
      "loss": 0.0126,
      "step": 8218
    },
    {
      "epoch": 2.204076159828372,
      "grad_norm": 0.294224745701501,
      "learning_rate": 1.995329319000388e-06,
      "loss": 0.0151,
      "step": 8219
    },
    {
      "epoch": 2.2043443282381334,
      "grad_norm": 0.18915102883293755,
      "learning_rate": 1.9940824100562676e-06,
      "loss": 0.0121,
      "step": 8220
    },
    {
      "epoch": 2.204612496647895,
      "grad_norm": 0.22905938352199806,
      "learning_rate": 1.992835793790363e-06,
      "loss": 0.0111,
      "step": 8221
    },
    {
      "epoch": 2.2048806650576562,
      "grad_norm": 0.26532166813197205,
      "learning_rate": 1.9915894703240547e-06,
      "loss": 0.0163,
      "step": 8222
    },
    {
      "epoch": 2.2051488334674176,
      "grad_norm": 0.21107062451266365,
      "learning_rate": 1.990343439778691e-06,
      "loss": 0.0127,
      "step": 8223
    },
    {
      "epoch": 2.205417001877179,
      "grad_norm": 0.24248473672811346,
      "learning_rate": 1.9890977022755975e-06,
      "loss": 0.0108,
      "step": 8224
    },
    {
      "epoch": 2.20568517028694,
      "grad_norm": 0.2726595704197026,
      "learning_rate": 1.987852257936065e-06,
      "loss": 0.015,
      "step": 8225
    },
    {
      "epoch": 2.2059533386967014,
      "grad_norm": 0.2174774113329928,
      "learning_rate": 1.986607106881363e-06,
      "loss": 0.0118,
      "step": 8226
    },
    {
      "epoch": 2.206221507106463,
      "grad_norm": 0.2482659082055337,
      "learning_rate": 1.985362249232725e-06,
      "loss": 0.0165,
      "step": 8227
    },
    {
      "epoch": 2.206489675516224,
      "grad_norm": 0.4659567327914753,
      "learning_rate": 1.984117685111361e-06,
      "loss": 0.0206,
      "step": 8228
    },
    {
      "epoch": 2.2067578439259856,
      "grad_norm": 0.2656340354421756,
      "learning_rate": 1.982873414638455e-06,
      "loss": 0.0122,
      "step": 8229
    },
    {
      "epoch": 2.207026012335747,
      "grad_norm": 0.2864654563733647,
      "learning_rate": 1.9816294379351495e-06,
      "loss": 0.0172,
      "step": 8230
    },
    {
      "epoch": 2.207294180745508,
      "grad_norm": 0.16663532732096026,
      "learning_rate": 1.980385755122575e-06,
      "loss": 0.0088,
      "step": 8231
    },
    {
      "epoch": 2.2075623491552694,
      "grad_norm": 0.35957101558993265,
      "learning_rate": 1.97914236632182e-06,
      "loss": 0.0244,
      "step": 8232
    },
    {
      "epoch": 2.207830517565031,
      "grad_norm": 0.24452708773953669,
      "learning_rate": 1.9778992716539532e-06,
      "loss": 0.0131,
      "step": 8233
    },
    {
      "epoch": 2.208098685974792,
      "grad_norm": 0.22462888779702833,
      "learning_rate": 1.976656471240012e-06,
      "loss": 0.0132,
      "step": 8234
    },
    {
      "epoch": 2.2083668543845536,
      "grad_norm": 0.21915421301990007,
      "learning_rate": 1.9754139652010025e-06,
      "loss": 0.0135,
      "step": 8235
    },
    {
      "epoch": 2.208635022794315,
      "grad_norm": 0.30582069285245794,
      "learning_rate": 1.9741717536579057e-06,
      "loss": 0.0173,
      "step": 8236
    },
    {
      "epoch": 2.208903191204076,
      "grad_norm": 0.20515263502991268,
      "learning_rate": 1.9729298367316713e-06,
      "loss": 0.0113,
      "step": 8237
    },
    {
      "epoch": 2.2091713596138374,
      "grad_norm": 0.24090092209809588,
      "learning_rate": 1.9716882145432208e-06,
      "loss": 0.0127,
      "step": 8238
    },
    {
      "epoch": 2.2094395280235988,
      "grad_norm": 0.2078462199163122,
      "learning_rate": 1.970446887213451e-06,
      "loss": 0.0118,
      "step": 8239
    },
    {
      "epoch": 2.20970769643336,
      "grad_norm": 0.21345577677834673,
      "learning_rate": 1.969205854863223e-06,
      "loss": 0.0133,
      "step": 8240
    },
    {
      "epoch": 2.2099758648431216,
      "grad_norm": 0.215015818637565,
      "learning_rate": 1.967965117613378e-06,
      "loss": 0.0129,
      "step": 8241
    },
    {
      "epoch": 2.210244033252883,
      "grad_norm": 0.2564102862523625,
      "learning_rate": 1.9667246755847157e-06,
      "loss": 0.0158,
      "step": 8242
    },
    {
      "epoch": 2.210512201662644,
      "grad_norm": 0.2982154800674518,
      "learning_rate": 1.9654845288980183e-06,
      "loss": 0.0167,
      "step": 8243
    },
    {
      "epoch": 2.2107803700724054,
      "grad_norm": 0.35845882568888393,
      "learning_rate": 1.9642446776740383e-06,
      "loss": 0.0147,
      "step": 8244
    },
    {
      "epoch": 2.2110485384821668,
      "grad_norm": 0.1894928467332662,
      "learning_rate": 1.9630051220334915e-06,
      "loss": 0.0103,
      "step": 8245
    },
    {
      "epoch": 2.211316706891928,
      "grad_norm": 0.3650781574872496,
      "learning_rate": 1.961765862097076e-06,
      "loss": 0.0192,
      "step": 8246
    },
    {
      "epoch": 2.2115848753016896,
      "grad_norm": 0.2778592857967054,
      "learning_rate": 1.9605268979854493e-06,
      "loss": 0.0165,
      "step": 8247
    },
    {
      "epoch": 2.211853043711451,
      "grad_norm": 0.22881872251585364,
      "learning_rate": 1.959288229819251e-06,
      "loss": 0.0142,
      "step": 8248
    },
    {
      "epoch": 2.212121212121212,
      "grad_norm": 0.22489033690033597,
      "learning_rate": 1.958049857719083e-06,
      "loss": 0.012,
      "step": 8249
    },
    {
      "epoch": 2.2123893805309733,
      "grad_norm": 0.223428197727709,
      "learning_rate": 1.9568117818055244e-06,
      "loss": 0.0122,
      "step": 8250
    },
    {
      "epoch": 2.2126575489407347,
      "grad_norm": 0.3002490564547751,
      "learning_rate": 1.955574002199125e-06,
      "loss": 0.0167,
      "step": 8251
    },
    {
      "epoch": 2.212925717350496,
      "grad_norm": 0.2954094443479846,
      "learning_rate": 1.9543365190204e-06,
      "loss": 0.0187,
      "step": 8252
    },
    {
      "epoch": 2.2131938857602576,
      "grad_norm": 0.2878556904645038,
      "learning_rate": 1.953099332389844e-06,
      "loss": 0.0191,
      "step": 8253
    },
    {
      "epoch": 2.213462054170019,
      "grad_norm": 0.18114150611054838,
      "learning_rate": 1.9518624424279166e-06,
      "loss": 0.0081,
      "step": 8254
    },
    {
      "epoch": 2.21373022257978,
      "grad_norm": 0.18329833849845795,
      "learning_rate": 1.9506258492550485e-06,
      "loss": 0.0107,
      "step": 8255
    },
    {
      "epoch": 2.2139983909895413,
      "grad_norm": 0.18942554235284997,
      "learning_rate": 1.9493895529916474e-06,
      "loss": 0.011,
      "step": 8256
    },
    {
      "epoch": 2.2142665593993027,
      "grad_norm": 0.19741336802141043,
      "learning_rate": 1.948153553758085e-06,
      "loss": 0.0104,
      "step": 8257
    },
    {
      "epoch": 2.214534727809064,
      "grad_norm": 0.17645697159424503,
      "learning_rate": 1.94691785167471e-06,
      "loss": 0.0105,
      "step": 8258
    },
    {
      "epoch": 2.2148028962188255,
      "grad_norm": 0.22691199148149038,
      "learning_rate": 1.9456824468618365e-06,
      "loss": 0.0131,
      "step": 8259
    },
    {
      "epoch": 2.215071064628587,
      "grad_norm": 0.24612984871556332,
      "learning_rate": 1.9444473394397535e-06,
      "loss": 0.0146,
      "step": 8260
    },
    {
      "epoch": 2.215339233038348,
      "grad_norm": 0.2614212515859232,
      "learning_rate": 1.943212529528723e-06,
      "loss": 0.0125,
      "step": 8261
    },
    {
      "epoch": 2.2156074014481093,
      "grad_norm": 0.2975679419960307,
      "learning_rate": 1.941978017248971e-06,
      "loss": 0.0156,
      "step": 8262
    },
    {
      "epoch": 2.2158755698578707,
      "grad_norm": 0.19591205733018469,
      "learning_rate": 1.940743802720702e-06,
      "loss": 0.0121,
      "step": 8263
    },
    {
      "epoch": 2.216143738267632,
      "grad_norm": 0.5660046526438169,
      "learning_rate": 1.9395098860640856e-06,
      "loss": 0.0495,
      "step": 8264
    },
    {
      "epoch": 2.2164119066773935,
      "grad_norm": 0.25615785085624326,
      "learning_rate": 1.9382762673992673e-06,
      "loss": 0.0148,
      "step": 8265
    },
    {
      "epoch": 2.216680075087155,
      "grad_norm": 0.2860856280241198,
      "learning_rate": 1.937042946846361e-06,
      "loss": 0.0105,
      "step": 8266
    },
    {
      "epoch": 2.216948243496916,
      "grad_norm": 0.20781741558592928,
      "learning_rate": 1.935809924525448e-06,
      "loss": 0.0108,
      "step": 8267
    },
    {
      "epoch": 2.2172164119066773,
      "grad_norm": 0.2593300934729011,
      "learning_rate": 1.93457720055659e-06,
      "loss": 0.018,
      "step": 8268
    },
    {
      "epoch": 2.2174845803164387,
      "grad_norm": 0.30742140921589506,
      "learning_rate": 1.9333447750598095e-06,
      "loss": 0.019,
      "step": 8269
    },
    {
      "epoch": 2.2177527487262,
      "grad_norm": 0.2561469110538905,
      "learning_rate": 1.932112648155107e-06,
      "loss": 0.0156,
      "step": 8270
    },
    {
      "epoch": 2.2180209171359615,
      "grad_norm": 0.27682924356337,
      "learning_rate": 1.9308808199624518e-06,
      "loss": 0.0164,
      "step": 8271
    },
    {
      "epoch": 2.218289085545723,
      "grad_norm": 0.2036876804075994,
      "learning_rate": 1.929649290601781e-06,
      "loss": 0.0118,
      "step": 8272
    },
    {
      "epoch": 2.218557253955484,
      "grad_norm": 0.21986702651922632,
      "learning_rate": 1.928418060193009e-06,
      "loss": 0.0121,
      "step": 8273
    },
    {
      "epoch": 2.2188254223652453,
      "grad_norm": 0.3104317301641841,
      "learning_rate": 1.927187128856014e-06,
      "loss": 0.0233,
      "step": 8274
    },
    {
      "epoch": 2.2190935907750067,
      "grad_norm": 0.2479554695197249,
      "learning_rate": 1.925956496710652e-06,
      "loss": 0.0135,
      "step": 8275
    },
    {
      "epoch": 2.219361759184768,
      "grad_norm": 0.2922416719432511,
      "learning_rate": 1.9247261638767427e-06,
      "loss": 0.0225,
      "step": 8276
    },
    {
      "epoch": 2.2196299275945295,
      "grad_norm": 0.23988489862933485,
      "learning_rate": 1.923496130474084e-06,
      "loss": 0.0151,
      "step": 8277
    },
    {
      "epoch": 2.219898096004291,
      "grad_norm": 0.2335797090472106,
      "learning_rate": 1.9222663966224386e-06,
      "loss": 0.0125,
      "step": 8278
    },
    {
      "epoch": 2.220166264414052,
      "grad_norm": 0.3362073759996765,
      "learning_rate": 1.9210369624415413e-06,
      "loss": 0.0193,
      "step": 8279
    },
    {
      "epoch": 2.2204344328238133,
      "grad_norm": 0.2321747879687417,
      "learning_rate": 1.9198078280511013e-06,
      "loss": 0.0194,
      "step": 8280
    },
    {
      "epoch": 2.2207026012335747,
      "grad_norm": 0.2524216600750648,
      "learning_rate": 1.918578993570796e-06,
      "loss": 0.0202,
      "step": 8281
    },
    {
      "epoch": 2.220970769643336,
      "grad_norm": 0.27983711289265273,
      "learning_rate": 1.917350459120272e-06,
      "loss": 0.0174,
      "step": 8282
    },
    {
      "epoch": 2.2212389380530975,
      "grad_norm": 0.2452037039069059,
      "learning_rate": 1.9161222248191515e-06,
      "loss": 0.0156,
      "step": 8283
    },
    {
      "epoch": 2.221507106462859,
      "grad_norm": 0.2364984564688047,
      "learning_rate": 1.91489429078702e-06,
      "loss": 0.0114,
      "step": 8284
    },
    {
      "epoch": 2.22177527487262,
      "grad_norm": 0.21775613990598863,
      "learning_rate": 1.9136666571434425e-06,
      "loss": 0.0135,
      "step": 8285
    },
    {
      "epoch": 2.2220434432823812,
      "grad_norm": 0.23605524853830348,
      "learning_rate": 1.912439324007947e-06,
      "loss": 0.0152,
      "step": 8286
    },
    {
      "epoch": 2.2223116116921426,
      "grad_norm": 0.25674699993789263,
      "learning_rate": 1.911212291500037e-06,
      "loss": 0.0173,
      "step": 8287
    },
    {
      "epoch": 2.222579780101904,
      "grad_norm": 0.2052191202252737,
      "learning_rate": 1.9099855597391876e-06,
      "loss": 0.0114,
      "step": 8288
    },
    {
      "epoch": 2.2228479485116655,
      "grad_norm": 0.26539626889977636,
      "learning_rate": 1.90875912884484e-06,
      "loss": 0.0144,
      "step": 8289
    },
    {
      "epoch": 2.223116116921427,
      "grad_norm": 0.2513821837348073,
      "learning_rate": 1.9075329989364076e-06,
      "loss": 0.0146,
      "step": 8290
    },
    {
      "epoch": 2.223384285331188,
      "grad_norm": 0.2310165207486927,
      "learning_rate": 1.9063071701332785e-06,
      "loss": 0.0176,
      "step": 8291
    },
    {
      "epoch": 2.223652453740949,
      "grad_norm": 0.24337359694745206,
      "learning_rate": 1.9050816425548046e-06,
      "loss": 0.0104,
      "step": 8292
    },
    {
      "epoch": 2.2239206221507106,
      "grad_norm": 0.32868182284790426,
      "learning_rate": 1.903856416320316e-06,
      "loss": 0.0174,
      "step": 8293
    },
    {
      "epoch": 2.224188790560472,
      "grad_norm": 0.18983830065859716,
      "learning_rate": 1.9026314915491067e-06,
      "loss": 0.0111,
      "step": 8294
    },
    {
      "epoch": 2.2244569589702334,
      "grad_norm": 0.2261355359206126,
      "learning_rate": 1.9014068683604475e-06,
      "loss": 0.0097,
      "step": 8295
    },
    {
      "epoch": 2.224725127379995,
      "grad_norm": 0.2719942161409358,
      "learning_rate": 1.9001825468735729e-06,
      "loss": 0.0161,
      "step": 8296
    },
    {
      "epoch": 2.224993295789756,
      "grad_norm": 0.27268868535095575,
      "learning_rate": 1.8989585272076937e-06,
      "loss": 0.0176,
      "step": 8297
    },
    {
      "epoch": 2.225261464199517,
      "grad_norm": 0.2729738206201802,
      "learning_rate": 1.8977348094819914e-06,
      "loss": 0.0157,
      "step": 8298
    },
    {
      "epoch": 2.2255296326092786,
      "grad_norm": 0.20276190528928145,
      "learning_rate": 1.8965113938156132e-06,
      "loss": 0.0118,
      "step": 8299
    },
    {
      "epoch": 2.22579780101904,
      "grad_norm": 0.30885195790218806,
      "learning_rate": 1.8952882803276822e-06,
      "loss": 0.0203,
      "step": 8300
    },
    {
      "epoch": 2.2260659694288014,
      "grad_norm": 0.26372368002897234,
      "learning_rate": 1.8940654691372884e-06,
      "loss": 0.0164,
      "step": 8301
    },
    {
      "epoch": 2.226334137838563,
      "grad_norm": 0.38067120211407274,
      "learning_rate": 1.8928429603634924e-06,
      "loss": 0.0157,
      "step": 8302
    },
    {
      "epoch": 2.226602306248324,
      "grad_norm": 0.24902300390960355,
      "learning_rate": 1.8916207541253295e-06,
      "loss": 0.0151,
      "step": 8303
    },
    {
      "epoch": 2.226870474658085,
      "grad_norm": 0.24593242992777087,
      "learning_rate": 1.8903988505417998e-06,
      "loss": 0.013,
      "step": 8304
    },
    {
      "epoch": 2.2271386430678466,
      "grad_norm": 0.29686637895574625,
      "learning_rate": 1.8891772497318795e-06,
      "loss": 0.0123,
      "step": 8305
    },
    {
      "epoch": 2.227406811477608,
      "grad_norm": 0.2765502319911622,
      "learning_rate": 1.8879559518145096e-06,
      "loss": 0.0135,
      "step": 8306
    },
    {
      "epoch": 2.2276749798873694,
      "grad_norm": 0.3273650551803932,
      "learning_rate": 1.8867349569086064e-06,
      "loss": 0.0145,
      "step": 8307
    },
    {
      "epoch": 2.227943148297131,
      "grad_norm": 0.3585962002939518,
      "learning_rate": 1.8855142651330566e-06,
      "loss": 0.0327,
      "step": 8308
    },
    {
      "epoch": 2.2282113167068918,
      "grad_norm": 0.23510924218910548,
      "learning_rate": 1.8842938766067127e-06,
      "loss": 0.0169,
      "step": 8309
    },
    {
      "epoch": 2.228479485116653,
      "grad_norm": 0.32140421696187305,
      "learning_rate": 1.8830737914484033e-06,
      "loss": 0.0172,
      "step": 8310
    },
    {
      "epoch": 2.2287476535264146,
      "grad_norm": 0.2639823590373743,
      "learning_rate": 1.8818540097769218e-06,
      "loss": 0.0187,
      "step": 8311
    },
    {
      "epoch": 2.229015821936176,
      "grad_norm": 0.22021570393667517,
      "learning_rate": 1.8806345317110387e-06,
      "loss": 0.0128,
      "step": 8312
    },
    {
      "epoch": 2.2292839903459374,
      "grad_norm": 0.2528034563400032,
      "learning_rate": 1.879415357369488e-06,
      "loss": 0.0113,
      "step": 8313
    },
    {
      "epoch": 2.2295521587556983,
      "grad_norm": 0.2159421474713139,
      "learning_rate": 1.87819648687098e-06,
      "loss": 0.0112,
      "step": 8314
    },
    {
      "epoch": 2.2298203271654597,
      "grad_norm": 0.25896997214402545,
      "learning_rate": 1.8769779203341925e-06,
      "loss": 0.014,
      "step": 8315
    },
    {
      "epoch": 2.230088495575221,
      "grad_norm": 0.20804231297209,
      "learning_rate": 1.8757596578777715e-06,
      "loss": 0.0113,
      "step": 8316
    },
    {
      "epoch": 2.2303566639849826,
      "grad_norm": 0.3032048472512723,
      "learning_rate": 1.8745416996203375e-06,
      "loss": 0.0179,
      "step": 8317
    },
    {
      "epoch": 2.230624832394744,
      "grad_norm": 0.439151122404767,
      "learning_rate": 1.8733240456804823e-06,
      "loss": 0.013,
      "step": 8318
    },
    {
      "epoch": 2.2308930008045054,
      "grad_norm": 0.22085748003204903,
      "learning_rate": 1.8721066961767626e-06,
      "loss": 0.0115,
      "step": 8319
    },
    {
      "epoch": 2.2311611692142668,
      "grad_norm": 0.2832245193928407,
      "learning_rate": 1.8708896512277102e-06,
      "loss": 0.0162,
      "step": 8320
    },
    {
      "epoch": 2.2314293376240277,
      "grad_norm": 0.2526843371098077,
      "learning_rate": 1.8696729109518235e-06,
      "loss": 0.0137,
      "step": 8321
    },
    {
      "epoch": 2.231697506033789,
      "grad_norm": 0.25348089304872545,
      "learning_rate": 1.8684564754675765e-06,
      "loss": 0.0119,
      "step": 8322
    },
    {
      "epoch": 2.2319656744435505,
      "grad_norm": 0.2002955813734136,
      "learning_rate": 1.8672403448934069e-06,
      "loss": 0.0092,
      "step": 8323
    },
    {
      "epoch": 2.232233842853312,
      "grad_norm": 0.1947161156217244,
      "learning_rate": 1.8660245193477266e-06,
      "loss": 0.0104,
      "step": 8324
    },
    {
      "epoch": 2.2325020112630733,
      "grad_norm": 0.29650823424436046,
      "learning_rate": 1.864808998948921e-06,
      "loss": 0.0154,
      "step": 8325
    },
    {
      "epoch": 2.2327701796728343,
      "grad_norm": 0.21422757560315422,
      "learning_rate": 1.863593783815339e-06,
      "loss": 0.0107,
      "step": 8326
    },
    {
      "epoch": 2.2330383480825957,
      "grad_norm": 0.28149428245118413,
      "learning_rate": 1.8623788740653032e-06,
      "loss": 0.017,
      "step": 8327
    },
    {
      "epoch": 2.233306516492357,
      "grad_norm": 0.24903829824693105,
      "learning_rate": 1.8611642698171045e-06,
      "loss": 0.012,
      "step": 8328
    },
    {
      "epoch": 2.2335746849021185,
      "grad_norm": 0.407864035131312,
      "learning_rate": 1.8599499711890079e-06,
      "loss": 0.0216,
      "step": 8329
    },
    {
      "epoch": 2.23384285331188,
      "grad_norm": 0.25157355224235883,
      "learning_rate": 1.858735978299247e-06,
      "loss": 0.0143,
      "step": 8330
    },
    {
      "epoch": 2.2341110217216413,
      "grad_norm": 0.46909416865291176,
      "learning_rate": 1.8575222912660224e-06,
      "loss": 0.0111,
      "step": 8331
    },
    {
      "epoch": 2.2343791901314027,
      "grad_norm": 0.30403674282606535,
      "learning_rate": 1.856308910207511e-06,
      "loss": 0.029,
      "step": 8332
    },
    {
      "epoch": 2.2346473585411637,
      "grad_norm": 0.3014268162125151,
      "learning_rate": 1.855095835241853e-06,
      "loss": 0.0183,
      "step": 8333
    },
    {
      "epoch": 2.234915526950925,
      "grad_norm": 0.2245817312371467,
      "learning_rate": 1.8538830664871644e-06,
      "loss": 0.0125,
      "step": 8334
    },
    {
      "epoch": 2.2351836953606865,
      "grad_norm": 0.24011462910908046,
      "learning_rate": 1.85267060406153e-06,
      "loss": 0.0153,
      "step": 8335
    },
    {
      "epoch": 2.235451863770448,
      "grad_norm": 0.23971004533726453,
      "learning_rate": 1.8514584480830012e-06,
      "loss": 0.0131,
      "step": 8336
    },
    {
      "epoch": 2.2357200321802093,
      "grad_norm": 0.18248817492219066,
      "learning_rate": 1.8502465986696062e-06,
      "loss": 0.0153,
      "step": 8337
    },
    {
      "epoch": 2.2359882005899703,
      "grad_norm": 0.3221824040820477,
      "learning_rate": 1.8490350559393367e-06,
      "loss": 0.0175,
      "step": 8338
    },
    {
      "epoch": 2.2362563689997317,
      "grad_norm": 0.3242498264003011,
      "learning_rate": 1.8478238200101573e-06,
      "loss": 0.0143,
      "step": 8339
    },
    {
      "epoch": 2.236524537409493,
      "grad_norm": 0.24828076698751095,
      "learning_rate": 1.8466128910000048e-06,
      "loss": 0.0121,
      "step": 8340
    },
    {
      "epoch": 2.2367927058192545,
      "grad_norm": 0.22528977817329923,
      "learning_rate": 1.8454022690267815e-06,
      "loss": 0.0122,
      "step": 8341
    },
    {
      "epoch": 2.237060874229016,
      "grad_norm": 0.24623456950600076,
      "learning_rate": 1.8441919542083658e-06,
      "loss": 0.0133,
      "step": 8342
    },
    {
      "epoch": 2.2373290426387773,
      "grad_norm": 0.2312126761601223,
      "learning_rate": 1.8429819466625993e-06,
      "loss": 0.0173,
      "step": 8343
    },
    {
      "epoch": 2.2375972110485387,
      "grad_norm": 0.28949892537224914,
      "learning_rate": 1.8417722465073e-06,
      "loss": 0.015,
      "step": 8344
    },
    {
      "epoch": 2.2378653794582997,
      "grad_norm": 0.31800230655328876,
      "learning_rate": 1.840562853860251e-06,
      "loss": 0.0113,
      "step": 8345
    },
    {
      "epoch": 2.238133547868061,
      "grad_norm": 0.28330561279559063,
      "learning_rate": 1.8393537688392088e-06,
      "loss": 0.0108,
      "step": 8346
    },
    {
      "epoch": 2.2384017162778225,
      "grad_norm": 0.41487225225241675,
      "learning_rate": 1.8381449915619003e-06,
      "loss": 0.0193,
      "step": 8347
    },
    {
      "epoch": 2.238669884687584,
      "grad_norm": 0.22765504947372817,
      "learning_rate": 1.8369365221460178e-06,
      "loss": 0.012,
      "step": 8348
    },
    {
      "epoch": 2.2389380530973453,
      "grad_norm": 0.2116448219753899,
      "learning_rate": 1.8357283607092302e-06,
      "loss": 0.009,
      "step": 8349
    },
    {
      "epoch": 2.2392062215071062,
      "grad_norm": 0.4486320902391489,
      "learning_rate": 1.8345205073691708e-06,
      "loss": 0.0236,
      "step": 8350
    },
    {
      "epoch": 2.2394743899168676,
      "grad_norm": 0.2427698266046734,
      "learning_rate": 1.8333129622434441e-06,
      "loss": 0.0129,
      "step": 8351
    },
    {
      "epoch": 2.239742558326629,
      "grad_norm": 0.2639546069728409,
      "learning_rate": 1.8321057254496283e-06,
      "loss": 0.0146,
      "step": 8352
    },
    {
      "epoch": 2.2400107267363905,
      "grad_norm": 0.35224524366660154,
      "learning_rate": 1.8308987971052662e-06,
      "loss": 0.0221,
      "step": 8353
    },
    {
      "epoch": 2.240278895146152,
      "grad_norm": 0.1962515099187816,
      "learning_rate": 1.8296921773278754e-06,
      "loss": 0.0123,
      "step": 8354
    },
    {
      "epoch": 2.2405470635559133,
      "grad_norm": 0.29625539643446247,
      "learning_rate": 1.8284858662349391e-06,
      "loss": 0.0127,
      "step": 8355
    },
    {
      "epoch": 2.2408152319656747,
      "grad_norm": 0.3294326863685458,
      "learning_rate": 1.8272798639439132e-06,
      "loss": 0.0173,
      "step": 8356
    },
    {
      "epoch": 2.2410834003754356,
      "grad_norm": 0.22753157609650246,
      "learning_rate": 1.8260741705722257e-06,
      "loss": 0.0104,
      "step": 8357
    },
    {
      "epoch": 2.241351568785197,
      "grad_norm": 0.2541140642483386,
      "learning_rate": 1.8248687862372683e-06,
      "loss": 0.0119,
      "step": 8358
    },
    {
      "epoch": 2.2416197371949584,
      "grad_norm": 0.2155614256869364,
      "learning_rate": 1.8236637110564081e-06,
      "loss": 0.0111,
      "step": 8359
    },
    {
      "epoch": 2.24188790560472,
      "grad_norm": 0.24373517582836976,
      "learning_rate": 1.8224589451469788e-06,
      "loss": 0.0141,
      "step": 8360
    },
    {
      "epoch": 2.2421560740144812,
      "grad_norm": 0.29753368477557174,
      "learning_rate": 1.8212544886262867e-06,
      "loss": 0.0184,
      "step": 8361
    },
    {
      "epoch": 2.242424242424242,
      "grad_norm": 0.2775801036783729,
      "learning_rate": 1.8200503416116056e-06,
      "loss": 0.0169,
      "step": 8362
    },
    {
      "epoch": 2.2426924108340036,
      "grad_norm": 0.2495354974777359,
      "learning_rate": 1.8188465042201786e-06,
      "loss": 0.0102,
      "step": 8363
    },
    {
      "epoch": 2.242960579243765,
      "grad_norm": 0.24388044664602188,
      "learning_rate": 1.817642976569224e-06,
      "loss": 0.0162,
      "step": 8364
    },
    {
      "epoch": 2.2432287476535264,
      "grad_norm": 0.2828977941671306,
      "learning_rate": 1.8164397587759208e-06,
      "loss": 0.0159,
      "step": 8365
    },
    {
      "epoch": 2.243496916063288,
      "grad_norm": 0.23560104623648043,
      "learning_rate": 1.815236850957427e-06,
      "loss": 0.0116,
      "step": 8366
    },
    {
      "epoch": 2.2437650844730492,
      "grad_norm": 0.2780074544742447,
      "learning_rate": 1.8140342532308675e-06,
      "loss": 0.0156,
      "step": 8367
    },
    {
      "epoch": 2.2440332528828106,
      "grad_norm": 0.287518522201214,
      "learning_rate": 1.812831965713332e-06,
      "loss": 0.0139,
      "step": 8368
    },
    {
      "epoch": 2.2443014212925716,
      "grad_norm": 0.22456352279964373,
      "learning_rate": 1.811629988521888e-06,
      "loss": 0.0146,
      "step": 8369
    },
    {
      "epoch": 2.244569589702333,
      "grad_norm": 0.3462136169085823,
      "learning_rate": 1.8104283217735651e-06,
      "loss": 0.018,
      "step": 8370
    },
    {
      "epoch": 2.2448377581120944,
      "grad_norm": 0.2360960588022182,
      "learning_rate": 1.8092269655853706e-06,
      "loss": 0.0142,
      "step": 8371
    },
    {
      "epoch": 2.245105926521856,
      "grad_norm": 0.20852790876816335,
      "learning_rate": 1.8080259200742735e-06,
      "loss": 0.0127,
      "step": 8372
    },
    {
      "epoch": 2.245374094931617,
      "grad_norm": 0.2910053259722999,
      "learning_rate": 1.8068251853572177e-06,
      "loss": 0.0153,
      "step": 8373
    },
    {
      "epoch": 2.245642263341378,
      "grad_norm": 0.2834484633699396,
      "learning_rate": 1.805624761551119e-06,
      "loss": 0.0167,
      "step": 8374
    },
    {
      "epoch": 2.2459104317511396,
      "grad_norm": 0.203046579914804,
      "learning_rate": 1.8044246487728535e-06,
      "loss": 0.0096,
      "step": 8375
    },
    {
      "epoch": 2.246178600160901,
      "grad_norm": 0.22024456760589958,
      "learning_rate": 1.8032248471392755e-06,
      "loss": 0.0118,
      "step": 8376
    },
    {
      "epoch": 2.2464467685706624,
      "grad_norm": 0.25249396309268335,
      "learning_rate": 1.8020253567672085e-06,
      "loss": 0.0154,
      "step": 8377
    },
    {
      "epoch": 2.246714936980424,
      "grad_norm": 0.38038584475962667,
      "learning_rate": 1.800826177773441e-06,
      "loss": 0.0207,
      "step": 8378
    },
    {
      "epoch": 2.246983105390185,
      "grad_norm": 0.33561147818734655,
      "learning_rate": 1.7996273102747363e-06,
      "loss": 0.0143,
      "step": 8379
    },
    {
      "epoch": 2.247251273799946,
      "grad_norm": 0.24747128263755447,
      "learning_rate": 1.7984287543878216e-06,
      "loss": 0.0156,
      "step": 8380
    },
    {
      "epoch": 2.2475194422097076,
      "grad_norm": 0.3108950197889388,
      "learning_rate": 1.7972305102294007e-06,
      "loss": 0.0151,
      "step": 8381
    },
    {
      "epoch": 2.247787610619469,
      "grad_norm": 0.2875559243437829,
      "learning_rate": 1.7960325779161408e-06,
      "loss": 0.0135,
      "step": 8382
    },
    {
      "epoch": 2.2480557790292304,
      "grad_norm": 0.22202555936019988,
      "learning_rate": 1.7948349575646817e-06,
      "loss": 0.0103,
      "step": 8383
    },
    {
      "epoch": 2.2483239474389918,
      "grad_norm": 0.2519429999289391,
      "learning_rate": 1.793637649291635e-06,
      "loss": 0.0175,
      "step": 8384
    },
    {
      "epoch": 2.248592115848753,
      "grad_norm": 0.28930475977184633,
      "learning_rate": 1.7924406532135764e-06,
      "loss": 0.0167,
      "step": 8385
    },
    {
      "epoch": 2.248860284258514,
      "grad_norm": 0.2354980726770266,
      "learning_rate": 1.7912439694470575e-06,
      "loss": 0.0127,
      "step": 8386
    },
    {
      "epoch": 2.2491284526682755,
      "grad_norm": 0.2518805655968204,
      "learning_rate": 1.7900475981085935e-06,
      "loss": 0.013,
      "step": 8387
    },
    {
      "epoch": 2.249396621078037,
      "grad_norm": 0.3403962671344334,
      "learning_rate": 1.788851539314671e-06,
      "loss": 0.0198,
      "step": 8388
    },
    {
      "epoch": 2.2496647894877984,
      "grad_norm": 0.26899681165811345,
      "learning_rate": 1.7876557931817507e-06,
      "loss": 0.0154,
      "step": 8389
    },
    {
      "epoch": 2.2499329578975598,
      "grad_norm": 0.27006601080909026,
      "learning_rate": 1.7864603598262549e-06,
      "loss": 0.0172,
      "step": 8390
    },
    {
      "epoch": 2.250201126307321,
      "grad_norm": 0.26206043491755426,
      "learning_rate": 1.7852652393645842e-06,
      "loss": 0.0168,
      "step": 8391
    },
    {
      "epoch": 2.2504692947170826,
      "grad_norm": 0.26925943706433847,
      "learning_rate": 1.7840704319131009e-06,
      "loss": 0.0168,
      "step": 8392
    },
    {
      "epoch": 2.2507374631268435,
      "grad_norm": 0.35960691979775616,
      "learning_rate": 1.782875937588141e-06,
      "loss": 0.016,
      "step": 8393
    },
    {
      "epoch": 2.251005631536605,
      "grad_norm": 0.3016779320410955,
      "learning_rate": 1.7816817565060113e-06,
      "loss": 0.0154,
      "step": 8394
    },
    {
      "epoch": 2.2512737999463663,
      "grad_norm": 0.202546778624593,
      "learning_rate": 1.780487888782983e-06,
      "loss": 0.0108,
      "step": 8395
    },
    {
      "epoch": 2.2515419683561277,
      "grad_norm": 0.19196112769990653,
      "learning_rate": 1.779294334535303e-06,
      "loss": 0.0119,
      "step": 8396
    },
    {
      "epoch": 2.251810136765889,
      "grad_norm": 0.20738712280280852,
      "learning_rate": 1.7781010938791815e-06,
      "loss": 0.0107,
      "step": 8397
    },
    {
      "epoch": 2.25207830517565,
      "grad_norm": 0.26993265985758114,
      "learning_rate": 1.7769081669308035e-06,
      "loss": 0.0127,
      "step": 8398
    },
    {
      "epoch": 2.2523464735854115,
      "grad_norm": 0.3232650437046919,
      "learning_rate": 1.775715553806321e-06,
      "loss": 0.0149,
      "step": 8399
    },
    {
      "epoch": 2.252614641995173,
      "grad_norm": 0.20868487809385922,
      "learning_rate": 1.7745232546218522e-06,
      "loss": 0.0118,
      "step": 8400
    },
    {
      "epoch": 2.2528828104049343,
      "grad_norm": 0.15214608610035332,
      "learning_rate": 1.7733312694934929e-06,
      "loss": 0.0071,
      "step": 8401
    },
    {
      "epoch": 2.2531509788146957,
      "grad_norm": 0.23290841853588823,
      "learning_rate": 1.7721395985372997e-06,
      "loss": 0.0129,
      "step": 8402
    },
    {
      "epoch": 2.253419147224457,
      "grad_norm": 0.3342736360530436,
      "learning_rate": 1.7709482418693036e-06,
      "loss": 0.0243,
      "step": 8403
    },
    {
      "epoch": 2.2536873156342185,
      "grad_norm": 0.23649168326749187,
      "learning_rate": 1.7697571996055058e-06,
      "loss": 0.0107,
      "step": 8404
    },
    {
      "epoch": 2.2539554840439795,
      "grad_norm": 0.21544977778078253,
      "learning_rate": 1.7685664718618716e-06,
      "loss": 0.0093,
      "step": 8405
    },
    {
      "epoch": 2.254223652453741,
      "grad_norm": 0.21530024485611202,
      "learning_rate": 1.7673760587543425e-06,
      "loss": 0.0088,
      "step": 8406
    },
    {
      "epoch": 2.2544918208635023,
      "grad_norm": 0.24820648766956802,
      "learning_rate": 1.766185960398823e-06,
      "loss": 0.0169,
      "step": 8407
    },
    {
      "epoch": 2.2547599892732637,
      "grad_norm": 0.19072420963277156,
      "learning_rate": 1.7649961769111922e-06,
      "loss": 0.0144,
      "step": 8408
    },
    {
      "epoch": 2.255028157683025,
      "grad_norm": 0.19346045257770214,
      "learning_rate": 1.763806708407293e-06,
      "loss": 0.0119,
      "step": 8409
    },
    {
      "epoch": 2.255296326092786,
      "grad_norm": 0.26910231680635227,
      "learning_rate": 1.7626175550029446e-06,
      "loss": 0.0139,
      "step": 8410
    },
    {
      "epoch": 2.2555644945025475,
      "grad_norm": 0.2779314829663427,
      "learning_rate": 1.7614287168139298e-06,
      "loss": 0.0189,
      "step": 8411
    },
    {
      "epoch": 2.255832662912309,
      "grad_norm": 0.2443602650053386,
      "learning_rate": 1.760240193956001e-06,
      "loss": 0.0101,
      "step": 8412
    },
    {
      "epoch": 2.2561008313220703,
      "grad_norm": 0.1945014191963831,
      "learning_rate": 1.7590519865448829e-06,
      "loss": 0.0086,
      "step": 8413
    },
    {
      "epoch": 2.2563689997318317,
      "grad_norm": 0.2162503105149135,
      "learning_rate": 1.7578640946962699e-06,
      "loss": 0.013,
      "step": 8414
    },
    {
      "epoch": 2.256637168141593,
      "grad_norm": 0.2962144829908356,
      "learning_rate": 1.7566765185258205e-06,
      "loss": 0.0169,
      "step": 8415
    },
    {
      "epoch": 2.256905336551354,
      "grad_norm": 0.18888010985306866,
      "learning_rate": 1.75548925814917e-06,
      "loss": 0.0103,
      "step": 8416
    },
    {
      "epoch": 2.2571735049611155,
      "grad_norm": 0.19608193591592452,
      "learning_rate": 1.7543023136819144e-06,
      "loss": 0.0097,
      "step": 8417
    },
    {
      "epoch": 2.257441673370877,
      "grad_norm": 0.29292365143455823,
      "learning_rate": 1.753115685239627e-06,
      "loss": 0.0125,
      "step": 8418
    },
    {
      "epoch": 2.2577098417806383,
      "grad_norm": 0.27850452980986307,
      "learning_rate": 1.7519293729378429e-06,
      "loss": 0.0143,
      "step": 8419
    },
    {
      "epoch": 2.2579780101903997,
      "grad_norm": 0.24212283742353258,
      "learning_rate": 1.750743376892073e-06,
      "loss": 0.0098,
      "step": 8420
    },
    {
      "epoch": 2.258246178600161,
      "grad_norm": 0.217117261100985,
      "learning_rate": 1.7495576972177952e-06,
      "loss": 0.0111,
      "step": 8421
    },
    {
      "epoch": 2.258514347009922,
      "grad_norm": 0.19667290683446428,
      "learning_rate": 1.7483723340304547e-06,
      "loss": 0.011,
      "step": 8422
    },
    {
      "epoch": 2.2587825154196834,
      "grad_norm": 0.4082059929411152,
      "learning_rate": 1.747187287445467e-06,
      "loss": 0.0186,
      "step": 8423
    },
    {
      "epoch": 2.259050683829445,
      "grad_norm": 0.18226474843049434,
      "learning_rate": 1.7460025575782158e-06,
      "loss": 0.0093,
      "step": 8424
    },
    {
      "epoch": 2.2593188522392063,
      "grad_norm": 0.2087185802700878,
      "learning_rate": 1.744818144544056e-06,
      "loss": 0.0108,
      "step": 8425
    },
    {
      "epoch": 2.2595870206489677,
      "grad_norm": 0.24860243094253248,
      "learning_rate": 1.743634048458313e-06,
      "loss": 0.0173,
      "step": 8426
    },
    {
      "epoch": 2.259855189058729,
      "grad_norm": 0.37384976359409355,
      "learning_rate": 1.7424502694362755e-06,
      "loss": 0.0206,
      "step": 8427
    },
    {
      "epoch": 2.26012335746849,
      "grad_norm": 0.21340279001700604,
      "learning_rate": 1.7412668075932082e-06,
      "loss": 0.0129,
      "step": 8428
    },
    {
      "epoch": 2.2603915258782514,
      "grad_norm": 0.26589111252871844,
      "learning_rate": 1.7400836630443386e-06,
      "loss": 0.0204,
      "step": 8429
    },
    {
      "epoch": 2.260659694288013,
      "grad_norm": 0.2154091391984328,
      "learning_rate": 1.7389008359048677e-06,
      "loss": 0.0119,
      "step": 8430
    },
    {
      "epoch": 2.2609278626977742,
      "grad_norm": 0.2795333436193349,
      "learning_rate": 1.7377183262899662e-06,
      "loss": 0.0138,
      "step": 8431
    },
    {
      "epoch": 2.2611960311075356,
      "grad_norm": 0.4037599656868992,
      "learning_rate": 1.736536134314768e-06,
      "loss": 0.0163,
      "step": 8432
    },
    {
      "epoch": 2.261464199517297,
      "grad_norm": 0.22674526989891602,
      "learning_rate": 1.735354260094384e-06,
      "loss": 0.0124,
      "step": 8433
    },
    {
      "epoch": 2.261732367927058,
      "grad_norm": 0.327041357828816,
      "learning_rate": 1.7341727037438866e-06,
      "loss": 0.0133,
      "step": 8434
    },
    {
      "epoch": 2.2620005363368194,
      "grad_norm": 0.2737326461304354,
      "learning_rate": 1.7329914653783241e-06,
      "loss": 0.0146,
      "step": 8435
    },
    {
      "epoch": 2.262268704746581,
      "grad_norm": 0.27410649926241926,
      "learning_rate": 1.7318105451127087e-06,
      "loss": 0.0141,
      "step": 8436
    },
    {
      "epoch": 2.262536873156342,
      "grad_norm": 0.19795372617334916,
      "learning_rate": 1.7306299430620222e-06,
      "loss": 0.0119,
      "step": 8437
    },
    {
      "epoch": 2.2628050415661036,
      "grad_norm": 0.4027353533822718,
      "learning_rate": 1.7294496593412198e-06,
      "loss": 0.0203,
      "step": 8438
    },
    {
      "epoch": 2.263073209975865,
      "grad_norm": 0.2637329834433796,
      "learning_rate": 1.7282696940652188e-06,
      "loss": 0.018,
      "step": 8439
    },
    {
      "epoch": 2.263341378385626,
      "grad_norm": 0.22831583786030787,
      "learning_rate": 1.7270900473489138e-06,
      "loss": 0.0104,
      "step": 8440
    },
    {
      "epoch": 2.2636095467953874,
      "grad_norm": 0.23113302301453473,
      "learning_rate": 1.7259107193071594e-06,
      "loss": 0.0168,
      "step": 8441
    },
    {
      "epoch": 2.263877715205149,
      "grad_norm": 0.20839182480001603,
      "learning_rate": 1.7247317100547855e-06,
      "loss": 0.0121,
      "step": 8442
    },
    {
      "epoch": 2.26414588361491,
      "grad_norm": 0.3121122235370847,
      "learning_rate": 1.7235530197065913e-06,
      "loss": 0.0245,
      "step": 8443
    },
    {
      "epoch": 2.2644140520246716,
      "grad_norm": 0.2634070448337121,
      "learning_rate": 1.722374648377339e-06,
      "loss": 0.0136,
      "step": 8444
    },
    {
      "epoch": 2.264682220434433,
      "grad_norm": 0.24115031838263482,
      "learning_rate": 1.7211965961817672e-06,
      "loss": 0.0119,
      "step": 8445
    },
    {
      "epoch": 2.264950388844194,
      "grad_norm": 0.22925340229063987,
      "learning_rate": 1.720018863234576e-06,
      "loss": 0.0122,
      "step": 8446
    },
    {
      "epoch": 2.2652185572539554,
      "grad_norm": 0.19698281015379107,
      "learning_rate": 1.7188414496504413e-06,
      "loss": 0.0134,
      "step": 8447
    },
    {
      "epoch": 2.265486725663717,
      "grad_norm": 0.20768577143007977,
      "learning_rate": 1.7176643555440036e-06,
      "loss": 0.0128,
      "step": 8448
    },
    {
      "epoch": 2.265754894073478,
      "grad_norm": 0.22495302651168775,
      "learning_rate": 1.7164875810298714e-06,
      "loss": 0.0105,
      "step": 8449
    },
    {
      "epoch": 2.2660230624832396,
      "grad_norm": 0.46937709758612356,
      "learning_rate": 1.715311126222628e-06,
      "loss": 0.013,
      "step": 8450
    },
    {
      "epoch": 2.266291230893001,
      "grad_norm": 0.23662344246469108,
      "learning_rate": 1.714134991236817e-06,
      "loss": 0.0165,
      "step": 8451
    },
    {
      "epoch": 2.266559399302762,
      "grad_norm": 0.3666947739177268,
      "learning_rate": 1.7129591761869585e-06,
      "loss": 0.014,
      "step": 8452
    },
    {
      "epoch": 2.2668275677125234,
      "grad_norm": 0.1530048273735077,
      "learning_rate": 1.71178368118754e-06,
      "loss": 0.0063,
      "step": 8453
    },
    {
      "epoch": 2.2670957361222848,
      "grad_norm": 0.24259126111725363,
      "learning_rate": 1.7106085063530125e-06,
      "loss": 0.0168,
      "step": 8454
    },
    {
      "epoch": 2.267363904532046,
      "grad_norm": 0.1991659782595108,
      "learning_rate": 1.7094336517978033e-06,
      "loss": 0.0126,
      "step": 8455
    },
    {
      "epoch": 2.2676320729418076,
      "grad_norm": 0.2532396945980159,
      "learning_rate": 1.7082591176363018e-06,
      "loss": 0.0111,
      "step": 8456
    },
    {
      "epoch": 2.267900241351569,
      "grad_norm": 0.30459714695482926,
      "learning_rate": 1.707084903982872e-06,
      "loss": 0.0267,
      "step": 8457
    },
    {
      "epoch": 2.26816840976133,
      "grad_norm": 0.2433995567309358,
      "learning_rate": 1.7059110109518407e-06,
      "loss": 0.0106,
      "step": 8458
    },
    {
      "epoch": 2.2684365781710913,
      "grad_norm": 0.28978017889972923,
      "learning_rate": 1.7047374386575104e-06,
      "loss": 0.0162,
      "step": 8459
    },
    {
      "epoch": 2.2687047465808527,
      "grad_norm": 0.29282937259234343,
      "learning_rate": 1.7035641872141474e-06,
      "loss": 0.023,
      "step": 8460
    },
    {
      "epoch": 2.268972914990614,
      "grad_norm": 0.2201059834557375,
      "learning_rate": 1.702391256735985e-06,
      "loss": 0.0093,
      "step": 8461
    },
    {
      "epoch": 2.2692410834003756,
      "grad_norm": 0.22789475962591008,
      "learning_rate": 1.7012186473372317e-06,
      "loss": 0.0113,
      "step": 8462
    },
    {
      "epoch": 2.2695092518101365,
      "grad_norm": 0.273189725525553,
      "learning_rate": 1.7000463591320621e-06,
      "loss": 0.0178,
      "step": 8463
    },
    {
      "epoch": 2.269777420219898,
      "grad_norm": 0.20273513360110934,
      "learning_rate": 1.6988743922346156e-06,
      "loss": 0.0099,
      "step": 8464
    },
    {
      "epoch": 2.2700455886296593,
      "grad_norm": 0.29250991970413154,
      "learning_rate": 1.6977027467590068e-06,
      "loss": 0.0216,
      "step": 8465
    },
    {
      "epoch": 2.2703137570394207,
      "grad_norm": 0.2723107381632108,
      "learning_rate": 1.6965314228193126e-06,
      "loss": 0.0192,
      "step": 8466
    },
    {
      "epoch": 2.270581925449182,
      "grad_norm": 0.20941681535076634,
      "learning_rate": 1.6953604205295848e-06,
      "loss": 0.0133,
      "step": 8467
    },
    {
      "epoch": 2.2708500938589435,
      "grad_norm": 0.27709954285805866,
      "learning_rate": 1.6941897400038376e-06,
      "loss": 0.015,
      "step": 8468
    },
    {
      "epoch": 2.271118262268705,
      "grad_norm": 0.23605061823897994,
      "learning_rate": 1.6930193813560586e-06,
      "loss": 0.0127,
      "step": 8469
    },
    {
      "epoch": 2.271386430678466,
      "grad_norm": 0.2826037160527682,
      "learning_rate": 1.6918493447002043e-06,
      "loss": 0.0145,
      "step": 8470
    },
    {
      "epoch": 2.2716545990882273,
      "grad_norm": 0.26291507719180107,
      "learning_rate": 1.6906796301501965e-06,
      "loss": 0.0135,
      "step": 8471
    },
    {
      "epoch": 2.2719227674979887,
      "grad_norm": 0.24340630299311553,
      "learning_rate": 1.6895102378199252e-06,
      "loss": 0.0151,
      "step": 8472
    },
    {
      "epoch": 2.27219093590775,
      "grad_norm": 0.2568415515716202,
      "learning_rate": 1.6883411678232548e-06,
      "loss": 0.0202,
      "step": 8473
    },
    {
      "epoch": 2.2724591043175115,
      "grad_norm": 0.16282139438450577,
      "learning_rate": 1.6871724202740103e-06,
      "loss": 0.0083,
      "step": 8474
    },
    {
      "epoch": 2.2727272727272725,
      "grad_norm": 0.2775997368887086,
      "learning_rate": 1.6860039952859941e-06,
      "loss": 0.0132,
      "step": 8475
    },
    {
      "epoch": 2.272995441137034,
      "grad_norm": 0.23970074685438542,
      "learning_rate": 1.6848358929729687e-06,
      "loss": 0.0114,
      "step": 8476
    },
    {
      "epoch": 2.2732636095467953,
      "grad_norm": 0.25299634110264096,
      "learning_rate": 1.683668113448672e-06,
      "loss": 0.0178,
      "step": 8477
    },
    {
      "epoch": 2.2735317779565567,
      "grad_norm": 0.24066257414749712,
      "learning_rate": 1.6825006568268048e-06,
      "loss": 0.0135,
      "step": 8478
    },
    {
      "epoch": 2.273799946366318,
      "grad_norm": 0.25926075723699976,
      "learning_rate": 1.6813335232210416e-06,
      "loss": 0.0129,
      "step": 8479
    },
    {
      "epoch": 2.2740681147760795,
      "grad_norm": 0.3643012234394694,
      "learning_rate": 1.6801667127450233e-06,
      "loss": 0.0142,
      "step": 8480
    },
    {
      "epoch": 2.274336283185841,
      "grad_norm": 0.22886255552419288,
      "learning_rate": 1.6790002255123567e-06,
      "loss": 0.0122,
      "step": 8481
    },
    {
      "epoch": 2.274604451595602,
      "grad_norm": 0.2621957705359423,
      "learning_rate": 1.6778340616366228e-06,
      "loss": 0.0124,
      "step": 8482
    },
    {
      "epoch": 2.2748726200053633,
      "grad_norm": 0.3351720043750013,
      "learning_rate": 1.6766682212313662e-06,
      "loss": 0.0142,
      "step": 8483
    },
    {
      "epoch": 2.2751407884151247,
      "grad_norm": 0.22377702206473182,
      "learning_rate": 1.6755027044100996e-06,
      "loss": 0.0128,
      "step": 8484
    },
    {
      "epoch": 2.275408956824886,
      "grad_norm": 0.29362666190855174,
      "learning_rate": 1.6743375112863103e-06,
      "loss": 0.0193,
      "step": 8485
    },
    {
      "epoch": 2.2756771252346475,
      "grad_norm": 0.3482797014707081,
      "learning_rate": 1.6731726419734461e-06,
      "loss": 0.0231,
      "step": 8486
    },
    {
      "epoch": 2.2759452936444085,
      "grad_norm": 0.2406188999772896,
      "learning_rate": 1.672008096584931e-06,
      "loss": 0.0177,
      "step": 8487
    },
    {
      "epoch": 2.27621346205417,
      "grad_norm": 0.24758160982945265,
      "learning_rate": 1.6708438752341506e-06,
      "loss": 0.015,
      "step": 8488
    },
    {
      "epoch": 2.2764816304639313,
      "grad_norm": 0.27442449128690044,
      "learning_rate": 1.669679978034463e-06,
      "loss": 0.0132,
      "step": 8489
    },
    {
      "epoch": 2.2767497988736927,
      "grad_norm": 0.2841118250840289,
      "learning_rate": 1.668516405099196e-06,
      "loss": 0.0144,
      "step": 8490
    },
    {
      "epoch": 2.277017967283454,
      "grad_norm": 0.1854022618083752,
      "learning_rate": 1.6673531565416395e-06,
      "loss": 0.008,
      "step": 8491
    },
    {
      "epoch": 2.2772861356932155,
      "grad_norm": 0.22814781678986457,
      "learning_rate": 1.66619023247506e-06,
      "loss": 0.015,
      "step": 8492
    },
    {
      "epoch": 2.277554304102977,
      "grad_norm": 0.20524171405366431,
      "learning_rate": 1.6650276330126853e-06,
      "loss": 0.0143,
      "step": 8493
    },
    {
      "epoch": 2.277822472512738,
      "grad_norm": 0.21054793591251186,
      "learning_rate": 1.6638653582677166e-06,
      "loss": 0.009,
      "step": 8494
    },
    {
      "epoch": 2.2780906409224992,
      "grad_norm": 0.2718074338655996,
      "learning_rate": 1.6627034083533216e-06,
      "loss": 0.0167,
      "step": 8495
    },
    {
      "epoch": 2.2783588093322606,
      "grad_norm": 0.28732617270328803,
      "learning_rate": 1.661541783382633e-06,
      "loss": 0.0123,
      "step": 8496
    },
    {
      "epoch": 2.278626977742022,
      "grad_norm": 0.4803803441162348,
      "learning_rate": 1.6603804834687597e-06,
      "loss": 0.0255,
      "step": 8497
    },
    {
      "epoch": 2.2788951461517835,
      "grad_norm": 0.24658436804267175,
      "learning_rate": 1.6592195087247703e-06,
      "loss": 0.0136,
      "step": 8498
    },
    {
      "epoch": 2.2791633145615444,
      "grad_norm": 0.2600345584599022,
      "learning_rate": 1.658058859263708e-06,
      "loss": 0.0148,
      "step": 8499
    },
    {
      "epoch": 2.279431482971306,
      "grad_norm": 0.2806213527335778,
      "learning_rate": 1.6568985351985833e-06,
      "loss": 0.0182,
      "step": 8500
    },
    {
      "epoch": 2.2796996513810672,
      "grad_norm": 0.21882891356023207,
      "learning_rate": 1.6557385366423706e-06,
      "loss": 0.0124,
      "step": 8501
    },
    {
      "epoch": 2.2799678197908286,
      "grad_norm": 0.2590529117257352,
      "learning_rate": 1.6545788637080195e-06,
      "loss": 0.0152,
      "step": 8502
    },
    {
      "epoch": 2.28023598820059,
      "grad_norm": 0.32070816632245663,
      "learning_rate": 1.6534195165084417e-06,
      "loss": 0.0209,
      "step": 8503
    },
    {
      "epoch": 2.2805041566103514,
      "grad_norm": 0.2245510799972336,
      "learning_rate": 1.6522604951565213e-06,
      "loss": 0.0127,
      "step": 8504
    },
    {
      "epoch": 2.280772325020113,
      "grad_norm": 0.2627587646071397,
      "learning_rate": 1.6511017997651069e-06,
      "loss": 0.0114,
      "step": 8505
    },
    {
      "epoch": 2.281040493429874,
      "grad_norm": 0.24922569063116717,
      "learning_rate": 1.6499434304470192e-06,
      "loss": 0.0147,
      "step": 8506
    },
    {
      "epoch": 2.281308661839635,
      "grad_norm": 0.28996286727518145,
      "learning_rate": 1.6487853873150489e-06,
      "loss": 0.0164,
      "step": 8507
    },
    {
      "epoch": 2.2815768302493966,
      "grad_norm": 0.28964302425281013,
      "learning_rate": 1.6476276704819443e-06,
      "loss": 0.0173,
      "step": 8508
    },
    {
      "epoch": 2.281844998659158,
      "grad_norm": 0.21446319948636575,
      "learning_rate": 1.6464702800604337e-06,
      "loss": 0.0077,
      "step": 8509
    },
    {
      "epoch": 2.2821131670689194,
      "grad_norm": 0.26190773480088075,
      "learning_rate": 1.6453132161632095e-06,
      "loss": 0.0159,
      "step": 8510
    },
    {
      "epoch": 2.2823813354786804,
      "grad_norm": 0.20840110109380103,
      "learning_rate": 1.64415647890293e-06,
      "loss": 0.0133,
      "step": 8511
    },
    {
      "epoch": 2.282649503888442,
      "grad_norm": 0.20761017796542586,
      "learning_rate": 1.6430000683922258e-06,
      "loss": 0.0088,
      "step": 8512
    },
    {
      "epoch": 2.282917672298203,
      "grad_norm": 0.21272298851348512,
      "learning_rate": 1.6418439847436907e-06,
      "loss": 0.0148,
      "step": 8513
    },
    {
      "epoch": 2.2831858407079646,
      "grad_norm": 0.23052055012735603,
      "learning_rate": 1.6406882280698926e-06,
      "loss": 0.0131,
      "step": 8514
    },
    {
      "epoch": 2.283454009117726,
      "grad_norm": 0.21278876835311675,
      "learning_rate": 1.6395327984833614e-06,
      "loss": 0.0115,
      "step": 8515
    },
    {
      "epoch": 2.2837221775274874,
      "grad_norm": 0.27951236105264277,
      "learning_rate": 1.6383776960965997e-06,
      "loss": 0.0145,
      "step": 8516
    },
    {
      "epoch": 2.283990345937249,
      "grad_norm": 0.39106713856218195,
      "learning_rate": 1.6372229210220785e-06,
      "loss": 0.0292,
      "step": 8517
    },
    {
      "epoch": 2.2842585143470098,
      "grad_norm": 0.27699602435211956,
      "learning_rate": 1.636068473372232e-06,
      "loss": 0.0139,
      "step": 8518
    },
    {
      "epoch": 2.284526682756771,
      "grad_norm": 0.21526888536503336,
      "learning_rate": 1.6349143532594702e-06,
      "loss": 0.0084,
      "step": 8519
    },
    {
      "epoch": 2.2847948511665326,
      "grad_norm": 0.26269675252639224,
      "learning_rate": 1.6337605607961604e-06,
      "loss": 0.0166,
      "step": 8520
    },
    {
      "epoch": 2.285063019576294,
      "grad_norm": 0.21093303678308473,
      "learning_rate": 1.6326070960946478e-06,
      "loss": 0.0142,
      "step": 8521
    },
    {
      "epoch": 2.2853311879860554,
      "grad_norm": 0.22947905171018737,
      "learning_rate": 1.6314539592672435e-06,
      "loss": 0.0155,
      "step": 8522
    },
    {
      "epoch": 2.2855993563958164,
      "grad_norm": 0.3304083339666778,
      "learning_rate": 1.6303011504262223e-06,
      "loss": 0.0179,
      "step": 8523
    },
    {
      "epoch": 2.2858675248055778,
      "grad_norm": 0.2898016325597004,
      "learning_rate": 1.6291486696838332e-06,
      "loss": 0.0145,
      "step": 8524
    },
    {
      "epoch": 2.286135693215339,
      "grad_norm": 0.2439958636102099,
      "learning_rate": 1.627996517152287e-06,
      "loss": 0.0105,
      "step": 8525
    },
    {
      "epoch": 2.2864038616251006,
      "grad_norm": 0.2693337846878799,
      "learning_rate": 1.6268446929437677e-06,
      "loss": 0.0143,
      "step": 8526
    },
    {
      "epoch": 2.286672030034862,
      "grad_norm": 0.19800695752480713,
      "learning_rate": 1.6256931971704266e-06,
      "loss": 0.0091,
      "step": 8527
    },
    {
      "epoch": 2.2869401984446234,
      "grad_norm": 0.2736790571679425,
      "learning_rate": 1.6245420299443788e-06,
      "loss": 0.0152,
      "step": 8528
    },
    {
      "epoch": 2.2872083668543848,
      "grad_norm": 0.2796611359984947,
      "learning_rate": 1.6233911913777135e-06,
      "loss": 0.015,
      "step": 8529
    },
    {
      "epoch": 2.2874765352641457,
      "grad_norm": 0.2698950595977366,
      "learning_rate": 1.622240681582481e-06,
      "loss": 0.0122,
      "step": 8530
    },
    {
      "epoch": 2.287744703673907,
      "grad_norm": 0.23332941311845956,
      "learning_rate": 1.6210905006707079e-06,
      "loss": 0.0142,
      "step": 8531
    },
    {
      "epoch": 2.2880128720836685,
      "grad_norm": 0.23352858651534042,
      "learning_rate": 1.6199406487543817e-06,
      "loss": 0.0106,
      "step": 8532
    },
    {
      "epoch": 2.28828104049343,
      "grad_norm": 0.3628079292751081,
      "learning_rate": 1.6187911259454586e-06,
      "loss": 0.0232,
      "step": 8533
    },
    {
      "epoch": 2.2885492089031914,
      "grad_norm": 0.21606384361387734,
      "learning_rate": 1.6176419323558689e-06,
      "loss": 0.0104,
      "step": 8534
    },
    {
      "epoch": 2.2888173773129523,
      "grad_norm": 0.23478407077827018,
      "learning_rate": 1.6164930680975021e-06,
      "loss": 0.0117,
      "step": 8535
    },
    {
      "epoch": 2.2890855457227137,
      "grad_norm": 0.27461262288016375,
      "learning_rate": 1.6153445332822243e-06,
      "loss": 0.0193,
      "step": 8536
    },
    {
      "epoch": 2.289353714132475,
      "grad_norm": 0.27065145211918784,
      "learning_rate": 1.6141963280218615e-06,
      "loss": 0.0155,
      "step": 8537
    },
    {
      "epoch": 2.2896218825422365,
      "grad_norm": 0.5295796368434448,
      "learning_rate": 1.6130484524282125e-06,
      "loss": 0.0219,
      "step": 8538
    },
    {
      "epoch": 2.289890050951998,
      "grad_norm": 0.16323762444684314,
      "learning_rate": 1.611900906613046e-06,
      "loss": 0.0098,
      "step": 8539
    },
    {
      "epoch": 2.2901582193617593,
      "grad_norm": 0.2892312485584017,
      "learning_rate": 1.6107536906880904e-06,
      "loss": 0.0188,
      "step": 8540
    },
    {
      "epoch": 2.2904263877715207,
      "grad_norm": 0.2436998962697862,
      "learning_rate": 1.6096068047650521e-06,
      "loss": 0.0136,
      "step": 8541
    },
    {
      "epoch": 2.2906945561812817,
      "grad_norm": 0.1988970918027792,
      "learning_rate": 1.6084602489555963e-06,
      "loss": 0.0098,
      "step": 8542
    },
    {
      "epoch": 2.290962724591043,
      "grad_norm": 0.2960631853201012,
      "learning_rate": 1.607314023371363e-06,
      "loss": 0.0148,
      "step": 8543
    },
    {
      "epoch": 2.2912308930008045,
      "grad_norm": 0.24372428053667514,
      "learning_rate": 1.6061681281239554e-06,
      "loss": 0.0108,
      "step": 8544
    },
    {
      "epoch": 2.291499061410566,
      "grad_norm": 0.24917036764393638,
      "learning_rate": 1.605022563324946e-06,
      "loss": 0.0146,
      "step": 8545
    },
    {
      "epoch": 2.2917672298203273,
      "grad_norm": 0.4487712075695106,
      "learning_rate": 1.6038773290858773e-06,
      "loss": 0.015,
      "step": 8546
    },
    {
      "epoch": 2.2920353982300883,
      "grad_norm": 0.17495563739516928,
      "learning_rate": 1.6027324255182547e-06,
      "loss": 0.0092,
      "step": 8547
    },
    {
      "epoch": 2.2923035666398497,
      "grad_norm": 0.2066927858919382,
      "learning_rate": 1.6015878527335566e-06,
      "loss": 0.0117,
      "step": 8548
    },
    {
      "epoch": 2.292571735049611,
      "grad_norm": 0.23330762011233286,
      "learning_rate": 1.600443610843228e-06,
      "loss": 0.0127,
      "step": 8549
    },
    {
      "epoch": 2.2928399034593725,
      "grad_norm": 0.2403612175731807,
      "learning_rate": 1.599299699958678e-06,
      "loss": 0.0139,
      "step": 8550
    },
    {
      "epoch": 2.293108071869134,
      "grad_norm": 0.2096230403278886,
      "learning_rate": 1.5981561201912893e-06,
      "loss": 0.0107,
      "step": 8551
    },
    {
      "epoch": 2.2933762402788953,
      "grad_norm": 0.18043890025319256,
      "learning_rate": 1.5970128716524052e-06,
      "loss": 0.0098,
      "step": 8552
    },
    {
      "epoch": 2.2936444086886567,
      "grad_norm": 0.28130873570064524,
      "learning_rate": 1.595869954453345e-06,
      "loss": 0.0125,
      "step": 8553
    },
    {
      "epoch": 2.2939125770984177,
      "grad_norm": 0.2985108464714816,
      "learning_rate": 1.5947273687053882e-06,
      "loss": 0.0196,
      "step": 8554
    },
    {
      "epoch": 2.294180745508179,
      "grad_norm": 0.19109555128876005,
      "learning_rate": 1.5935851145197867e-06,
      "loss": 0.0111,
      "step": 8555
    },
    {
      "epoch": 2.2944489139179405,
      "grad_norm": 0.2799930477074702,
      "learning_rate": 1.5924431920077615e-06,
      "loss": 0.0168,
      "step": 8556
    },
    {
      "epoch": 2.294717082327702,
      "grad_norm": 0.2061159324620164,
      "learning_rate": 1.5913016012804927e-06,
      "loss": 0.0114,
      "step": 8557
    },
    {
      "epoch": 2.2949852507374633,
      "grad_norm": 0.29094974679386515,
      "learning_rate": 1.5901603424491375e-06,
      "loss": 0.0157,
      "step": 8558
    },
    {
      "epoch": 2.2952534191472242,
      "grad_norm": 0.2748932428405859,
      "learning_rate": 1.5890194156248178e-06,
      "loss": 0.0185,
      "step": 8559
    },
    {
      "epoch": 2.2955215875569857,
      "grad_norm": 0.2340296208759539,
      "learning_rate": 1.5878788209186203e-06,
      "loss": 0.011,
      "step": 8560
    },
    {
      "epoch": 2.295789755966747,
      "grad_norm": 0.30880371525182687,
      "learning_rate": 1.5867385584416052e-06,
      "loss": 0.0146,
      "step": 8561
    },
    {
      "epoch": 2.2960579243765085,
      "grad_norm": 0.24634459825291338,
      "learning_rate": 1.585598628304792e-06,
      "loss": 0.0141,
      "step": 8562
    },
    {
      "epoch": 2.29632609278627,
      "grad_norm": 0.23238923951575596,
      "learning_rate": 1.5844590306191776e-06,
      "loss": 0.012,
      "step": 8563
    },
    {
      "epoch": 2.2965942611960313,
      "grad_norm": 0.22994407720841128,
      "learning_rate": 1.583319765495717e-06,
      "loss": 0.0108,
      "step": 8564
    },
    {
      "epoch": 2.2968624296057927,
      "grad_norm": 0.22202755640472555,
      "learning_rate": 1.5821808330453403e-06,
      "loss": 0.0138,
      "step": 8565
    },
    {
      "epoch": 2.2971305980155536,
      "grad_norm": 0.2306430729146389,
      "learning_rate": 1.5810422333789432e-06,
      "loss": 0.0096,
      "step": 8566
    },
    {
      "epoch": 2.297398766425315,
      "grad_norm": 0.2519542834760024,
      "learning_rate": 1.5799039666073852e-06,
      "loss": 0.0148,
      "step": 8567
    },
    {
      "epoch": 2.2976669348350764,
      "grad_norm": 0.20352706729632353,
      "learning_rate": 1.5787660328414993e-06,
      "loss": 0.0136,
      "step": 8568
    },
    {
      "epoch": 2.297935103244838,
      "grad_norm": 0.19152395116537818,
      "learning_rate": 1.5776284321920814e-06,
      "loss": 0.0113,
      "step": 8569
    },
    {
      "epoch": 2.2982032716545993,
      "grad_norm": 0.2470719405727053,
      "learning_rate": 1.5764911647698956e-06,
      "loss": 0.0135,
      "step": 8570
    },
    {
      "epoch": 2.29847144006436,
      "grad_norm": 0.24031869693112035,
      "learning_rate": 1.5753542306856774e-06,
      "loss": 0.0119,
      "step": 8571
    },
    {
      "epoch": 2.2987396084741216,
      "grad_norm": 0.2096520176333196,
      "learning_rate": 1.574217630050124e-06,
      "loss": 0.0126,
      "step": 8572
    },
    {
      "epoch": 2.299007776883883,
      "grad_norm": 0.477953625726046,
      "learning_rate": 1.5730813629739055e-06,
      "loss": 0.0121,
      "step": 8573
    },
    {
      "epoch": 2.2992759452936444,
      "grad_norm": 0.2057460610459564,
      "learning_rate": 1.5719454295676555e-06,
      "loss": 0.0104,
      "step": 8574
    },
    {
      "epoch": 2.299544113703406,
      "grad_norm": 0.18764641461850087,
      "learning_rate": 1.5708098299419778e-06,
      "loss": 0.0114,
      "step": 8575
    },
    {
      "epoch": 2.2998122821131672,
      "grad_norm": 0.23646356723862264,
      "learning_rate": 1.569674564207444e-06,
      "loss": 0.011,
      "step": 8576
    },
    {
      "epoch": 2.3000804505229286,
      "grad_norm": 0.25307598388373187,
      "learning_rate": 1.5685396324745888e-06,
      "loss": 0.0124,
      "step": 8577
    },
    {
      "epoch": 2.3003486189326896,
      "grad_norm": 0.25461928705345865,
      "learning_rate": 1.567405034853921e-06,
      "loss": 0.0116,
      "step": 8578
    },
    {
      "epoch": 2.300616787342451,
      "grad_norm": 0.2671985608749617,
      "learning_rate": 1.56627077145591e-06,
      "loss": 0.0144,
      "step": 8579
    },
    {
      "epoch": 2.3008849557522124,
      "grad_norm": 0.2755625660026213,
      "learning_rate": 1.565136842390999e-06,
      "loss": 0.0199,
      "step": 8580
    },
    {
      "epoch": 2.301153124161974,
      "grad_norm": 0.2247513374118995,
      "learning_rate": 1.5640032477695942e-06,
      "loss": 0.0135,
      "step": 8581
    },
    {
      "epoch": 2.3014212925717352,
      "grad_norm": 0.2494479077495726,
      "learning_rate": 1.5628699877020698e-06,
      "loss": 0.0145,
      "step": 8582
    },
    {
      "epoch": 2.301689460981496,
      "grad_norm": 0.3995341164074962,
      "learning_rate": 1.5617370622987703e-06,
      "loss": 0.018,
      "step": 8583
    },
    {
      "epoch": 2.3019576293912576,
      "grad_norm": 0.17771201803975975,
      "learning_rate": 1.560604471670003e-06,
      "loss": 0.0082,
      "step": 8584
    },
    {
      "epoch": 2.302225797801019,
      "grad_norm": 0.262371696980084,
      "learning_rate": 1.5594722159260473e-06,
      "loss": 0.0127,
      "step": 8585
    },
    {
      "epoch": 2.3024939662107804,
      "grad_norm": 0.2607488845230623,
      "learning_rate": 1.5583402951771492e-06,
      "loss": 0.0096,
      "step": 8586
    },
    {
      "epoch": 2.302762134620542,
      "grad_norm": 0.2438573230479588,
      "learning_rate": 1.557208709533517e-06,
      "loss": 0.0115,
      "step": 8587
    },
    {
      "epoch": 2.303030303030303,
      "grad_norm": 0.27221245642541997,
      "learning_rate": 1.556077459105334e-06,
      "loss": 0.0155,
      "step": 8588
    },
    {
      "epoch": 2.3032984714400646,
      "grad_norm": 0.1935712054688396,
      "learning_rate": 1.5549465440027444e-06,
      "loss": 0.0113,
      "step": 8589
    },
    {
      "epoch": 2.3035666398498256,
      "grad_norm": 0.2583407282458997,
      "learning_rate": 1.5538159643358642e-06,
      "loss": 0.0153,
      "step": 8590
    },
    {
      "epoch": 2.303834808259587,
      "grad_norm": 0.20017089471870775,
      "learning_rate": 1.5526857202147727e-06,
      "loss": 0.0121,
      "step": 8591
    },
    {
      "epoch": 2.3041029766693484,
      "grad_norm": 0.24224659808717244,
      "learning_rate": 1.551555811749522e-06,
      "loss": 0.014,
      "step": 8592
    },
    {
      "epoch": 2.30437114507911,
      "grad_norm": 0.251945168325294,
      "learning_rate": 1.5504262390501268e-06,
      "loss": 0.0154,
      "step": 8593
    },
    {
      "epoch": 2.304639313488871,
      "grad_norm": 0.18602493445541723,
      "learning_rate": 1.549297002226568e-06,
      "loss": 0.0091,
      "step": 8594
    },
    {
      "epoch": 2.304907481898632,
      "grad_norm": 0.21581372215885616,
      "learning_rate": 1.548168101388799e-06,
      "loss": 0.0161,
      "step": 8595
    },
    {
      "epoch": 2.3051756503083936,
      "grad_norm": 0.3066193808260382,
      "learning_rate": 1.547039536646739e-06,
      "loss": 0.0156,
      "step": 8596
    },
    {
      "epoch": 2.305443818718155,
      "grad_norm": 0.15517015279682161,
      "learning_rate": 1.54591130811027e-06,
      "loss": 0.0068,
      "step": 8597
    },
    {
      "epoch": 2.3057119871279164,
      "grad_norm": 0.3332669783422645,
      "learning_rate": 1.5447834158892483e-06,
      "loss": 0.0229,
      "step": 8598
    },
    {
      "epoch": 2.3059801555376778,
      "grad_norm": 0.18656299848711694,
      "learning_rate": 1.543655860093491e-06,
      "loss": 0.0119,
      "step": 8599
    },
    {
      "epoch": 2.306248323947439,
      "grad_norm": 0.2690538790953686,
      "learning_rate": 1.542528640832787e-06,
      "loss": 0.018,
      "step": 8600
    },
    {
      "epoch": 2.3065164923572,
      "grad_norm": 0.24971500485351675,
      "learning_rate": 1.5414017582168883e-06,
      "loss": 0.0186,
      "step": 8601
    },
    {
      "epoch": 2.3067846607669615,
      "grad_norm": 0.3341652588727936,
      "learning_rate": 1.5402752123555181e-06,
      "loss": 0.0156,
      "step": 8602
    },
    {
      "epoch": 2.307052829176723,
      "grad_norm": 0.26926704644825905,
      "learning_rate": 1.5391490033583674e-06,
      "loss": 0.0152,
      "step": 8603
    },
    {
      "epoch": 2.3073209975864843,
      "grad_norm": 0.23616964014569905,
      "learning_rate": 1.538023131335089e-06,
      "loss": 0.0151,
      "step": 8604
    },
    {
      "epoch": 2.3075891659962458,
      "grad_norm": 0.21770847723912887,
      "learning_rate": 1.536897596395307e-06,
      "loss": 0.0168,
      "step": 8605
    },
    {
      "epoch": 2.307857334406007,
      "grad_norm": 0.19345158612519578,
      "learning_rate": 1.53577239864861e-06,
      "loss": 0.0127,
      "step": 8606
    },
    {
      "epoch": 2.308125502815768,
      "grad_norm": 0.47884970037923785,
      "learning_rate": 1.5346475382045578e-06,
      "loss": 0.0199,
      "step": 8607
    },
    {
      "epoch": 2.3083936712255295,
      "grad_norm": 0.7507193422746499,
      "learning_rate": 1.5335230151726755e-06,
      "loss": 0.0177,
      "step": 8608
    },
    {
      "epoch": 2.308661839635291,
      "grad_norm": 0.23263385954289825,
      "learning_rate": 1.5323988296624526e-06,
      "loss": 0.0141,
      "step": 8609
    },
    {
      "epoch": 2.3089300080450523,
      "grad_norm": 0.24689395138637943,
      "learning_rate": 1.5312749817833511e-06,
      "loss": 0.0123,
      "step": 8610
    },
    {
      "epoch": 2.3091981764548137,
      "grad_norm": 0.20075668765169405,
      "learning_rate": 1.5301514716447935e-06,
      "loss": 0.0142,
      "step": 8611
    },
    {
      "epoch": 2.309466344864575,
      "grad_norm": 0.2737769088693984,
      "learning_rate": 1.5290282993561745e-06,
      "loss": 0.0163,
      "step": 8612
    },
    {
      "epoch": 2.309734513274336,
      "grad_norm": 0.2264524107132518,
      "learning_rate": 1.5279054650268571e-06,
      "loss": 0.0137,
      "step": 8613
    },
    {
      "epoch": 2.3100026816840975,
      "grad_norm": 0.26502545491268126,
      "learning_rate": 1.526782968766164e-06,
      "loss": 0.0213,
      "step": 8614
    },
    {
      "epoch": 2.310270850093859,
      "grad_norm": 0.3013048384046786,
      "learning_rate": 1.5256608106833942e-06,
      "loss": 0.018,
      "step": 8615
    },
    {
      "epoch": 2.3105390185036203,
      "grad_norm": 0.4271957220770806,
      "learning_rate": 1.5245389908878067e-06,
      "loss": 0.0137,
      "step": 8616
    },
    {
      "epoch": 2.3108071869133817,
      "grad_norm": 0.2765798965047091,
      "learning_rate": 1.523417509488629e-06,
      "loss": 0.014,
      "step": 8617
    },
    {
      "epoch": 2.311075355323143,
      "grad_norm": 0.21569703900886006,
      "learning_rate": 1.5222963665950597e-06,
      "loss": 0.0114,
      "step": 8618
    },
    {
      "epoch": 2.311343523732904,
      "grad_norm": 0.4386624662922243,
      "learning_rate": 1.5211755623162588e-06,
      "loss": 0.0151,
      "step": 8619
    },
    {
      "epoch": 2.3116116921426655,
      "grad_norm": 0.2995373571648426,
      "learning_rate": 1.5200550967613585e-06,
      "loss": 0.0185,
      "step": 8620
    },
    {
      "epoch": 2.311879860552427,
      "grad_norm": 0.28069588176177623,
      "learning_rate": 1.5189349700394529e-06,
      "loss": 0.0117,
      "step": 8621
    },
    {
      "epoch": 2.3121480289621883,
      "grad_norm": 0.30552362939894545,
      "learning_rate": 1.5178151822596077e-06,
      "loss": 0.0174,
      "step": 8622
    },
    {
      "epoch": 2.3124161973719497,
      "grad_norm": 0.23695501271099612,
      "learning_rate": 1.516695733530854e-06,
      "loss": 0.009,
      "step": 8623
    },
    {
      "epoch": 2.312684365781711,
      "grad_norm": 0.2830322535231299,
      "learning_rate": 1.5155766239621877e-06,
      "loss": 0.0167,
      "step": 8624
    },
    {
      "epoch": 2.312952534191472,
      "grad_norm": 0.2555145572284112,
      "learning_rate": 1.5144578536625758e-06,
      "loss": 0.0143,
      "step": 8625
    },
    {
      "epoch": 2.3132207026012335,
      "grad_norm": 0.2345941727158562,
      "learning_rate": 1.5133394227409476e-06,
      "loss": 0.016,
      "step": 8626
    },
    {
      "epoch": 2.313488871010995,
      "grad_norm": 0.30258556998965835,
      "learning_rate": 1.512221331306204e-06,
      "loss": 0.0215,
      "step": 8627
    },
    {
      "epoch": 2.3137570394207563,
      "grad_norm": 0.2281024957062922,
      "learning_rate": 1.5111035794672096e-06,
      "loss": 0.0086,
      "step": 8628
    },
    {
      "epoch": 2.3140252078305177,
      "grad_norm": 0.22569010762628666,
      "learning_rate": 1.5099861673327953e-06,
      "loss": 0.0124,
      "step": 8629
    },
    {
      "epoch": 2.314293376240279,
      "grad_norm": 0.25352845493748566,
      "learning_rate": 1.5088690950117641e-06,
      "loss": 0.011,
      "step": 8630
    },
    {
      "epoch": 2.31456154465004,
      "grad_norm": 0.25123579416985403,
      "learning_rate": 1.507752362612878e-06,
      "loss": 0.0153,
      "step": 8631
    },
    {
      "epoch": 2.3148297130598015,
      "grad_norm": 0.2758104481989522,
      "learning_rate": 1.5066359702448746e-06,
      "loss": 0.016,
      "step": 8632
    },
    {
      "epoch": 2.315097881469563,
      "grad_norm": 0.27401065533755614,
      "learning_rate": 1.5055199180164503e-06,
      "loss": 0.0138,
      "step": 8633
    },
    {
      "epoch": 2.3153660498793243,
      "grad_norm": 0.2266539853038789,
      "learning_rate": 1.504404206036274e-06,
      "loss": 0.0112,
      "step": 8634
    },
    {
      "epoch": 2.3156342182890857,
      "grad_norm": 0.200067612866986,
      "learning_rate": 1.5032888344129815e-06,
      "loss": 0.0068,
      "step": 8635
    },
    {
      "epoch": 2.3159023866988466,
      "grad_norm": 0.2388457188268222,
      "learning_rate": 1.5021738032551697e-06,
      "loss": 0.0119,
      "step": 8636
    },
    {
      "epoch": 2.316170555108608,
      "grad_norm": 0.2645702212712605,
      "learning_rate": 1.5010591126714096e-06,
      "loss": 0.0148,
      "step": 8637
    },
    {
      "epoch": 2.3164387235183694,
      "grad_norm": 0.2908803805266686,
      "learning_rate": 1.4999447627702329e-06,
      "loss": 0.0188,
      "step": 8638
    },
    {
      "epoch": 2.316706891928131,
      "grad_norm": 0.24610175218367125,
      "learning_rate": 1.4988307536601421e-06,
      "loss": 0.011,
      "step": 8639
    },
    {
      "epoch": 2.3169750603378922,
      "grad_norm": 0.24572047607030081,
      "learning_rate": 1.4977170854496082e-06,
      "loss": 0.0144,
      "step": 8640
    },
    {
      "epoch": 2.3172432287476536,
      "grad_norm": 0.19903839603043907,
      "learning_rate": 1.4966037582470605e-06,
      "loss": 0.0105,
      "step": 8641
    },
    {
      "epoch": 2.317511397157415,
      "grad_norm": 0.25698726669715444,
      "learning_rate": 1.4954907721609046e-06,
      "loss": 0.0125,
      "step": 8642
    },
    {
      "epoch": 2.317779565567176,
      "grad_norm": 0.2505192152892468,
      "learning_rate": 1.4943781272995073e-06,
      "loss": 0.0139,
      "step": 8643
    },
    {
      "epoch": 2.3180477339769374,
      "grad_norm": 0.33031862154365244,
      "learning_rate": 1.493265823771204e-06,
      "loss": 0.0137,
      "step": 8644
    },
    {
      "epoch": 2.318315902386699,
      "grad_norm": 0.5223775938305318,
      "learning_rate": 1.4921538616842983e-06,
      "loss": 0.0235,
      "step": 8645
    },
    {
      "epoch": 2.3185840707964602,
      "grad_norm": 0.24126975842564075,
      "learning_rate": 1.491042241147057e-06,
      "loss": 0.0114,
      "step": 8646
    },
    {
      "epoch": 2.3188522392062216,
      "grad_norm": 0.180003647600566,
      "learning_rate": 1.4899309622677177e-06,
      "loss": 0.0093,
      "step": 8647
    },
    {
      "epoch": 2.3191204076159826,
      "grad_norm": 0.23500087383834684,
      "learning_rate": 1.4888200251544803e-06,
      "loss": 0.0113,
      "step": 8648
    },
    {
      "epoch": 2.319388576025744,
      "grad_norm": 0.25511855656872956,
      "learning_rate": 1.487709429915517e-06,
      "loss": 0.0164,
      "step": 8649
    },
    {
      "epoch": 2.3196567444355054,
      "grad_norm": 0.2983068090196593,
      "learning_rate": 1.4865991766589593e-06,
      "loss": 0.0142,
      "step": 8650
    },
    {
      "epoch": 2.319924912845267,
      "grad_norm": 0.26417925216705035,
      "learning_rate": 1.4854892654929121e-06,
      "loss": 0.0153,
      "step": 8651
    },
    {
      "epoch": 2.320193081255028,
      "grad_norm": 0.3135512194675503,
      "learning_rate": 1.4843796965254475e-06,
      "loss": 0.0162,
      "step": 8652
    },
    {
      "epoch": 2.3204612496647896,
      "grad_norm": 0.340245926696201,
      "learning_rate": 1.483270469864595e-06,
      "loss": 0.0199,
      "step": 8653
    },
    {
      "epoch": 2.320729418074551,
      "grad_norm": 0.27072191202557566,
      "learning_rate": 1.4821615856183597e-06,
      "loss": 0.018,
      "step": 8654
    },
    {
      "epoch": 2.320997586484312,
      "grad_norm": 0.27330535554225177,
      "learning_rate": 1.481053043894713e-06,
      "loss": 0.0127,
      "step": 8655
    },
    {
      "epoch": 2.3212657548940734,
      "grad_norm": 0.23340843466556335,
      "learning_rate": 1.4799448448015874e-06,
      "loss": 0.0125,
      "step": 8656
    },
    {
      "epoch": 2.321533923303835,
      "grad_norm": 0.5090462783103876,
      "learning_rate": 1.4788369884468879e-06,
      "loss": 0.0202,
      "step": 8657
    },
    {
      "epoch": 2.321802091713596,
      "grad_norm": 0.20328981688919176,
      "learning_rate": 1.4777294749384812e-06,
      "loss": 0.0105,
      "step": 8658
    },
    {
      "epoch": 2.3220702601233576,
      "grad_norm": 0.2681476936830618,
      "learning_rate": 1.4766223043842055e-06,
      "loss": 0.0145,
      "step": 8659
    },
    {
      "epoch": 2.3223384285331186,
      "grad_norm": 0.2405206618204203,
      "learning_rate": 1.4755154768918595e-06,
      "loss": 0.0136,
      "step": 8660
    },
    {
      "epoch": 2.32260659694288,
      "grad_norm": 0.2585878013461934,
      "learning_rate": 1.4744089925692146e-06,
      "loss": 0.0117,
      "step": 8661
    },
    {
      "epoch": 2.3228747653526414,
      "grad_norm": 0.21775794137735757,
      "learning_rate": 1.4733028515240078e-06,
      "loss": 0.0127,
      "step": 8662
    },
    {
      "epoch": 2.3231429337624028,
      "grad_norm": 0.256179844960534,
      "learning_rate": 1.4721970538639368e-06,
      "loss": 0.0144,
      "step": 8663
    },
    {
      "epoch": 2.323411102172164,
      "grad_norm": 0.2557355753713361,
      "learning_rate": 1.4710915996966745e-06,
      "loss": 0.0138,
      "step": 8664
    },
    {
      "epoch": 2.3236792705819256,
      "grad_norm": 0.27140538148902027,
      "learning_rate": 1.4699864891298543e-06,
      "loss": 0.014,
      "step": 8665
    },
    {
      "epoch": 2.323947438991687,
      "grad_norm": 0.2261401722738106,
      "learning_rate": 1.468881722271076e-06,
      "loss": 0.0133,
      "step": 8666
    },
    {
      "epoch": 2.324215607401448,
      "grad_norm": 0.28234224604889957,
      "learning_rate": 1.467777299227911e-06,
      "loss": 0.0192,
      "step": 8667
    },
    {
      "epoch": 2.3244837758112094,
      "grad_norm": 0.24259925820230865,
      "learning_rate": 1.4666732201078903e-06,
      "loss": 0.0172,
      "step": 8668
    },
    {
      "epoch": 2.3247519442209708,
      "grad_norm": 0.22426983024369054,
      "learning_rate": 1.46556948501852e-06,
      "loss": 0.0142,
      "step": 8669
    },
    {
      "epoch": 2.325020112630732,
      "grad_norm": 0.2409633800842328,
      "learning_rate": 1.4644660940672628e-06,
      "loss": 0.0148,
      "step": 8670
    },
    {
      "epoch": 2.3252882810404936,
      "grad_norm": 0.20695553915390003,
      "learning_rate": 1.4633630473615557e-06,
      "loss": 0.0112,
      "step": 8671
    },
    {
      "epoch": 2.3255564494502545,
      "grad_norm": 0.2786438906176772,
      "learning_rate": 1.4622603450088007e-06,
      "loss": 0.0155,
      "step": 8672
    },
    {
      "epoch": 2.325824617860016,
      "grad_norm": 0.22704071141338902,
      "learning_rate": 1.4611579871163616e-06,
      "loss": 0.0102,
      "step": 8673
    },
    {
      "epoch": 2.3260927862697773,
      "grad_norm": 0.27249013841703296,
      "learning_rate": 1.4600559737915754e-06,
      "loss": 0.0147,
      "step": 8674
    },
    {
      "epoch": 2.3263609546795387,
      "grad_norm": 0.53912171181824,
      "learning_rate": 1.4589543051417394e-06,
      "loss": 0.0189,
      "step": 8675
    },
    {
      "epoch": 2.3266291230893,
      "grad_norm": 0.21709447295229015,
      "learning_rate": 1.4578529812741226e-06,
      "loss": 0.0134,
      "step": 8676
    },
    {
      "epoch": 2.3268972914990615,
      "grad_norm": 0.3740660350860136,
      "learning_rate": 1.4567520022959575e-06,
      "loss": 0.0195,
      "step": 8677
    },
    {
      "epoch": 2.327165459908823,
      "grad_norm": 0.23147447070010047,
      "learning_rate": 1.4556513683144403e-06,
      "loss": 0.0139,
      "step": 8678
    },
    {
      "epoch": 2.327433628318584,
      "grad_norm": 0.2503128661520003,
      "learning_rate": 1.4545510794367413e-06,
      "loss": 0.0145,
      "step": 8679
    },
    {
      "epoch": 2.3277017967283453,
      "grad_norm": 0.2210827532651394,
      "learning_rate": 1.4534511357699894e-06,
      "loss": 0.0137,
      "step": 8680
    },
    {
      "epoch": 2.3279699651381067,
      "grad_norm": 0.2590759723543358,
      "learning_rate": 1.4523515374212842e-06,
      "loss": 0.0122,
      "step": 8681
    },
    {
      "epoch": 2.328238133547868,
      "grad_norm": 0.196989088260634,
      "learning_rate": 1.4512522844976927e-06,
      "loss": 0.0078,
      "step": 8682
    },
    {
      "epoch": 2.3285063019576295,
      "grad_norm": 0.22556322722783412,
      "learning_rate": 1.4501533771062426e-06,
      "loss": 0.0126,
      "step": 8683
    },
    {
      "epoch": 2.3287744703673905,
      "grad_norm": 0.2604605854809054,
      "learning_rate": 1.4490548153539358e-06,
      "loss": 0.0129,
      "step": 8684
    },
    {
      "epoch": 2.329042638777152,
      "grad_norm": 0.45427552336200566,
      "learning_rate": 1.4479565993477317e-06,
      "loss": 0.0153,
      "step": 8685
    },
    {
      "epoch": 2.3293108071869133,
      "grad_norm": 0.2349043096559891,
      "learning_rate": 1.4468587291945651e-06,
      "loss": 0.0148,
      "step": 8686
    },
    {
      "epoch": 2.3295789755966747,
      "grad_norm": 0.26467804452173194,
      "learning_rate": 1.4457612050013288e-06,
      "loss": 0.0165,
      "step": 8687
    },
    {
      "epoch": 2.329847144006436,
      "grad_norm": 0.28971480344462647,
      "learning_rate": 1.4446640268748875e-06,
      "loss": 0.0169,
      "step": 8688
    },
    {
      "epoch": 2.3301153124161975,
      "grad_norm": 0.30258972850569443,
      "learning_rate": 1.4435671949220742e-06,
      "loss": 0.0165,
      "step": 8689
    },
    {
      "epoch": 2.330383480825959,
      "grad_norm": 0.1887149669584492,
      "learning_rate": 1.4424707092496777e-06,
      "loss": 0.0101,
      "step": 8690
    },
    {
      "epoch": 2.33065164923572,
      "grad_norm": 0.1861058472565932,
      "learning_rate": 1.4413745699644633e-06,
      "loss": 0.0096,
      "step": 8691
    },
    {
      "epoch": 2.3309198176454813,
      "grad_norm": 0.2149779785766391,
      "learning_rate": 1.4402787771731602e-06,
      "loss": 0.0112,
      "step": 8692
    },
    {
      "epoch": 2.3311879860552427,
      "grad_norm": 0.29153411132394547,
      "learning_rate": 1.4391833309824604e-06,
      "loss": 0.0154,
      "step": 8693
    },
    {
      "epoch": 2.331456154465004,
      "grad_norm": 0.3105750627854313,
      "learning_rate": 1.4380882314990274e-06,
      "loss": 0.0116,
      "step": 8694
    },
    {
      "epoch": 2.3317243228747655,
      "grad_norm": 0.22455057006404402,
      "learning_rate": 1.4369934788294854e-06,
      "loss": 0.0105,
      "step": 8695
    },
    {
      "epoch": 2.3319924912845265,
      "grad_norm": 0.26440743961439617,
      "learning_rate": 1.43589907308043e-06,
      "loss": 0.0195,
      "step": 8696
    },
    {
      "epoch": 2.332260659694288,
      "grad_norm": 0.22502744550643716,
      "learning_rate": 1.4348050143584175e-06,
      "loss": 0.0157,
      "step": 8697
    },
    {
      "epoch": 2.3325288281040493,
      "grad_norm": 0.23474988374050096,
      "learning_rate": 1.4337113027699757e-06,
      "loss": 0.0136,
      "step": 8698
    },
    {
      "epoch": 2.3327969965138107,
      "grad_norm": 0.2563789742052498,
      "learning_rate": 1.4326179384215976e-06,
      "loss": 0.0152,
      "step": 8699
    },
    {
      "epoch": 2.333065164923572,
      "grad_norm": 0.29205139396130114,
      "learning_rate": 1.431524921419738e-06,
      "loss": 0.019,
      "step": 8700
    },
    {
      "epoch": 2.3333333333333335,
      "grad_norm": 0.3274340907847781,
      "learning_rate": 1.4304322518708258e-06,
      "loss": 0.0222,
      "step": 8701
    },
    {
      "epoch": 2.333601501743095,
      "grad_norm": 0.19762588085895882,
      "learning_rate": 1.429339929881245e-06,
      "loss": 0.0145,
      "step": 8702
    },
    {
      "epoch": 2.333869670152856,
      "grad_norm": 0.22875533841975274,
      "learning_rate": 1.4282479555573559e-06,
      "loss": 0.011,
      "step": 8703
    },
    {
      "epoch": 2.3341378385626173,
      "grad_norm": 0.24973670746102064,
      "learning_rate": 1.4271563290054813e-06,
      "loss": 0.0185,
      "step": 8704
    },
    {
      "epoch": 2.3344060069723787,
      "grad_norm": 0.24010374259600617,
      "learning_rate": 1.4260650503319078e-06,
      "loss": 0.0149,
      "step": 8705
    },
    {
      "epoch": 2.33467417538214,
      "grad_norm": 0.2249552374506792,
      "learning_rate": 1.4249741196428935e-06,
      "loss": 0.016,
      "step": 8706
    },
    {
      "epoch": 2.3349423437919015,
      "grad_norm": 0.2938645353662435,
      "learning_rate": 1.4238835370446564e-06,
      "loss": 0.024,
      "step": 8707
    },
    {
      "epoch": 2.3352105122016624,
      "grad_norm": 0.21096546303225347,
      "learning_rate": 1.4227933026433844e-06,
      "loss": 0.0108,
      "step": 8708
    },
    {
      "epoch": 2.335478680611424,
      "grad_norm": 0.24438435878892928,
      "learning_rate": 1.4217034165452332e-06,
      "loss": 0.0121,
      "step": 8709
    },
    {
      "epoch": 2.3357468490211852,
      "grad_norm": 0.21412421639257484,
      "learning_rate": 1.4206138788563185e-06,
      "loss": 0.0116,
      "step": 8710
    },
    {
      "epoch": 2.3360150174309466,
      "grad_norm": 0.24530523152328118,
      "learning_rate": 1.4195246896827287e-06,
      "loss": 0.0212,
      "step": 8711
    },
    {
      "epoch": 2.336283185840708,
      "grad_norm": 0.18449482618449511,
      "learning_rate": 1.4184358491305123e-06,
      "loss": 0.0098,
      "step": 8712
    },
    {
      "epoch": 2.3365513542504694,
      "grad_norm": 0.24340499543134247,
      "learning_rate": 1.4173473573056894e-06,
      "loss": 0.0141,
      "step": 8713
    },
    {
      "epoch": 2.336819522660231,
      "grad_norm": 0.15628529029745822,
      "learning_rate": 1.4162592143142435e-06,
      "loss": 0.0069,
      "step": 8714
    },
    {
      "epoch": 2.337087691069992,
      "grad_norm": 0.22775841085893547,
      "learning_rate": 1.4151714202621214e-06,
      "loss": 0.0131,
      "step": 8715
    },
    {
      "epoch": 2.337355859479753,
      "grad_norm": 0.20057119975694843,
      "learning_rate": 1.414083975255242e-06,
      "loss": 0.0113,
      "step": 8716
    },
    {
      "epoch": 2.3376240278895146,
      "grad_norm": 0.20190872508930574,
      "learning_rate": 1.412996879399484e-06,
      "loss": 0.0126,
      "step": 8717
    },
    {
      "epoch": 2.337892196299276,
      "grad_norm": 0.2661059827505163,
      "learning_rate": 1.4119101328006984e-06,
      "loss": 0.0164,
      "step": 8718
    },
    {
      "epoch": 2.3381603647090374,
      "grad_norm": 0.26761286583020666,
      "learning_rate": 1.4108237355646954e-06,
      "loss": 0.0129,
      "step": 8719
    },
    {
      "epoch": 2.3384285331187984,
      "grad_norm": 0.26475320826499826,
      "learning_rate": 1.4097376877972556e-06,
      "loss": 0.0134,
      "step": 8720
    },
    {
      "epoch": 2.33869670152856,
      "grad_norm": 0.24252686662367595,
      "learning_rate": 1.4086519896041274e-06,
      "loss": 0.0159,
      "step": 8721
    },
    {
      "epoch": 2.338964869938321,
      "grad_norm": 0.23551756950743247,
      "learning_rate": 1.4075666410910178e-06,
      "loss": 0.0103,
      "step": 8722
    },
    {
      "epoch": 2.3392330383480826,
      "grad_norm": 0.22981124267325467,
      "learning_rate": 1.4064816423636091e-06,
      "loss": 0.0135,
      "step": 8723
    },
    {
      "epoch": 2.339501206757844,
      "grad_norm": 0.24957913105906604,
      "learning_rate": 1.4053969935275403e-06,
      "loss": 0.0149,
      "step": 8724
    },
    {
      "epoch": 2.3397693751676054,
      "grad_norm": 0.3834226660751616,
      "learning_rate": 1.4043126946884238e-06,
      "loss": 0.0142,
      "step": 8725
    },
    {
      "epoch": 2.340037543577367,
      "grad_norm": 0.2233007333266325,
      "learning_rate": 1.4032287459518346e-06,
      "loss": 0.0133,
      "step": 8726
    },
    {
      "epoch": 2.340305711987128,
      "grad_norm": 0.21591740359232955,
      "learning_rate": 1.4021451474233111e-06,
      "loss": 0.0109,
      "step": 8727
    },
    {
      "epoch": 2.340573880396889,
      "grad_norm": 0.22120398769901672,
      "learning_rate": 1.4010618992083646e-06,
      "loss": 0.0149,
      "step": 8728
    },
    {
      "epoch": 2.3408420488066506,
      "grad_norm": 0.22924108163063697,
      "learning_rate": 1.3999790014124636e-06,
      "loss": 0.0089,
      "step": 8729
    },
    {
      "epoch": 2.341110217216412,
      "grad_norm": 0.23076648069175332,
      "learning_rate": 1.39889645414105e-06,
      "loss": 0.0125,
      "step": 8730
    },
    {
      "epoch": 2.3413783856261734,
      "grad_norm": 0.20233388140574485,
      "learning_rate": 1.3978142574995296e-06,
      "loss": 0.0127,
      "step": 8731
    },
    {
      "epoch": 2.3416465540359344,
      "grad_norm": 0.21478124204734136,
      "learning_rate": 1.3967324115932696e-06,
      "loss": 0.0096,
      "step": 8732
    },
    {
      "epoch": 2.3419147224456958,
      "grad_norm": 0.37381161905392574,
      "learning_rate": 1.3956509165276095e-06,
      "loss": 0.0157,
      "step": 8733
    },
    {
      "epoch": 2.342182890855457,
      "grad_norm": 0.21164052749809473,
      "learning_rate": 1.394569772407849e-06,
      "loss": 0.0084,
      "step": 8734
    },
    {
      "epoch": 2.3424510592652186,
      "grad_norm": 0.2580662464620983,
      "learning_rate": 1.3934889793392575e-06,
      "loss": 0.0139,
      "step": 8735
    },
    {
      "epoch": 2.34271922767498,
      "grad_norm": 0.4764464248250427,
      "learning_rate": 1.3924085374270701e-06,
      "loss": 0.0232,
      "step": 8736
    },
    {
      "epoch": 2.3429873960847414,
      "grad_norm": 0.21738121815564201,
      "learning_rate": 1.3913284467764859e-06,
      "loss": 0.0134,
      "step": 8737
    },
    {
      "epoch": 2.343255564494503,
      "grad_norm": 0.24962066159399496,
      "learning_rate": 1.3902487074926702e-06,
      "loss": 0.0135,
      "step": 8738
    },
    {
      "epoch": 2.3435237329042637,
      "grad_norm": 0.24091110165201918,
      "learning_rate": 1.389169319680752e-06,
      "loss": 0.0157,
      "step": 8739
    },
    {
      "epoch": 2.343791901314025,
      "grad_norm": 0.3273392600999916,
      "learning_rate": 1.388090283445831e-06,
      "loss": 0.016,
      "step": 8740
    },
    {
      "epoch": 2.3440600697237866,
      "grad_norm": 0.20532480895440475,
      "learning_rate": 1.387011598892971e-06,
      "loss": 0.0121,
      "step": 8741
    },
    {
      "epoch": 2.344328238133548,
      "grad_norm": 0.25469043473168734,
      "learning_rate": 1.385933266127198e-06,
      "loss": 0.0139,
      "step": 8742
    },
    {
      "epoch": 2.3445964065433094,
      "grad_norm": 0.24838213309250026,
      "learning_rate": 1.384855285253509e-06,
      "loss": 0.0201,
      "step": 8743
    },
    {
      "epoch": 2.3448645749530703,
      "grad_norm": 0.3622427493343319,
      "learning_rate": 1.3837776563768613e-06,
      "loss": 0.0137,
      "step": 8744
    },
    {
      "epoch": 2.3451327433628317,
      "grad_norm": 0.23862192449280178,
      "learning_rate": 1.3827003796021838e-06,
      "loss": 0.0075,
      "step": 8745
    },
    {
      "epoch": 2.345400911772593,
      "grad_norm": 0.22864217964395311,
      "learning_rate": 1.3816234550343655e-06,
      "loss": 0.011,
      "step": 8746
    },
    {
      "epoch": 2.3456690801823545,
      "grad_norm": 0.2614900229516161,
      "learning_rate": 1.3805468827782642e-06,
      "loss": 0.0149,
      "step": 8747
    },
    {
      "epoch": 2.345937248592116,
      "grad_norm": 0.28310098599203615,
      "learning_rate": 1.3794706629387056e-06,
      "loss": 0.0196,
      "step": 8748
    },
    {
      "epoch": 2.3462054170018773,
      "grad_norm": 0.2840282487576676,
      "learning_rate": 1.3783947956204756e-06,
      "loss": 0.0139,
      "step": 8749
    },
    {
      "epoch": 2.3464735854116388,
      "grad_norm": 0.29103374726116016,
      "learning_rate": 1.3773192809283282e-06,
      "loss": 0.0143,
      "step": 8750
    },
    {
      "epoch": 2.3467417538213997,
      "grad_norm": 0.25237074371255347,
      "learning_rate": 1.3762441189669855e-06,
      "loss": 0.014,
      "step": 8751
    },
    {
      "epoch": 2.347009922231161,
      "grad_norm": 0.2563584933085516,
      "learning_rate": 1.3751693098411312e-06,
      "loss": 0.0134,
      "step": 8752
    },
    {
      "epoch": 2.3472780906409225,
      "grad_norm": 0.2812388417712608,
      "learning_rate": 1.3740948536554182e-06,
      "loss": 0.0164,
      "step": 8753
    },
    {
      "epoch": 2.347546259050684,
      "grad_norm": 0.2543324919796652,
      "learning_rate": 1.3730207505144617e-06,
      "loss": 0.0133,
      "step": 8754
    },
    {
      "epoch": 2.3478144274604453,
      "grad_norm": 0.24450088550378657,
      "learning_rate": 1.3719470005228463e-06,
      "loss": 0.0152,
      "step": 8755
    },
    {
      "epoch": 2.3480825958702063,
      "grad_norm": 0.24091163965857523,
      "learning_rate": 1.3708736037851177e-06,
      "loss": 0.0131,
      "step": 8756
    },
    {
      "epoch": 2.3483507642799677,
      "grad_norm": 0.26512613416436936,
      "learning_rate": 1.3698005604057918e-06,
      "loss": 0.0135,
      "step": 8757
    },
    {
      "epoch": 2.348618932689729,
      "grad_norm": 0.2392380160697273,
      "learning_rate": 1.368727870489348e-06,
      "loss": 0.0106,
      "step": 8758
    },
    {
      "epoch": 2.3488871010994905,
      "grad_norm": 0.3020065200032434,
      "learning_rate": 1.3676555341402298e-06,
      "loss": 0.0147,
      "step": 8759
    },
    {
      "epoch": 2.349155269509252,
      "grad_norm": 0.19561372886869396,
      "learning_rate": 1.36658355146285e-06,
      "loss": 0.0096,
      "step": 8760
    },
    {
      "epoch": 2.3494234379190133,
      "grad_norm": 0.2249178552632823,
      "learning_rate": 1.3655119225615832e-06,
      "loss": 0.0122,
      "step": 8761
    },
    {
      "epoch": 2.3496916063287747,
      "grad_norm": 0.23630131147452707,
      "learning_rate": 1.3644406475407689e-06,
      "loss": 0.0115,
      "step": 8762
    },
    {
      "epoch": 2.3499597747385357,
      "grad_norm": 0.17393779661894931,
      "learning_rate": 1.363369726504719e-06,
      "loss": 0.0091,
      "step": 8763
    },
    {
      "epoch": 2.350227943148297,
      "grad_norm": 0.2834533902175608,
      "learning_rate": 1.3622991595577018e-06,
      "loss": 0.011,
      "step": 8764
    },
    {
      "epoch": 2.3504961115580585,
      "grad_norm": 0.23481106304763782,
      "learning_rate": 1.3612289468039592e-06,
      "loss": 0.0134,
      "step": 8765
    },
    {
      "epoch": 2.35076427996782,
      "grad_norm": 0.31338252159442564,
      "learning_rate": 1.3601590883476917e-06,
      "loss": 0.0226,
      "step": 8766
    },
    {
      "epoch": 2.3510324483775813,
      "grad_norm": 0.32927384652862585,
      "learning_rate": 1.3590895842930702e-06,
      "loss": 0.0138,
      "step": 8767
    },
    {
      "epoch": 2.3513006167873423,
      "grad_norm": 0.2919477831314323,
      "learning_rate": 1.3580204347442304e-06,
      "loss": 0.018,
      "step": 8768
    },
    {
      "epoch": 2.3515687851971037,
      "grad_norm": 0.24778176043977437,
      "learning_rate": 1.3569516398052707e-06,
      "loss": 0.0133,
      "step": 8769
    },
    {
      "epoch": 2.351836953606865,
      "grad_norm": 0.2753183559944742,
      "learning_rate": 1.3558831995802597e-06,
      "loss": 0.0156,
      "step": 8770
    },
    {
      "epoch": 2.3521051220166265,
      "grad_norm": 0.22321299850041598,
      "learning_rate": 1.354815114173224e-06,
      "loss": 0.0122,
      "step": 8771
    },
    {
      "epoch": 2.352373290426388,
      "grad_norm": 0.2744825607349537,
      "learning_rate": 1.3537473836881655e-06,
      "loss": 0.0182,
      "step": 8772
    },
    {
      "epoch": 2.3526414588361493,
      "grad_norm": 0.2550076113894432,
      "learning_rate": 1.3526800082290425e-06,
      "loss": 0.0167,
      "step": 8773
    },
    {
      "epoch": 2.3529096272459102,
      "grad_norm": 0.27622058249258497,
      "learning_rate": 1.351612987899783e-06,
      "loss": 0.0143,
      "step": 8774
    },
    {
      "epoch": 2.3531777956556716,
      "grad_norm": 0.28347254876238986,
      "learning_rate": 1.3505463228042814e-06,
      "loss": 0.0187,
      "step": 8775
    },
    {
      "epoch": 2.353445964065433,
      "grad_norm": 0.2851610549181346,
      "learning_rate": 1.3494800130463943e-06,
      "loss": 0.0144,
      "step": 8776
    },
    {
      "epoch": 2.3537141324751945,
      "grad_norm": 0.306055209688671,
      "learning_rate": 1.348414058729946e-06,
      "loss": 0.0116,
      "step": 8777
    },
    {
      "epoch": 2.353982300884956,
      "grad_norm": 0.26858599344716333,
      "learning_rate": 1.3473484599587277e-06,
      "loss": 0.0161,
      "step": 8778
    },
    {
      "epoch": 2.3542504692947173,
      "grad_norm": 0.7184244746239631,
      "learning_rate": 1.3462832168364908e-06,
      "loss": 0.0169,
      "step": 8779
    },
    {
      "epoch": 2.3545186377044782,
      "grad_norm": 0.26973482776098645,
      "learning_rate": 1.3452183294669586e-06,
      "loss": 0.0156,
      "step": 8780
    },
    {
      "epoch": 2.3547868061142396,
      "grad_norm": 0.2332269645852738,
      "learning_rate": 1.3441537979538122e-06,
      "loss": 0.0148,
      "step": 8781
    },
    {
      "epoch": 2.355054974524001,
      "grad_norm": 0.2457826690013042,
      "learning_rate": 1.3430896224007062e-06,
      "loss": 0.0127,
      "step": 8782
    },
    {
      "epoch": 2.3553231429337624,
      "grad_norm": 0.2023712817341855,
      "learning_rate": 1.342025802911253e-06,
      "loss": 0.0128,
      "step": 8783
    },
    {
      "epoch": 2.355591311343524,
      "grad_norm": 0.29414674598457763,
      "learning_rate": 1.3409623395890358e-06,
      "loss": 0.0196,
      "step": 8784
    },
    {
      "epoch": 2.3558594797532852,
      "grad_norm": 0.24553793483254582,
      "learning_rate": 1.3398992325376042e-06,
      "loss": 0.0124,
      "step": 8785
    },
    {
      "epoch": 2.356127648163046,
      "grad_norm": 0.2736543228878304,
      "learning_rate": 1.3388364818604638e-06,
      "loss": 0.0205,
      "step": 8786
    },
    {
      "epoch": 2.3563958165728076,
      "grad_norm": 0.24233062247216783,
      "learning_rate": 1.337774087661095e-06,
      "loss": 0.0158,
      "step": 8787
    },
    {
      "epoch": 2.356663984982569,
      "grad_norm": 0.22995513994594335,
      "learning_rate": 1.336712050042942e-06,
      "loss": 0.0138,
      "step": 8788
    },
    {
      "epoch": 2.3569321533923304,
      "grad_norm": 0.3001133876697425,
      "learning_rate": 1.3356503691094092e-06,
      "loss": 0.0191,
      "step": 8789
    },
    {
      "epoch": 2.357200321802092,
      "grad_norm": 0.23956923387351903,
      "learning_rate": 1.3345890449638726e-06,
      "loss": 0.0142,
      "step": 8790
    },
    {
      "epoch": 2.3574684902118532,
      "grad_norm": 0.29043022241683497,
      "learning_rate": 1.3335280777096676e-06,
      "loss": 0.0191,
      "step": 8791
    },
    {
      "epoch": 2.357736658621614,
      "grad_norm": 0.25988682334278884,
      "learning_rate": 1.332467467450101e-06,
      "loss": 0.0129,
      "step": 8792
    },
    {
      "epoch": 2.3580048270313756,
      "grad_norm": 0.22658708464542276,
      "learning_rate": 1.3314072142884388e-06,
      "loss": 0.0122,
      "step": 8793
    },
    {
      "epoch": 2.358272995441137,
      "grad_norm": 0.2797839439996834,
      "learning_rate": 1.3303473183279158e-06,
      "loss": 0.0163,
      "step": 8794
    },
    {
      "epoch": 2.3585411638508984,
      "grad_norm": 0.24922688482613706,
      "learning_rate": 1.3292877796717335e-06,
      "loss": 0.0124,
      "step": 8795
    },
    {
      "epoch": 2.35880933226066,
      "grad_norm": 0.2962999612129481,
      "learning_rate": 1.3282285984230532e-06,
      "loss": 0.0201,
      "step": 8796
    },
    {
      "epoch": 2.359077500670421,
      "grad_norm": 0.3000251792134309,
      "learning_rate": 1.3271697746850066e-06,
      "loss": 0.0159,
      "step": 8797
    },
    {
      "epoch": 2.359345669080182,
      "grad_norm": 0.19414631907929364,
      "learning_rate": 1.3261113085606886e-06,
      "loss": 0.0132,
      "step": 8798
    },
    {
      "epoch": 2.3596138374899436,
      "grad_norm": 0.2034764083800131,
      "learning_rate": 1.3250532001531568e-06,
      "loss": 0.0107,
      "step": 8799
    },
    {
      "epoch": 2.359882005899705,
      "grad_norm": 0.290166773619518,
      "learning_rate": 1.3239954495654394e-06,
      "loss": 0.0141,
      "step": 8800
    },
    {
      "epoch": 2.3601501743094664,
      "grad_norm": 0.2962176899656806,
      "learning_rate": 1.322938056900524e-06,
      "loss": 0.0225,
      "step": 8801
    },
    {
      "epoch": 2.360418342719228,
      "grad_norm": 0.271721296400545,
      "learning_rate": 1.3218810222613694e-06,
      "loss": 0.0131,
      "step": 8802
    },
    {
      "epoch": 2.360686511128989,
      "grad_norm": 0.2017296912438486,
      "learning_rate": 1.3208243457508928e-06,
      "loss": 0.0104,
      "step": 8803
    },
    {
      "epoch": 2.36095467953875,
      "grad_norm": 0.33463941482740056,
      "learning_rate": 1.3197680274719816e-06,
      "loss": 0.024,
      "step": 8804
    },
    {
      "epoch": 2.3612228479485116,
      "grad_norm": 0.2824241849845443,
      "learning_rate": 1.3187120675274883e-06,
      "loss": 0.0142,
      "step": 8805
    },
    {
      "epoch": 2.361491016358273,
      "grad_norm": 0.26892092478512397,
      "learning_rate": 1.3176564660202256e-06,
      "loss": 0.0124,
      "step": 8806
    },
    {
      "epoch": 2.3617591847680344,
      "grad_norm": 0.34652967835942994,
      "learning_rate": 1.3166012230529784e-06,
      "loss": 0.0224,
      "step": 8807
    },
    {
      "epoch": 2.3620273531777958,
      "grad_norm": 0.2029031056031269,
      "learning_rate": 1.3155463387284889e-06,
      "loss": 0.0125,
      "step": 8808
    },
    {
      "epoch": 2.362295521587557,
      "grad_norm": 0.2623270919671144,
      "learning_rate": 1.3144918131494722e-06,
      "loss": 0.0133,
      "step": 8809
    },
    {
      "epoch": 2.362563689997318,
      "grad_norm": 0.22182874828442717,
      "learning_rate": 1.3134376464186028e-06,
      "loss": 0.0096,
      "step": 8810
    },
    {
      "epoch": 2.3628318584070795,
      "grad_norm": 0.2190420058664251,
      "learning_rate": 1.31238383863852e-06,
      "loss": 0.0173,
      "step": 8811
    },
    {
      "epoch": 2.363100026816841,
      "grad_norm": 0.29008216219437244,
      "learning_rate": 1.3113303899118346e-06,
      "loss": 0.0155,
      "step": 8812
    },
    {
      "epoch": 2.3633681952266024,
      "grad_norm": 0.22984360179359528,
      "learning_rate": 1.310277300341114e-06,
      "loss": 0.0135,
      "step": 8813
    },
    {
      "epoch": 2.3636363636363638,
      "grad_norm": 0.20017662413180887,
      "learning_rate": 1.3092245700288985e-06,
      "loss": 0.0131,
      "step": 8814
    },
    {
      "epoch": 2.363904532046125,
      "grad_norm": 0.265157379183699,
      "learning_rate": 1.3081721990776863e-06,
      "loss": 0.0122,
      "step": 8815
    },
    {
      "epoch": 2.364172700455886,
      "grad_norm": 0.3100513096766328,
      "learning_rate": 1.3071201875899453e-06,
      "loss": 0.0157,
      "step": 8816
    },
    {
      "epoch": 2.3644408688656475,
      "grad_norm": 0.2659911663172562,
      "learning_rate": 1.3060685356681084e-06,
      "loss": 0.0142,
      "step": 8817
    },
    {
      "epoch": 2.364709037275409,
      "grad_norm": 0.2160980153538925,
      "learning_rate": 1.3050172434145697e-06,
      "loss": 0.0134,
      "step": 8818
    },
    {
      "epoch": 2.3649772056851703,
      "grad_norm": 0.255943968011093,
      "learning_rate": 1.3039663109316936e-06,
      "loss": 0.0128,
      "step": 8819
    },
    {
      "epoch": 2.3652453740949317,
      "grad_norm": 0.33857607067218165,
      "learning_rate": 1.3029157383218039e-06,
      "loss": 0.0156,
      "step": 8820
    },
    {
      "epoch": 2.3655135425046927,
      "grad_norm": 0.21608554484041517,
      "learning_rate": 1.3018655256871925e-06,
      "loss": 0.012,
      "step": 8821
    },
    {
      "epoch": 2.365781710914454,
      "grad_norm": 0.3184395345812074,
      "learning_rate": 1.3008156731301202e-06,
      "loss": 0.0157,
      "step": 8822
    },
    {
      "epoch": 2.3660498793242155,
      "grad_norm": 0.25402450737454285,
      "learning_rate": 1.2997661807528011e-06,
      "loss": 0.0101,
      "step": 8823
    },
    {
      "epoch": 2.366318047733977,
      "grad_norm": 0.19659072128945074,
      "learning_rate": 1.298717048657427e-06,
      "loss": 0.0104,
      "step": 8824
    },
    {
      "epoch": 2.3665862161437383,
      "grad_norm": 0.2158454993872614,
      "learning_rate": 1.2976682769461451e-06,
      "loss": 0.0105,
      "step": 8825
    },
    {
      "epoch": 2.3668543845534997,
      "grad_norm": 0.2389747279511932,
      "learning_rate": 1.2966198657210738e-06,
      "loss": 0.0137,
      "step": 8826
    },
    {
      "epoch": 2.367122552963261,
      "grad_norm": 0.2700955864688488,
      "learning_rate": 1.2955718150842955e-06,
      "loss": 0.0136,
      "step": 8827
    },
    {
      "epoch": 2.367390721373022,
      "grad_norm": 0.2336387971828409,
      "learning_rate": 1.2945241251378531e-06,
      "loss": 0.014,
      "step": 8828
    },
    {
      "epoch": 2.3676588897827835,
      "grad_norm": 0.2981731815514084,
      "learning_rate": 1.2934767959837601e-06,
      "loss": 0.019,
      "step": 8829
    },
    {
      "epoch": 2.367927058192545,
      "grad_norm": 0.23071617343697035,
      "learning_rate": 1.2924298277239893e-06,
      "loss": 0.0129,
      "step": 8830
    },
    {
      "epoch": 2.3681952266023063,
      "grad_norm": 0.30180035200700134,
      "learning_rate": 1.2913832204604843e-06,
      "loss": 0.0151,
      "step": 8831
    },
    {
      "epoch": 2.3684633950120677,
      "grad_norm": 0.19731641864175456,
      "learning_rate": 1.2903369742951476e-06,
      "loss": 0.0113,
      "step": 8832
    },
    {
      "epoch": 2.3687315634218287,
      "grad_norm": 0.23816816914990033,
      "learning_rate": 1.2892910893298506e-06,
      "loss": 0.0136,
      "step": 8833
    },
    {
      "epoch": 2.36899973183159,
      "grad_norm": 0.2713304562654815,
      "learning_rate": 1.2882455656664316e-06,
      "loss": 0.0131,
      "step": 8834
    },
    {
      "epoch": 2.3692679002413515,
      "grad_norm": 0.2696318193781521,
      "learning_rate": 1.2872004034066843e-06,
      "loss": 0.0177,
      "step": 8835
    },
    {
      "epoch": 2.369536068651113,
      "grad_norm": 0.5978011517919262,
      "learning_rate": 1.2861556026523765e-06,
      "loss": 0.0295,
      "step": 8836
    },
    {
      "epoch": 2.3698042370608743,
      "grad_norm": 0.24738344049179034,
      "learning_rate": 1.2851111635052388e-06,
      "loss": 0.0156,
      "step": 8837
    },
    {
      "epoch": 2.3700724054706357,
      "grad_norm": 0.21850065252202885,
      "learning_rate": 1.284067086066963e-06,
      "loss": 0.0099,
      "step": 8838
    },
    {
      "epoch": 2.370340573880397,
      "grad_norm": 0.23101054793043033,
      "learning_rate": 1.2830233704392108e-06,
      "loss": 0.0091,
      "step": 8839
    },
    {
      "epoch": 2.370608742290158,
      "grad_norm": 0.2907302675671745,
      "learning_rate": 1.2819800167236034e-06,
      "loss": 0.0156,
      "step": 8840
    },
    {
      "epoch": 2.3708769106999195,
      "grad_norm": 0.21538506819431902,
      "learning_rate": 1.2809370250217324e-06,
      "loss": 0.012,
      "step": 8841
    },
    {
      "epoch": 2.371145079109681,
      "grad_norm": 0.20854084638980305,
      "learning_rate": 1.2798943954351472e-06,
      "loss": 0.0121,
      "step": 8842
    },
    {
      "epoch": 2.3714132475194423,
      "grad_norm": 0.24099820408389627,
      "learning_rate": 1.2788521280653682e-06,
      "loss": 0.0095,
      "step": 8843
    },
    {
      "epoch": 2.3716814159292037,
      "grad_norm": 0.1694392024663969,
      "learning_rate": 1.2778102230138795e-06,
      "loss": 0.0085,
      "step": 8844
    },
    {
      "epoch": 2.3719495843389646,
      "grad_norm": 0.2509983520713743,
      "learning_rate": 1.276768680382126e-06,
      "loss": 0.0136,
      "step": 8845
    },
    {
      "epoch": 2.372217752748726,
      "grad_norm": 0.2690816501472869,
      "learning_rate": 1.2757275002715219e-06,
      "loss": 0.0154,
      "step": 8846
    },
    {
      "epoch": 2.3724859211584874,
      "grad_norm": 0.2233873316102051,
      "learning_rate": 1.2746866827834443e-06,
      "loss": 0.01,
      "step": 8847
    },
    {
      "epoch": 2.372754089568249,
      "grad_norm": 0.2623754794963019,
      "learning_rate": 1.2736462280192318e-06,
      "loss": 0.0148,
      "step": 8848
    },
    {
      "epoch": 2.3730222579780103,
      "grad_norm": 0.2960986759332379,
      "learning_rate": 1.2726061360801945e-06,
      "loss": 0.0198,
      "step": 8849
    },
    {
      "epoch": 2.3732904263877717,
      "grad_norm": 0.24959970144102056,
      "learning_rate": 1.2715664070675998e-06,
      "loss": 0.0162,
      "step": 8850
    },
    {
      "epoch": 2.373558594797533,
      "grad_norm": 0.22214111939214695,
      "learning_rate": 1.270527041082687e-06,
      "loss": 0.014,
      "step": 8851
    },
    {
      "epoch": 2.373826763207294,
      "grad_norm": 0.23426650578312616,
      "learning_rate": 1.2694880382266534e-06,
      "loss": 0.0117,
      "step": 8852
    },
    {
      "epoch": 2.3740949316170554,
      "grad_norm": 0.33907043249678986,
      "learning_rate": 1.2684493986006646e-06,
      "loss": 0.019,
      "step": 8853
    },
    {
      "epoch": 2.374363100026817,
      "grad_norm": 0.24163505567158972,
      "learning_rate": 1.2674111223058526e-06,
      "loss": 0.011,
      "step": 8854
    },
    {
      "epoch": 2.3746312684365782,
      "grad_norm": 0.25438166133148776,
      "learning_rate": 1.2663732094433084e-06,
      "loss": 0.0143,
      "step": 8855
    },
    {
      "epoch": 2.3748994368463396,
      "grad_norm": 0.26791961151219207,
      "learning_rate": 1.2653356601140936e-06,
      "loss": 0.0234,
      "step": 8856
    },
    {
      "epoch": 2.3751676052561006,
      "grad_norm": 0.290136182835645,
      "learning_rate": 1.2642984744192289e-06,
      "loss": 0.0139,
      "step": 8857
    },
    {
      "epoch": 2.375435773665862,
      "grad_norm": 0.2653624783908322,
      "learning_rate": 1.2632616524597052e-06,
      "loss": 0.0128,
      "step": 8858
    },
    {
      "epoch": 2.3757039420756234,
      "grad_norm": 0.22840591599123197,
      "learning_rate": 1.2622251943364733e-06,
      "loss": 0.0117,
      "step": 8859
    },
    {
      "epoch": 2.375972110485385,
      "grad_norm": 0.2407417968247365,
      "learning_rate": 1.2611891001504489e-06,
      "loss": 0.0118,
      "step": 8860
    },
    {
      "epoch": 2.376240278895146,
      "grad_norm": 0.26844069939663145,
      "learning_rate": 1.2601533700025177e-06,
      "loss": 0.0159,
      "step": 8861
    },
    {
      "epoch": 2.3765084473049076,
      "grad_norm": 0.2932895122780787,
      "learning_rate": 1.259118003993522e-06,
      "loss": 0.0161,
      "step": 8862
    },
    {
      "epoch": 2.376776615714669,
      "grad_norm": 0.2978579334284393,
      "learning_rate": 1.258083002224274e-06,
      "loss": 0.0136,
      "step": 8863
    },
    {
      "epoch": 2.37704478412443,
      "grad_norm": 0.24840312389348312,
      "learning_rate": 1.2570483647955512e-06,
      "loss": 0.0103,
      "step": 8864
    },
    {
      "epoch": 2.3773129525341914,
      "grad_norm": 0.23327446557141393,
      "learning_rate": 1.2560140918080905e-06,
      "loss": 0.0149,
      "step": 8865
    },
    {
      "epoch": 2.377581120943953,
      "grad_norm": 0.2182026620305688,
      "learning_rate": 1.2549801833625986e-06,
      "loss": 0.0092,
      "step": 8866
    },
    {
      "epoch": 2.377849289353714,
      "grad_norm": 0.2452548195302081,
      "learning_rate": 1.253946639559742e-06,
      "loss": 0.0141,
      "step": 8867
    },
    {
      "epoch": 2.3781174577634756,
      "grad_norm": 0.25182169487687767,
      "learning_rate": 1.2529134605001575e-06,
      "loss": 0.0215,
      "step": 8868
    },
    {
      "epoch": 2.3783856261732366,
      "grad_norm": 0.23953267623160315,
      "learning_rate": 1.251880646284439e-06,
      "loss": 0.0149,
      "step": 8869
    },
    {
      "epoch": 2.378653794582998,
      "grad_norm": 0.2866394907189888,
      "learning_rate": 1.2508481970131525e-06,
      "loss": 0.0169,
      "step": 8870
    },
    {
      "epoch": 2.3789219629927594,
      "grad_norm": 0.30830857756430546,
      "learning_rate": 1.2498161127868236e-06,
      "loss": 0.0181,
      "step": 8871
    },
    {
      "epoch": 2.379190131402521,
      "grad_norm": 0.2159241675216704,
      "learning_rate": 1.2487843937059419e-06,
      "loss": 0.011,
      "step": 8872
    },
    {
      "epoch": 2.379458299812282,
      "grad_norm": 0.33188015865712117,
      "learning_rate": 1.2477530398709637e-06,
      "loss": 0.0189,
      "step": 8873
    },
    {
      "epoch": 2.3797264682220436,
      "grad_norm": 0.2307720202914826,
      "learning_rate": 1.246722051382312e-06,
      "loss": 0.0147,
      "step": 8874
    },
    {
      "epoch": 2.379994636631805,
      "grad_norm": 0.25960144181444483,
      "learning_rate": 1.2456914283403681e-06,
      "loss": 0.0114,
      "step": 8875
    },
    {
      "epoch": 2.380262805041566,
      "grad_norm": 0.2567508372623785,
      "learning_rate": 1.2446611708454836e-06,
      "loss": 0.0128,
      "step": 8876
    },
    {
      "epoch": 2.3805309734513274,
      "grad_norm": 0.27431585564137334,
      "learning_rate": 1.2436312789979687e-06,
      "loss": 0.0174,
      "step": 8877
    },
    {
      "epoch": 2.3807991418610888,
      "grad_norm": 0.29117160031955036,
      "learning_rate": 1.2426017528981054e-06,
      "loss": 0.0158,
      "step": 8878
    },
    {
      "epoch": 2.38106731027085,
      "grad_norm": 0.22451579205646643,
      "learning_rate": 1.241572592646132e-06,
      "loss": 0.0122,
      "step": 8879
    },
    {
      "epoch": 2.3813354786806116,
      "grad_norm": 0.3350332403248255,
      "learning_rate": 1.2405437983422569e-06,
      "loss": 0.0152,
      "step": 8880
    },
    {
      "epoch": 2.3816036470903725,
      "grad_norm": 0.2547375728447541,
      "learning_rate": 1.239515370086652e-06,
      "loss": 0.012,
      "step": 8881
    },
    {
      "epoch": 2.381871815500134,
      "grad_norm": 0.21113936705633257,
      "learning_rate": 1.2384873079794518e-06,
      "loss": 0.0105,
      "step": 8882
    },
    {
      "epoch": 2.3821399839098953,
      "grad_norm": 0.23284560499567805,
      "learning_rate": 1.237459612120755e-06,
      "loss": 0.0141,
      "step": 8883
    },
    {
      "epoch": 2.3824081523196567,
      "grad_norm": 0.22330282190267034,
      "learning_rate": 1.2364322826106273e-06,
      "loss": 0.0127,
      "step": 8884
    },
    {
      "epoch": 2.382676320729418,
      "grad_norm": 0.22630918655180907,
      "learning_rate": 1.2354053195490944e-06,
      "loss": 0.0105,
      "step": 8885
    },
    {
      "epoch": 2.3829444891391796,
      "grad_norm": 0.21455447557134635,
      "learning_rate": 1.2343787230361526e-06,
      "loss": 0.0099,
      "step": 8886
    },
    {
      "epoch": 2.383212657548941,
      "grad_norm": 0.35751760158608464,
      "learning_rate": 1.2333524931717555e-06,
      "loss": 0.0173,
      "step": 8887
    },
    {
      "epoch": 2.383480825958702,
      "grad_norm": 0.2994492926919677,
      "learning_rate": 1.2323266300558274e-06,
      "loss": 0.0104,
      "step": 8888
    },
    {
      "epoch": 2.3837489943684633,
      "grad_norm": 0.26217533467522347,
      "learning_rate": 1.2313011337882507e-06,
      "loss": 0.015,
      "step": 8889
    },
    {
      "epoch": 2.3840171627782247,
      "grad_norm": 0.40953940755675033,
      "learning_rate": 1.230276004468877e-06,
      "loss": 0.0153,
      "step": 8890
    },
    {
      "epoch": 2.384285331187986,
      "grad_norm": 0.29969263474363145,
      "learning_rate": 1.2292512421975221e-06,
      "loss": 0.0165,
      "step": 8891
    },
    {
      "epoch": 2.3845534995977475,
      "grad_norm": 0.49915587275499396,
      "learning_rate": 1.228226847073961e-06,
      "loss": 0.0171,
      "step": 8892
    },
    {
      "epoch": 2.3848216680075085,
      "grad_norm": 0.23819661281975443,
      "learning_rate": 1.2272028191979391e-06,
      "loss": 0.0141,
      "step": 8893
    },
    {
      "epoch": 2.38508983641727,
      "grad_norm": 0.2981933504414854,
      "learning_rate": 1.2261791586691624e-06,
      "loss": 0.0155,
      "step": 8894
    },
    {
      "epoch": 2.3853580048270313,
      "grad_norm": 0.231977298884141,
      "learning_rate": 1.2251558655873003e-06,
      "loss": 0.0112,
      "step": 8895
    },
    {
      "epoch": 2.3856261732367927,
      "grad_norm": 0.4240465431319797,
      "learning_rate": 1.2241329400519908e-06,
      "loss": 0.0146,
      "step": 8896
    },
    {
      "epoch": 2.385894341646554,
      "grad_norm": 0.2734949785986456,
      "learning_rate": 1.223110382162831e-06,
      "loss": 0.0168,
      "step": 8897
    },
    {
      "epoch": 2.3861625100563155,
      "grad_norm": 0.288096114709897,
      "learning_rate": 1.2220881920193867e-06,
      "loss": 0.015,
      "step": 8898
    },
    {
      "epoch": 2.386430678466077,
      "grad_norm": 0.310926460675642,
      "learning_rate": 1.2210663697211839e-06,
      "loss": 0.0172,
      "step": 8899
    },
    {
      "epoch": 2.386698846875838,
      "grad_norm": 0.34860188484869303,
      "learning_rate": 1.2200449153677158e-06,
      "loss": 0.0119,
      "step": 8900
    },
    {
      "epoch": 2.3869670152855993,
      "grad_norm": 0.22544749190147256,
      "learning_rate": 1.2190238290584404e-06,
      "loss": 0.0109,
      "step": 8901
    },
    {
      "epoch": 2.3872351836953607,
      "grad_norm": 0.23962491533368455,
      "learning_rate": 1.2180031108927747e-06,
      "loss": 0.0145,
      "step": 8902
    },
    {
      "epoch": 2.387503352105122,
      "grad_norm": 0.238123442608474,
      "learning_rate": 1.2169827609701067e-06,
      "loss": 0.0168,
      "step": 8903
    },
    {
      "epoch": 2.3877715205148835,
      "grad_norm": 0.2504086348923343,
      "learning_rate": 1.2159627793897822e-06,
      "loss": 0.0151,
      "step": 8904
    },
    {
      "epoch": 2.3880396889246445,
      "grad_norm": 0.24208487719736688,
      "learning_rate": 1.2149431662511164e-06,
      "loss": 0.0125,
      "step": 8905
    },
    {
      "epoch": 2.388307857334406,
      "grad_norm": 0.2726884796493773,
      "learning_rate": 1.213923921653386e-06,
      "loss": 0.017,
      "step": 8906
    },
    {
      "epoch": 2.3885760257441673,
      "grad_norm": 0.24761661896238568,
      "learning_rate": 1.2129050456958296e-06,
      "loss": 0.0149,
      "step": 8907
    },
    {
      "epoch": 2.3888441941539287,
      "grad_norm": 0.2903650345166876,
      "learning_rate": 1.2118865384776557e-06,
      "loss": 0.0146,
      "step": 8908
    },
    {
      "epoch": 2.38911236256369,
      "grad_norm": 0.2355149650340627,
      "learning_rate": 1.2108684000980314e-06,
      "loss": 0.0155,
      "step": 8909
    },
    {
      "epoch": 2.3893805309734515,
      "grad_norm": 0.17379595049950194,
      "learning_rate": 1.209850630656092e-06,
      "loss": 0.0095,
      "step": 8910
    },
    {
      "epoch": 2.389648699383213,
      "grad_norm": 0.20716854153297726,
      "learning_rate": 1.2088332302509331e-06,
      "loss": 0.0109,
      "step": 8911
    },
    {
      "epoch": 2.389916867792974,
      "grad_norm": 0.2976507143272806,
      "learning_rate": 1.2078161989816167e-06,
      "loss": 0.0148,
      "step": 8912
    },
    {
      "epoch": 2.3901850362027353,
      "grad_norm": 0.19630014860374387,
      "learning_rate": 1.2067995369471707e-06,
      "loss": 0.0091,
      "step": 8913
    },
    {
      "epoch": 2.3904532046124967,
      "grad_norm": 0.2708699642449018,
      "learning_rate": 1.2057832442465812e-06,
      "loss": 0.0152,
      "step": 8914
    },
    {
      "epoch": 2.390721373022258,
      "grad_norm": 0.2450899109755606,
      "learning_rate": 1.2047673209788058e-06,
      "loss": 0.0138,
      "step": 8915
    },
    {
      "epoch": 2.3909895414320195,
      "grad_norm": 0.2498755211362414,
      "learning_rate": 1.2037517672427584e-06,
      "loss": 0.0153,
      "step": 8916
    },
    {
      "epoch": 2.3912577098417804,
      "grad_norm": 0.24208444461348191,
      "learning_rate": 1.2027365831373229e-06,
      "loss": 0.0114,
      "step": 8917
    },
    {
      "epoch": 2.391525878251542,
      "grad_norm": 0.27464987038746524,
      "learning_rate": 1.2017217687613459e-06,
      "loss": 0.0119,
      "step": 8918
    },
    {
      "epoch": 2.3917940466613032,
      "grad_norm": 0.19308007987759734,
      "learning_rate": 1.2007073242136358e-06,
      "loss": 0.0079,
      "step": 8919
    },
    {
      "epoch": 2.3920622150710646,
      "grad_norm": 0.27478667978946936,
      "learning_rate": 1.199693249592967e-06,
      "loss": 0.0129,
      "step": 8920
    },
    {
      "epoch": 2.392330383480826,
      "grad_norm": 0.3724189446196056,
      "learning_rate": 1.1986795449980754e-06,
      "loss": 0.0092,
      "step": 8921
    },
    {
      "epoch": 2.3925985518905875,
      "grad_norm": 0.28723330298404476,
      "learning_rate": 1.1976662105276636e-06,
      "loss": 0.0133,
      "step": 8922
    },
    {
      "epoch": 2.392866720300349,
      "grad_norm": 0.28292753178080493,
      "learning_rate": 1.1966532462804004e-06,
      "loss": 0.0181,
      "step": 8923
    },
    {
      "epoch": 2.39313488871011,
      "grad_norm": 0.1728750320496948,
      "learning_rate": 1.1956406523549107e-06,
      "loss": 0.008,
      "step": 8924
    },
    {
      "epoch": 2.3934030571198712,
      "grad_norm": 0.27196377618620465,
      "learning_rate": 1.1946284288497917e-06,
      "loss": 0.0121,
      "step": 8925
    },
    {
      "epoch": 2.3936712255296326,
      "grad_norm": 0.2575641529535603,
      "learning_rate": 1.1936165758635988e-06,
      "loss": 0.0132,
      "step": 8926
    },
    {
      "epoch": 2.393939393939394,
      "grad_norm": 0.26288909062894616,
      "learning_rate": 1.1926050934948546e-06,
      "loss": 0.0151,
      "step": 8927
    },
    {
      "epoch": 2.3942075623491554,
      "grad_norm": 0.2775865845145447,
      "learning_rate": 1.1915939818420435e-06,
      "loss": 0.0165,
      "step": 8928
    },
    {
      "epoch": 2.3944757307589164,
      "grad_norm": 0.24843238587928054,
      "learning_rate": 1.190583241003615e-06,
      "loss": 0.0115,
      "step": 8929
    },
    {
      "epoch": 2.394743899168678,
      "grad_norm": 0.2084481830677716,
      "learning_rate": 1.1895728710779835e-06,
      "loss": 0.011,
      "step": 8930
    },
    {
      "epoch": 2.395012067578439,
      "grad_norm": 0.2303314316177401,
      "learning_rate": 1.1885628721635256e-06,
      "loss": 0.0146,
      "step": 8931
    },
    {
      "epoch": 2.3952802359882006,
      "grad_norm": 0.348939984788385,
      "learning_rate": 1.1875532443585796e-06,
      "loss": 0.0194,
      "step": 8932
    },
    {
      "epoch": 2.395548404397962,
      "grad_norm": 0.2714156748144347,
      "learning_rate": 1.1865439877614542e-06,
      "loss": 0.0118,
      "step": 8933
    },
    {
      "epoch": 2.3958165728077234,
      "grad_norm": 0.21323958876179125,
      "learning_rate": 1.1855351024704148e-06,
      "loss": 0.0116,
      "step": 8934
    },
    {
      "epoch": 2.396084741217485,
      "grad_norm": 0.2824618230103708,
      "learning_rate": 1.1845265885836971e-06,
      "loss": 0.0148,
      "step": 8935
    },
    {
      "epoch": 2.396352909627246,
      "grad_norm": 0.2558652677011885,
      "learning_rate": 1.1835184461994936e-06,
      "loss": 0.0184,
      "step": 8936
    },
    {
      "epoch": 2.396621078037007,
      "grad_norm": 0.21179821311893277,
      "learning_rate": 1.1825106754159677e-06,
      "loss": 0.02,
      "step": 8937
    },
    {
      "epoch": 2.3968892464467686,
      "grad_norm": 0.22470867529926203,
      "learning_rate": 1.1815032763312412e-06,
      "loss": 0.0127,
      "step": 8938
    },
    {
      "epoch": 2.39715741485653,
      "grad_norm": 0.4285237613611845,
      "learning_rate": 1.1804962490434024e-06,
      "loss": 0.0163,
      "step": 8939
    },
    {
      "epoch": 2.3974255832662914,
      "grad_norm": 0.23467361244108367,
      "learning_rate": 1.1794895936505047e-06,
      "loss": 0.0088,
      "step": 8940
    },
    {
      "epoch": 2.3976937516760524,
      "grad_norm": 0.21376197820355894,
      "learning_rate": 1.1784833102505604e-06,
      "loss": 0.0112,
      "step": 8941
    },
    {
      "epoch": 2.3979619200858138,
      "grad_norm": 0.2841502083921377,
      "learning_rate": 1.1774773989415522e-06,
      "loss": 0.0146,
      "step": 8942
    },
    {
      "epoch": 2.398230088495575,
      "grad_norm": 0.29752993559893237,
      "learning_rate": 1.176471859821421e-06,
      "loss": 0.0137,
      "step": 8943
    },
    {
      "epoch": 2.3984982569053366,
      "grad_norm": 0.2300866100203763,
      "learning_rate": 1.1754666929880714e-06,
      "loss": 0.0129,
      "step": 8944
    },
    {
      "epoch": 2.398766425315098,
      "grad_norm": 0.20249095947609028,
      "learning_rate": 1.1744618985393779e-06,
      "loss": 0.0127,
      "step": 8945
    },
    {
      "epoch": 2.3990345937248594,
      "grad_norm": 0.23461288041588463,
      "learning_rate": 1.1734574765731715e-06,
      "loss": 0.012,
      "step": 8946
    },
    {
      "epoch": 2.3993027621346203,
      "grad_norm": 0.2571935265214276,
      "learning_rate": 1.1724534271872522e-06,
      "loss": 0.0145,
      "step": 8947
    },
    {
      "epoch": 2.3995709305443818,
      "grad_norm": 0.19539378077179323,
      "learning_rate": 1.1714497504793797e-06,
      "loss": 0.0088,
      "step": 8948
    },
    {
      "epoch": 2.399839098954143,
      "grad_norm": 0.20966524970058859,
      "learning_rate": 1.1704464465472802e-06,
      "loss": 0.0118,
      "step": 8949
    },
    {
      "epoch": 2.4001072673639046,
      "grad_norm": 0.24338336768991495,
      "learning_rate": 1.1694435154886446e-06,
      "loss": 0.011,
      "step": 8950
    },
    {
      "epoch": 2.400375435773666,
      "grad_norm": 0.2625569150316832,
      "learning_rate": 1.168440957401123e-06,
      "loss": 0.0154,
      "step": 8951
    },
    {
      "epoch": 2.4006436041834274,
      "grad_norm": 0.26753006701296095,
      "learning_rate": 1.1674387723823332e-06,
      "loss": 0.0153,
      "step": 8952
    },
    {
      "epoch": 2.4009117725931883,
      "grad_norm": 0.22908348152004687,
      "learning_rate": 1.1664369605298543e-06,
      "loss": 0.0139,
      "step": 8953
    },
    {
      "epoch": 2.4011799410029497,
      "grad_norm": 0.34288830887703775,
      "learning_rate": 1.1654355219412316e-06,
      "loss": 0.0216,
      "step": 8954
    },
    {
      "epoch": 2.401448109412711,
      "grad_norm": 0.2993738772339097,
      "learning_rate": 1.1644344567139716e-06,
      "loss": 0.0227,
      "step": 8955
    },
    {
      "epoch": 2.4017162778224725,
      "grad_norm": 0.2540094089158068,
      "learning_rate": 1.163433764945544e-06,
      "loss": 0.0179,
      "step": 8956
    },
    {
      "epoch": 2.401984446232234,
      "grad_norm": 0.3824701523030589,
      "learning_rate": 1.162433446733386e-06,
      "loss": 0.013,
      "step": 8957
    },
    {
      "epoch": 2.4022526146419954,
      "grad_norm": 0.29860198097002477,
      "learning_rate": 1.1614335021748934e-06,
      "loss": 0.0159,
      "step": 8958
    },
    {
      "epoch": 2.4025207830517563,
      "grad_norm": 0.23830131060735016,
      "learning_rate": 1.16043393136743e-06,
      "loss": 0.0207,
      "step": 8959
    },
    {
      "epoch": 2.4027889514615177,
      "grad_norm": 0.2081124445473722,
      "learning_rate": 1.1594347344083218e-06,
      "loss": 0.0113,
      "step": 8960
    },
    {
      "epoch": 2.403057119871279,
      "grad_norm": 0.20663840139198966,
      "learning_rate": 1.1584359113948558e-06,
      "loss": 0.0131,
      "step": 8961
    },
    {
      "epoch": 2.4033252882810405,
      "grad_norm": 0.2665210583092656,
      "learning_rate": 1.1574374624242874e-06,
      "loss": 0.0133,
      "step": 8962
    },
    {
      "epoch": 2.403593456690802,
      "grad_norm": 0.2749575755414612,
      "learning_rate": 1.1564393875938295e-06,
      "loss": 0.0175,
      "step": 8963
    },
    {
      "epoch": 2.4038616251005633,
      "grad_norm": 0.25949339672251504,
      "learning_rate": 1.1554416870006657e-06,
      "loss": 0.0119,
      "step": 8964
    },
    {
      "epoch": 2.4041297935103243,
      "grad_norm": 0.2043555872963266,
      "learning_rate": 1.1544443607419364e-06,
      "loss": 0.0111,
      "step": 8965
    },
    {
      "epoch": 2.4043979619200857,
      "grad_norm": 0.22938426922413468,
      "learning_rate": 1.1534474089147501e-06,
      "loss": 0.0171,
      "step": 8966
    },
    {
      "epoch": 2.404666130329847,
      "grad_norm": 0.2442580051163986,
      "learning_rate": 1.1524508316161799e-06,
      "loss": 0.0111,
      "step": 8967
    },
    {
      "epoch": 2.4049342987396085,
      "grad_norm": 0.2676534371400079,
      "learning_rate": 1.1514546289432543e-06,
      "loss": 0.0148,
      "step": 8968
    },
    {
      "epoch": 2.40520246714937,
      "grad_norm": 0.24368749708155377,
      "learning_rate": 1.150458800992974e-06,
      "loss": 0.0134,
      "step": 8969
    },
    {
      "epoch": 2.4054706355591313,
      "grad_norm": 0.2455797409295501,
      "learning_rate": 1.1494633478623008e-06,
      "loss": 0.017,
      "step": 8970
    },
    {
      "epoch": 2.4057388039688923,
      "grad_norm": 0.21105909614053148,
      "learning_rate": 1.1484682696481575e-06,
      "loss": 0.0133,
      "step": 8971
    },
    {
      "epoch": 2.4060069723786537,
      "grad_norm": 0.2456015665349908,
      "learning_rate": 1.1474735664474346e-06,
      "loss": 0.0119,
      "step": 8972
    },
    {
      "epoch": 2.406275140788415,
      "grad_norm": 0.2373576103175216,
      "learning_rate": 1.1464792383569806e-06,
      "loss": 0.0115,
      "step": 8973
    },
    {
      "epoch": 2.4065433091981765,
      "grad_norm": 0.264275129918367,
      "learning_rate": 1.1454852854736138e-06,
      "loss": 0.0131,
      "step": 8974
    },
    {
      "epoch": 2.406811477607938,
      "grad_norm": 0.2188753446663117,
      "learning_rate": 1.1444917078941099e-06,
      "loss": 0.0112,
      "step": 8975
    },
    {
      "epoch": 2.4070796460176993,
      "grad_norm": 0.362218174027092,
      "learning_rate": 1.143498505715212e-06,
      "loss": 0.0227,
      "step": 8976
    },
    {
      "epoch": 2.4073478144274603,
      "grad_norm": 0.2568216875750802,
      "learning_rate": 1.1425056790336274e-06,
      "loss": 0.0147,
      "step": 8977
    },
    {
      "epoch": 2.4076159828372217,
      "grad_norm": 0.257097226202802,
      "learning_rate": 1.1415132279460217e-06,
      "loss": 0.015,
      "step": 8978
    },
    {
      "epoch": 2.407884151246983,
      "grad_norm": 0.267637388761784,
      "learning_rate": 1.1405211525490307e-06,
      "loss": 0.0153,
      "step": 8979
    },
    {
      "epoch": 2.4081523196567445,
      "grad_norm": 0.23261555592433777,
      "learning_rate": 1.1395294529392481e-06,
      "loss": 0.0113,
      "step": 8980
    },
    {
      "epoch": 2.408420488066506,
      "grad_norm": 0.2088945230170517,
      "learning_rate": 1.138538129213232e-06,
      "loss": 0.0135,
      "step": 8981
    },
    {
      "epoch": 2.4086886564762673,
      "grad_norm": 0.26757456713187644,
      "learning_rate": 1.1375471814675076e-06,
      "loss": 0.0172,
      "step": 8982
    },
    {
      "epoch": 2.4089568248860282,
      "grad_norm": 0.30707992635246745,
      "learning_rate": 1.136556609798558e-06,
      "loss": 0.0178,
      "step": 8983
    },
    {
      "epoch": 2.4092249932957897,
      "grad_norm": 0.18689621928705544,
      "learning_rate": 1.1355664143028356e-06,
      "loss": 0.0108,
      "step": 8984
    },
    {
      "epoch": 2.409493161705551,
      "grad_norm": 0.2204720872926068,
      "learning_rate": 1.1345765950767507e-06,
      "loss": 0.0138,
      "step": 8985
    },
    {
      "epoch": 2.4097613301153125,
      "grad_norm": 0.2252101783296155,
      "learning_rate": 1.1335871522166797e-06,
      "loss": 0.0099,
      "step": 8986
    },
    {
      "epoch": 2.410029498525074,
      "grad_norm": 0.3447113958971166,
      "learning_rate": 1.1325980858189639e-06,
      "loss": 0.0247,
      "step": 8987
    },
    {
      "epoch": 2.4102976669348353,
      "grad_norm": 0.20796525719621287,
      "learning_rate": 1.1316093959799034e-06,
      "loss": 0.0091,
      "step": 8988
    },
    {
      "epoch": 2.4105658353445962,
      "grad_norm": 0.2349923552570824,
      "learning_rate": 1.130621082795767e-06,
      "loss": 0.0135,
      "step": 8989
    },
    {
      "epoch": 2.4108340037543576,
      "grad_norm": 0.2330563868819737,
      "learning_rate": 1.1296331463627814e-06,
      "loss": 0.0152,
      "step": 8990
    },
    {
      "epoch": 2.411102172164119,
      "grad_norm": 0.221006971603871,
      "learning_rate": 1.1286455867771422e-06,
      "loss": 0.0118,
      "step": 8991
    },
    {
      "epoch": 2.4113703405738804,
      "grad_norm": 0.2586053294090216,
      "learning_rate": 1.1276584041350036e-06,
      "loss": 0.0259,
      "step": 8992
    },
    {
      "epoch": 2.411638508983642,
      "grad_norm": 0.35759300409319444,
      "learning_rate": 1.1266715985324843e-06,
      "loss": 0.0157,
      "step": 8993
    },
    {
      "epoch": 2.411906677393403,
      "grad_norm": 0.36323005406335307,
      "learning_rate": 1.1256851700656695e-06,
      "loss": 0.0151,
      "step": 8994
    },
    {
      "epoch": 2.412174845803164,
      "grad_norm": 0.20950723757364045,
      "learning_rate": 1.1246991188306018e-06,
      "loss": 0.0105,
      "step": 8995
    },
    {
      "epoch": 2.4124430142129256,
      "grad_norm": 0.29567405787535755,
      "learning_rate": 1.1237134449232922e-06,
      "loss": 0.0158,
      "step": 8996
    },
    {
      "epoch": 2.412711182622687,
      "grad_norm": 0.2931943115932395,
      "learning_rate": 1.122728148439715e-06,
      "loss": 0.0165,
      "step": 8997
    },
    {
      "epoch": 2.4129793510324484,
      "grad_norm": 0.2265225399155894,
      "learning_rate": 1.1217432294758023e-06,
      "loss": 0.0109,
      "step": 8998
    },
    {
      "epoch": 2.41324751944221,
      "grad_norm": 0.30373948040003457,
      "learning_rate": 1.1207586881274562e-06,
      "loss": 0.014,
      "step": 8999
    },
    {
      "epoch": 2.4135156878519712,
      "grad_norm": 0.2965281453965157,
      "learning_rate": 1.1197745244905362e-06,
      "loss": 0.0127,
      "step": 9000
    },
    {
      "epoch": 2.413783856261732,
      "grad_norm": 0.5315404855850755,
      "learning_rate": 1.1187907386608698e-06,
      "loss": 0.0174,
      "step": 9001
    },
    {
      "epoch": 2.4140520246714936,
      "grad_norm": 0.2215528231046045,
      "learning_rate": 1.117807330734244e-06,
      "loss": 0.0116,
      "step": 9002
    },
    {
      "epoch": 2.414320193081255,
      "grad_norm": 0.24203933761863441,
      "learning_rate": 1.1168243008064123e-06,
      "loss": 0.0104,
      "step": 9003
    },
    {
      "epoch": 2.4145883614910164,
      "grad_norm": 0.3103980599632524,
      "learning_rate": 1.1158416489730895e-06,
      "loss": 0.0187,
      "step": 9004
    },
    {
      "epoch": 2.414856529900778,
      "grad_norm": 0.3142940712897623,
      "learning_rate": 1.114859375329951e-06,
      "loss": 0.0164,
      "step": 9005
    },
    {
      "epoch": 2.4151246983105388,
      "grad_norm": 0.22536701560534933,
      "learning_rate": 1.113877479972642e-06,
      "loss": 0.0164,
      "step": 9006
    },
    {
      "epoch": 2.4153928667203,
      "grad_norm": 0.20302589638445268,
      "learning_rate": 1.112895962996764e-06,
      "loss": 0.0108,
      "step": 9007
    },
    {
      "epoch": 2.4156610351300616,
      "grad_norm": 0.3340456792769076,
      "learning_rate": 1.1119148244978856e-06,
      "loss": 0.0166,
      "step": 9008
    },
    {
      "epoch": 2.415929203539823,
      "grad_norm": 0.2820654035153035,
      "learning_rate": 1.1109340645715394e-06,
      "loss": 0.0246,
      "step": 9009
    },
    {
      "epoch": 2.4161973719495844,
      "grad_norm": 0.2615254818623496,
      "learning_rate": 1.1099536833132173e-06,
      "loss": 0.0161,
      "step": 9010
    },
    {
      "epoch": 2.416465540359346,
      "grad_norm": 0.21645441418326358,
      "learning_rate": 1.1089736808183782e-06,
      "loss": 0.0093,
      "step": 9011
    },
    {
      "epoch": 2.416733708769107,
      "grad_norm": 0.3305579819847133,
      "learning_rate": 1.1079940571824398e-06,
      "loss": 0.0255,
      "step": 9012
    },
    {
      "epoch": 2.417001877178868,
      "grad_norm": 0.2151647832925846,
      "learning_rate": 1.1070148125007874e-06,
      "loss": 0.0124,
      "step": 9013
    },
    {
      "epoch": 2.4172700455886296,
      "grad_norm": 0.3030285629924302,
      "learning_rate": 1.106035946868768e-06,
      "loss": 0.0202,
      "step": 9014
    },
    {
      "epoch": 2.417538213998391,
      "grad_norm": 0.2673913289134322,
      "learning_rate": 1.1050574603816905e-06,
      "loss": 0.0145,
      "step": 9015
    },
    {
      "epoch": 2.4178063824081524,
      "grad_norm": 0.25235668734710826,
      "learning_rate": 1.104079353134827e-06,
      "loss": 0.0134,
      "step": 9016
    },
    {
      "epoch": 2.418074550817914,
      "grad_norm": 0.26143734782930905,
      "learning_rate": 1.1031016252234123e-06,
      "loss": 0.0147,
      "step": 9017
    },
    {
      "epoch": 2.4183427192276747,
      "grad_norm": 0.32754674050455207,
      "learning_rate": 1.102124276742646e-06,
      "loss": 0.0237,
      "step": 9018
    },
    {
      "epoch": 2.418610887637436,
      "grad_norm": 0.21230924975657184,
      "learning_rate": 1.1011473077876916e-06,
      "loss": 0.0109,
      "step": 9019
    },
    {
      "epoch": 2.4188790560471976,
      "grad_norm": 0.2203401196432714,
      "learning_rate": 1.1001707184536714e-06,
      "loss": 0.0161,
      "step": 9020
    },
    {
      "epoch": 2.419147224456959,
      "grad_norm": 0.31127551350874344,
      "learning_rate": 1.099194508835676e-06,
      "loss": 0.0129,
      "step": 9021
    },
    {
      "epoch": 2.4194153928667204,
      "grad_norm": 0.40680028382634426,
      "learning_rate": 1.098218679028753e-06,
      "loss": 0.0189,
      "step": 9022
    },
    {
      "epoch": 2.4196835612764818,
      "grad_norm": 0.28671617662431553,
      "learning_rate": 1.09724322912792e-06,
      "loss": 0.0198,
      "step": 9023
    },
    {
      "epoch": 2.419951729686243,
      "grad_norm": 0.27722171814325464,
      "learning_rate": 1.0962681592281504e-06,
      "loss": 0.0175,
      "step": 9024
    },
    {
      "epoch": 2.420219898096004,
      "grad_norm": 0.21173922240727902,
      "learning_rate": 1.095293469424386e-06,
      "loss": 0.0114,
      "step": 9025
    },
    {
      "epoch": 2.4204880665057655,
      "grad_norm": 0.23040888032505438,
      "learning_rate": 1.094319159811531e-06,
      "loss": 0.0205,
      "step": 9026
    },
    {
      "epoch": 2.420756234915527,
      "grad_norm": 0.258905304366605,
      "learning_rate": 1.0933452304844505e-06,
      "loss": 0.0142,
      "step": 9027
    },
    {
      "epoch": 2.4210244033252883,
      "grad_norm": 0.22784106824899747,
      "learning_rate": 1.0923716815379714e-06,
      "loss": 0.0189,
      "step": 9028
    },
    {
      "epoch": 2.4212925717350497,
      "grad_norm": 0.17988995878797667,
      "learning_rate": 1.0913985130668881e-06,
      "loss": 0.0115,
      "step": 9029
    },
    {
      "epoch": 2.4215607401448107,
      "grad_norm": 0.526037192996088,
      "learning_rate": 1.0904257251659535e-06,
      "loss": 0.0221,
      "step": 9030
    },
    {
      "epoch": 2.421828908554572,
      "grad_norm": 0.21122668235110634,
      "learning_rate": 1.0894533179298877e-06,
      "loss": 0.0116,
      "step": 9031
    },
    {
      "epoch": 2.4220970769643335,
      "grad_norm": 0.25921485026370983,
      "learning_rate": 1.0884812914533683e-06,
      "loss": 0.0153,
      "step": 9032
    },
    {
      "epoch": 2.422365245374095,
      "grad_norm": 0.1783664618367453,
      "learning_rate": 1.0875096458310426e-06,
      "loss": 0.0103,
      "step": 9033
    },
    {
      "epoch": 2.4226334137838563,
      "grad_norm": 0.22760849341335523,
      "learning_rate": 1.086538381157513e-06,
      "loss": 0.0148,
      "step": 9034
    },
    {
      "epoch": 2.4229015821936177,
      "grad_norm": 0.20163762952074663,
      "learning_rate": 1.0855674975273517e-06,
      "loss": 0.0134,
      "step": 9035
    },
    {
      "epoch": 2.423169750603379,
      "grad_norm": 0.23594968666092236,
      "learning_rate": 1.0845969950350921e-06,
      "loss": 0.0135,
      "step": 9036
    },
    {
      "epoch": 2.42343791901314,
      "grad_norm": 0.2496630910614985,
      "learning_rate": 1.083626873775226e-06,
      "loss": 0.0109,
      "step": 9037
    },
    {
      "epoch": 2.4237060874229015,
      "grad_norm": 0.24237573922825387,
      "learning_rate": 1.0826571338422153e-06,
      "loss": 0.0129,
      "step": 9038
    },
    {
      "epoch": 2.423974255832663,
      "grad_norm": 0.35377429247880243,
      "learning_rate": 1.0816877753304777e-06,
      "loss": 0.0242,
      "step": 9039
    },
    {
      "epoch": 2.4242424242424243,
      "grad_norm": 0.2545030341173104,
      "learning_rate": 1.0807187983343998e-06,
      "loss": 0.0138,
      "step": 9040
    },
    {
      "epoch": 2.4245105926521857,
      "grad_norm": 0.27251373935521084,
      "learning_rate": 1.079750202948327e-06,
      "loss": 0.0203,
      "step": 9041
    },
    {
      "epoch": 2.4247787610619467,
      "grad_norm": 0.24023004743675008,
      "learning_rate": 1.0787819892665673e-06,
      "loss": 0.0167,
      "step": 9042
    },
    {
      "epoch": 2.425046929471708,
      "grad_norm": 0.2151738081314473,
      "learning_rate": 1.0778141573833962e-06,
      "loss": 0.0139,
      "step": 9043
    },
    {
      "epoch": 2.4253150978814695,
      "grad_norm": 0.23397875991047445,
      "learning_rate": 1.076846707393046e-06,
      "loss": 0.0118,
      "step": 9044
    },
    {
      "epoch": 2.425583266291231,
      "grad_norm": 0.2596426155955461,
      "learning_rate": 1.0758796393897163e-06,
      "loss": 0.0164,
      "step": 9045
    },
    {
      "epoch": 2.4258514347009923,
      "grad_norm": 0.3022000146567599,
      "learning_rate": 1.074912953467569e-06,
      "loss": 0.0182,
      "step": 9046
    },
    {
      "epoch": 2.4261196031107537,
      "grad_norm": 0.27741510558731025,
      "learning_rate": 1.0739466497207252e-06,
      "loss": 0.0182,
      "step": 9047
    },
    {
      "epoch": 2.426387771520515,
      "grad_norm": 0.22197595237692136,
      "learning_rate": 1.0729807282432737e-06,
      "loss": 0.0157,
      "step": 9048
    },
    {
      "epoch": 2.426655939930276,
      "grad_norm": 0.3109968939512372,
      "learning_rate": 1.0720151891292619e-06,
      "loss": 0.0186,
      "step": 9049
    },
    {
      "epoch": 2.4269241083400375,
      "grad_norm": 0.21830328781493383,
      "learning_rate": 1.0710500324727035e-06,
      "loss": 0.0141,
      "step": 9050
    },
    {
      "epoch": 2.427192276749799,
      "grad_norm": 0.2418846349771064,
      "learning_rate": 1.0700852583675708e-06,
      "loss": 0.0123,
      "step": 9051
    },
    {
      "epoch": 2.4274604451595603,
      "grad_norm": 0.25278427112603036,
      "learning_rate": 1.0691208669078045e-06,
      "loss": 0.0198,
      "step": 9052
    },
    {
      "epoch": 2.4277286135693217,
      "grad_norm": 0.24640597456561303,
      "learning_rate": 1.0681568581873031e-06,
      "loss": 0.0125,
      "step": 9053
    },
    {
      "epoch": 2.4279967819790826,
      "grad_norm": 0.2405237812770578,
      "learning_rate": 1.067193232299928e-06,
      "loss": 0.0137,
      "step": 9054
    },
    {
      "epoch": 2.428264950388844,
      "grad_norm": 0.1954389754531959,
      "learning_rate": 1.0662299893395062e-06,
      "loss": 0.0109,
      "step": 9055
    },
    {
      "epoch": 2.4285331187986055,
      "grad_norm": 0.21554416138708013,
      "learning_rate": 1.065267129399828e-06,
      "loss": 0.0103,
      "step": 9056
    },
    {
      "epoch": 2.428801287208367,
      "grad_norm": 0.2350961010780986,
      "learning_rate": 1.0643046525746414e-06,
      "loss": 0.0141,
      "step": 9057
    },
    {
      "epoch": 2.4290694556181283,
      "grad_norm": 0.26794093868997493,
      "learning_rate": 1.0633425589576636e-06,
      "loss": 0.0186,
      "step": 9058
    },
    {
      "epoch": 2.4293376240278897,
      "grad_norm": 0.2753119583072202,
      "learning_rate": 1.0623808486425674e-06,
      "loss": 0.0147,
      "step": 9059
    },
    {
      "epoch": 2.429605792437651,
      "grad_norm": 0.3086129590470031,
      "learning_rate": 1.0614195217229957e-06,
      "loss": 0.0234,
      "step": 9060
    },
    {
      "epoch": 2.429873960847412,
      "grad_norm": 0.20758031101122348,
      "learning_rate": 1.0604585782925465e-06,
      "loss": 0.013,
      "step": 9061
    },
    {
      "epoch": 2.4301421292571734,
      "grad_norm": 0.23980291178912524,
      "learning_rate": 1.059498018444786e-06,
      "loss": 0.0112,
      "step": 9062
    },
    {
      "epoch": 2.430410297666935,
      "grad_norm": 0.24558960006564906,
      "learning_rate": 1.0585378422732435e-06,
      "loss": 0.0131,
      "step": 9063
    },
    {
      "epoch": 2.4306784660766962,
      "grad_norm": 0.24156801034628894,
      "learning_rate": 1.0575780498714067e-06,
      "loss": 0.0137,
      "step": 9064
    },
    {
      "epoch": 2.4309466344864576,
      "grad_norm": 0.22565744533826532,
      "learning_rate": 1.0566186413327274e-06,
      "loss": 0.0134,
      "step": 9065
    },
    {
      "epoch": 2.4312148028962186,
      "grad_norm": 0.2812350174966112,
      "learning_rate": 1.0556596167506223e-06,
      "loss": 0.0109,
      "step": 9066
    },
    {
      "epoch": 2.43148297130598,
      "grad_norm": 0.28176904720687085,
      "learning_rate": 1.0547009762184667e-06,
      "loss": 0.0122,
      "step": 9067
    },
    {
      "epoch": 2.4317511397157414,
      "grad_norm": 0.3192953045772836,
      "learning_rate": 1.0537427198296046e-06,
      "loss": 0.0172,
      "step": 9068
    },
    {
      "epoch": 2.432019308125503,
      "grad_norm": 0.1968057586420274,
      "learning_rate": 1.052784847677335e-06,
      "loss": 0.0119,
      "step": 9069
    },
    {
      "epoch": 2.4322874765352642,
      "grad_norm": 0.24941351032200065,
      "learning_rate": 1.0518273598549267e-06,
      "loss": 0.015,
      "step": 9070
    },
    {
      "epoch": 2.4325556449450256,
      "grad_norm": 0.19255185992171772,
      "learning_rate": 1.0508702564556045e-06,
      "loss": 0.01,
      "step": 9071
    },
    {
      "epoch": 2.432823813354787,
      "grad_norm": 0.20205885516707056,
      "learning_rate": 1.0499135375725606e-06,
      "loss": 0.0131,
      "step": 9072
    },
    {
      "epoch": 2.433091981764548,
      "grad_norm": 0.2938080741917749,
      "learning_rate": 1.0489572032989498e-06,
      "loss": 0.0184,
      "step": 9073
    },
    {
      "epoch": 2.4333601501743094,
      "grad_norm": 0.23834134069507926,
      "learning_rate": 1.048001253727885e-06,
      "loss": 0.0119,
      "step": 9074
    },
    {
      "epoch": 2.433628318584071,
      "grad_norm": 0.2765386961751854,
      "learning_rate": 1.0470456889524473e-06,
      "loss": 0.0175,
      "step": 9075
    },
    {
      "epoch": 2.433896486993832,
      "grad_norm": 0.2646075931019891,
      "learning_rate": 1.046090509065676e-06,
      "loss": 0.0166,
      "step": 9076
    },
    {
      "epoch": 2.4341646554035936,
      "grad_norm": 0.25916736941990226,
      "learning_rate": 1.0451357141605728e-06,
      "loss": 0.0154,
      "step": 9077
    },
    {
      "epoch": 2.4344328238133546,
      "grad_norm": 0.21120882958734571,
      "learning_rate": 1.0441813043301064e-06,
      "loss": 0.0091,
      "step": 9078
    },
    {
      "epoch": 2.434700992223116,
      "grad_norm": 0.2941808092924196,
      "learning_rate": 1.0432272796672028e-06,
      "loss": 0.0121,
      "step": 9079
    },
    {
      "epoch": 2.4349691606328774,
      "grad_norm": 0.916290487157292,
      "learning_rate": 1.0422736402647548e-06,
      "loss": 0.0156,
      "step": 9080
    },
    {
      "epoch": 2.435237329042639,
      "grad_norm": 0.26048983439635126,
      "learning_rate": 1.041320386215614e-06,
      "loss": 0.0157,
      "step": 9081
    },
    {
      "epoch": 2.4355054974524,
      "grad_norm": 0.21262406827303099,
      "learning_rate": 1.0403675176125971e-06,
      "loss": 0.0125,
      "step": 9082
    },
    {
      "epoch": 2.4357736658621616,
      "grad_norm": 0.26436806913504657,
      "learning_rate": 1.0394150345484838e-06,
      "loss": 0.0135,
      "step": 9083
    },
    {
      "epoch": 2.436041834271923,
      "grad_norm": 0.25486126802829767,
      "learning_rate": 1.0384629371160122e-06,
      "loss": 0.0137,
      "step": 9084
    },
    {
      "epoch": 2.436310002681684,
      "grad_norm": 0.25583689636177115,
      "learning_rate": 1.0375112254078878e-06,
      "loss": 0.012,
      "step": 9085
    },
    {
      "epoch": 2.4365781710914454,
      "grad_norm": 0.20381927491149499,
      "learning_rate": 1.0365598995167736e-06,
      "loss": 0.0111,
      "step": 9086
    },
    {
      "epoch": 2.4368463395012068,
      "grad_norm": 0.20003463016584686,
      "learning_rate": 1.0356089595353008e-06,
      "loss": 0.0103,
      "step": 9087
    },
    {
      "epoch": 2.437114507910968,
      "grad_norm": 0.23715222517157653,
      "learning_rate": 1.0346584055560587e-06,
      "loss": 0.0128,
      "step": 9088
    },
    {
      "epoch": 2.4373826763207296,
      "grad_norm": 0.28597333888507026,
      "learning_rate": 1.0337082376715985e-06,
      "loss": 0.0159,
      "step": 9089
    },
    {
      "epoch": 2.4376508447304905,
      "grad_norm": 0.25194055810146543,
      "learning_rate": 1.032758455974438e-06,
      "loss": 0.0124,
      "step": 9090
    },
    {
      "epoch": 2.437919013140252,
      "grad_norm": 0.268752749077263,
      "learning_rate": 1.0318090605570523e-06,
      "loss": 0.0165,
      "step": 9091
    },
    {
      "epoch": 2.4381871815500133,
      "grad_norm": 0.6863306637929121,
      "learning_rate": 1.0308600515118834e-06,
      "loss": 0.0257,
      "step": 9092
    },
    {
      "epoch": 2.4384553499597748,
      "grad_norm": 0.22817493977227393,
      "learning_rate": 1.0299114289313344e-06,
      "loss": 0.0107,
      "step": 9093
    },
    {
      "epoch": 2.438723518369536,
      "grad_norm": 0.19288011275645034,
      "learning_rate": 1.028963192907768e-06,
      "loss": 0.0062,
      "step": 9094
    },
    {
      "epoch": 2.4389916867792976,
      "grad_norm": 0.21612729415146228,
      "learning_rate": 1.0280153435335139e-06,
      "loss": 0.0117,
      "step": 9095
    },
    {
      "epoch": 2.439259855189059,
      "grad_norm": 0.277536117826134,
      "learning_rate": 1.0270678809008594e-06,
      "loss": 0.0164,
      "step": 9096
    },
    {
      "epoch": 2.43952802359882,
      "grad_norm": 0.2918040960978186,
      "learning_rate": 1.0261208051020584e-06,
      "loss": 0.0137,
      "step": 9097
    },
    {
      "epoch": 2.4397961920085813,
      "grad_norm": 0.29771386613234274,
      "learning_rate": 1.0251741162293228e-06,
      "loss": 0.0165,
      "step": 9098
    },
    {
      "epoch": 2.4400643604183427,
      "grad_norm": 0.25496286299464455,
      "learning_rate": 1.0242278143748307e-06,
      "loss": 0.0108,
      "step": 9099
    },
    {
      "epoch": 2.440332528828104,
      "grad_norm": 0.20296479843569679,
      "learning_rate": 1.023281899630723e-06,
      "loss": 0.0106,
      "step": 9100
    },
    {
      "epoch": 2.4406006972378655,
      "grad_norm": 0.2729067625042114,
      "learning_rate": 1.0223363720890966e-06,
      "loss": 0.0113,
      "step": 9101
    },
    {
      "epoch": 2.4408688656476265,
      "grad_norm": 0.21198395923082908,
      "learning_rate": 1.0213912318420189e-06,
      "loss": 0.0111,
      "step": 9102
    },
    {
      "epoch": 2.441137034057388,
      "grad_norm": 0.2614452534793551,
      "learning_rate": 1.0204464789815122e-06,
      "loss": 0.0235,
      "step": 9103
    },
    {
      "epoch": 2.4414052024671493,
      "grad_norm": 0.2878785904861613,
      "learning_rate": 1.0195021135995658e-06,
      "loss": 0.0196,
      "step": 9104
    },
    {
      "epoch": 2.4416733708769107,
      "grad_norm": 0.23574699285158554,
      "learning_rate": 1.0185581357881325e-06,
      "loss": 0.0119,
      "step": 9105
    },
    {
      "epoch": 2.441941539286672,
      "grad_norm": 0.32303435496754435,
      "learning_rate": 1.0176145456391219e-06,
      "loss": 0.0128,
      "step": 9106
    },
    {
      "epoch": 2.4422097076964335,
      "grad_norm": 0.28010618977728075,
      "learning_rate": 1.0166713432444108e-06,
      "loss": 0.015,
      "step": 9107
    },
    {
      "epoch": 2.442477876106195,
      "grad_norm": 0.5614975562030432,
      "learning_rate": 1.015728528695834e-06,
      "loss": 0.0194,
      "step": 9108
    },
    {
      "epoch": 2.442746044515956,
      "grad_norm": 0.22129816851623307,
      "learning_rate": 1.0147861020851929e-06,
      "loss": 0.0138,
      "step": 9109
    },
    {
      "epoch": 2.4430142129257173,
      "grad_norm": 0.2623286495545508,
      "learning_rate": 1.0138440635042495e-06,
      "loss": 0.0115,
      "step": 9110
    },
    {
      "epoch": 2.4432823813354787,
      "grad_norm": 0.32372058895317196,
      "learning_rate": 1.012902413044725e-06,
      "loss": 0.0127,
      "step": 9111
    },
    {
      "epoch": 2.44355054974524,
      "grad_norm": 0.1925924332488762,
      "learning_rate": 1.0119611507983102e-06,
      "loss": 0.0111,
      "step": 9112
    },
    {
      "epoch": 2.4438187181550015,
      "grad_norm": 0.2288066064517032,
      "learning_rate": 1.0110202768566474e-06,
      "loss": 0.0097,
      "step": 9113
    },
    {
      "epoch": 2.4440868865647625,
      "grad_norm": 0.26058351563234644,
      "learning_rate": 1.0100797913113491e-06,
      "loss": 0.0143,
      "step": 9114
    },
    {
      "epoch": 2.444355054974524,
      "grad_norm": 0.3050554251588954,
      "learning_rate": 1.0091396942539904e-06,
      "loss": 0.0157,
      "step": 9115
    },
    {
      "epoch": 2.4446232233842853,
      "grad_norm": 0.22465701055104542,
      "learning_rate": 1.0081999857761026e-06,
      "loss": 0.0135,
      "step": 9116
    },
    {
      "epoch": 2.4448913917940467,
      "grad_norm": 0.18604000115387828,
      "learning_rate": 1.007260665969186e-06,
      "loss": 0.0089,
      "step": 9117
    },
    {
      "epoch": 2.445159560203808,
      "grad_norm": 0.3226109923379357,
      "learning_rate": 1.0063217349246956e-06,
      "loss": 0.0176,
      "step": 9118
    },
    {
      "epoch": 2.4454277286135695,
      "grad_norm": 0.23008113046334286,
      "learning_rate": 1.005383192734057e-06,
      "loss": 0.0101,
      "step": 9119
    },
    {
      "epoch": 2.445695897023331,
      "grad_norm": 0.22023505625016718,
      "learning_rate": 1.0044450394886496e-06,
      "loss": 0.0106,
      "step": 9120
    },
    {
      "epoch": 2.445964065433092,
      "grad_norm": 0.2294634620987983,
      "learning_rate": 1.0035072752798208e-06,
      "loss": 0.0118,
      "step": 9121
    },
    {
      "epoch": 2.4462322338428533,
      "grad_norm": 0.16497875276822027,
      "learning_rate": 1.0025699001988793e-06,
      "loss": 0.0085,
      "step": 9122
    },
    {
      "epoch": 2.4465004022526147,
      "grad_norm": 0.2968024056540365,
      "learning_rate": 1.0016329143370929e-06,
      "loss": 0.0206,
      "step": 9123
    },
    {
      "epoch": 2.446768570662376,
      "grad_norm": 0.27301448401906914,
      "learning_rate": 1.000696317785695e-06,
      "loss": 0.0165,
      "step": 9124
    },
    {
      "epoch": 2.4470367390721375,
      "grad_norm": 0.40771136280637615,
      "learning_rate": 9.997601106358785e-07,
      "loss": 0.0192,
      "step": 9125
    },
    {
      "epoch": 2.4473049074818984,
      "grad_norm": 0.2600030053596715,
      "learning_rate": 9.98824292978799e-07,
      "loss": 0.0145,
      "step": 9126
    },
    {
      "epoch": 2.44757307589166,
      "grad_norm": 0.23410809789546616,
      "learning_rate": 9.978888649055757e-07,
      "loss": 0.0099,
      "step": 9127
    },
    {
      "epoch": 2.4478412443014212,
      "grad_norm": 0.34575040072198565,
      "learning_rate": 9.969538265072876e-07,
      "loss": 0.0169,
      "step": 9128
    },
    {
      "epoch": 2.4481094127111827,
      "grad_norm": 0.34864998962584937,
      "learning_rate": 9.960191778749783e-07,
      "loss": 0.0175,
      "step": 9129
    },
    {
      "epoch": 2.448377581120944,
      "grad_norm": 0.24647498095747095,
      "learning_rate": 9.950849190996503e-07,
      "loss": 0.0124,
      "step": 9130
    },
    {
      "epoch": 2.4486457495307055,
      "grad_norm": 0.2783685073756563,
      "learning_rate": 9.941510502722707e-07,
      "loss": 0.0211,
      "step": 9131
    },
    {
      "epoch": 2.4489139179404664,
      "grad_norm": 0.29366357305368285,
      "learning_rate": 9.932175714837695e-07,
      "loss": 0.0193,
      "step": 9132
    },
    {
      "epoch": 2.449182086350228,
      "grad_norm": 0.23399124493062584,
      "learning_rate": 9.922844828250339e-07,
      "loss": 0.0129,
      "step": 9133
    },
    {
      "epoch": 2.4494502547599892,
      "grad_norm": 0.3924118652029773,
      "learning_rate": 9.913517843869197e-07,
      "loss": 0.0213,
      "step": 9134
    },
    {
      "epoch": 2.4497184231697506,
      "grad_norm": 0.25679192881433266,
      "learning_rate": 9.904194762602382e-07,
      "loss": 0.0117,
      "step": 9135
    },
    {
      "epoch": 2.449986591579512,
      "grad_norm": 0.31301070974190776,
      "learning_rate": 9.894875585357678e-07,
      "loss": 0.0205,
      "step": 9136
    },
    {
      "epoch": 2.4502547599892734,
      "grad_norm": 0.3004325418934879,
      "learning_rate": 9.885560313042464e-07,
      "loss": 0.0169,
      "step": 9137
    },
    {
      "epoch": 2.4505229283990344,
      "grad_norm": 0.17051049485653105,
      "learning_rate": 9.876248946563732e-07,
      "loss": 0.0076,
      "step": 9138
    },
    {
      "epoch": 2.450791096808796,
      "grad_norm": 0.2517870537487761,
      "learning_rate": 9.866941486828119e-07,
      "loss": 0.0175,
      "step": 9139
    },
    {
      "epoch": 2.451059265218557,
      "grad_norm": 0.20846544256639823,
      "learning_rate": 9.857637934741854e-07,
      "loss": 0.0141,
      "step": 9140
    },
    {
      "epoch": 2.4513274336283186,
      "grad_norm": 0.26846130221518455,
      "learning_rate": 9.848338291210807e-07,
      "loss": 0.0184,
      "step": 9141
    },
    {
      "epoch": 2.45159560203808,
      "grad_norm": 0.30482564251824906,
      "learning_rate": 9.839042557140472e-07,
      "loss": 0.0137,
      "step": 9142
    },
    {
      "epoch": 2.4518637704478414,
      "grad_norm": 0.18919862392541226,
      "learning_rate": 9.829750733435928e-07,
      "loss": 0.0095,
      "step": 9143
    },
    {
      "epoch": 2.4521319388576024,
      "grad_norm": 0.20190728646544717,
      "learning_rate": 9.820462821001913e-07,
      "loss": 0.0093,
      "step": 9144
    },
    {
      "epoch": 2.452400107267364,
      "grad_norm": 0.22804541036683662,
      "learning_rate": 9.811178820742751e-07,
      "loss": 0.0103,
      "step": 9145
    },
    {
      "epoch": 2.452668275677125,
      "grad_norm": 0.24552412353800554,
      "learning_rate": 9.801898733562415e-07,
      "loss": 0.012,
      "step": 9146
    },
    {
      "epoch": 2.4529364440868866,
      "grad_norm": 0.24316902822533418,
      "learning_rate": 9.792622560364467e-07,
      "loss": 0.0131,
      "step": 9147
    },
    {
      "epoch": 2.453204612496648,
      "grad_norm": 0.27237942060933457,
      "learning_rate": 9.783350302052124e-07,
      "loss": 0.0169,
      "step": 9148
    },
    {
      "epoch": 2.4534727809064094,
      "grad_norm": 0.23938322070577253,
      "learning_rate": 9.774081959528186e-07,
      "loss": 0.013,
      "step": 9149
    },
    {
      "epoch": 2.4537409493161704,
      "grad_norm": 0.26649978161389626,
      "learning_rate": 9.76481753369508e-07,
      "loss": 0.0166,
      "step": 9150
    },
    {
      "epoch": 2.4540091177259318,
      "grad_norm": 0.2607979516496919,
      "learning_rate": 9.755557025454865e-07,
      "loss": 0.0125,
      "step": 9151
    },
    {
      "epoch": 2.454277286135693,
      "grad_norm": 0.2703144633188623,
      "learning_rate": 9.746300435709233e-07,
      "loss": 0.0115,
      "step": 9152
    },
    {
      "epoch": 2.4545454545454546,
      "grad_norm": 0.26188508396072446,
      "learning_rate": 9.737047765359443e-07,
      "loss": 0.0142,
      "step": 9153
    },
    {
      "epoch": 2.454813622955216,
      "grad_norm": 0.2435200020645698,
      "learning_rate": 9.727799015306427e-07,
      "loss": 0.0173,
      "step": 9154
    },
    {
      "epoch": 2.4550817913649774,
      "grad_norm": 0.234523563020842,
      "learning_rate": 9.718554186450686e-07,
      "loss": 0.011,
      "step": 9155
    },
    {
      "epoch": 2.4553499597747384,
      "grad_norm": 0.23507239376410358,
      "learning_rate": 9.70931327969239e-07,
      "loss": 0.0106,
      "step": 9156
    },
    {
      "epoch": 2.4556181281844998,
      "grad_norm": 0.28793634651708705,
      "learning_rate": 9.700076295931281e-07,
      "loss": 0.0168,
      "step": 9157
    },
    {
      "epoch": 2.455886296594261,
      "grad_norm": 0.20399842206377147,
      "learning_rate": 9.69084323606675e-07,
      "loss": 0.0126,
      "step": 9158
    },
    {
      "epoch": 2.4561544650040226,
      "grad_norm": 0.2059213327754596,
      "learning_rate": 9.681614100997806e-07,
      "loss": 0.0123,
      "step": 9159
    },
    {
      "epoch": 2.456422633413784,
      "grad_norm": 0.31585661275508586,
      "learning_rate": 9.672388891623035e-07,
      "loss": 0.0135,
      "step": 9160
    },
    {
      "epoch": 2.4566908018235454,
      "grad_norm": 0.24657097062572503,
      "learning_rate": 9.663167608840711e-07,
      "loss": 0.014,
      "step": 9161
    },
    {
      "epoch": 2.4569589702333063,
      "grad_norm": 0.26050884219817527,
      "learning_rate": 9.65395025354866e-07,
      "loss": 0.0139,
      "step": 9162
    },
    {
      "epoch": 2.4572271386430677,
      "grad_norm": 0.3087878877371412,
      "learning_rate": 9.644736826644347e-07,
      "loss": 0.019,
      "step": 9163
    },
    {
      "epoch": 2.457495307052829,
      "grad_norm": 0.2777293432405128,
      "learning_rate": 9.63552732902488e-07,
      "loss": 0.0151,
      "step": 9164
    },
    {
      "epoch": 2.4577634754625906,
      "grad_norm": 0.2003181352156441,
      "learning_rate": 9.626321761586937e-07,
      "loss": 0.0116,
      "step": 9165
    },
    {
      "epoch": 2.458031643872352,
      "grad_norm": 0.22121233788334033,
      "learning_rate": 9.617120125226864e-07,
      "loss": 0.0111,
      "step": 9166
    },
    {
      "epoch": 2.4582998122821134,
      "grad_norm": 0.2084194033245839,
      "learning_rate": 9.607922420840583e-07,
      "loss": 0.0109,
      "step": 9167
    },
    {
      "epoch": 2.4585679806918743,
      "grad_norm": 2.658693526867989,
      "learning_rate": 9.59872864932365e-07,
      "loss": 0.0242,
      "step": 9168
    },
    {
      "epoch": 2.4588361491016357,
      "grad_norm": 0.358666355856082,
      "learning_rate": 9.589538811571265e-07,
      "loss": 0.0237,
      "step": 9169
    },
    {
      "epoch": 2.459104317511397,
      "grad_norm": 0.22743863409401832,
      "learning_rate": 9.580352908478184e-07,
      "loss": 0.0094,
      "step": 9170
    },
    {
      "epoch": 2.4593724859211585,
      "grad_norm": 0.3505622264692479,
      "learning_rate": 9.57117094093884e-07,
      "loss": 0.0129,
      "step": 9171
    },
    {
      "epoch": 2.45964065433092,
      "grad_norm": 0.24496157889107473,
      "learning_rate": 9.561992909847235e-07,
      "loss": 0.0129,
      "step": 9172
    },
    {
      "epoch": 2.4599088227406813,
      "grad_norm": 0.18498047063961395,
      "learning_rate": 9.552818816097026e-07,
      "loss": 0.0094,
      "step": 9173
    },
    {
      "epoch": 2.4601769911504423,
      "grad_norm": 0.364705860626398,
      "learning_rate": 9.543648660581466e-07,
      "loss": 0.024,
      "step": 9174
    },
    {
      "epoch": 2.4604451595602037,
      "grad_norm": 0.5617333184932559,
      "learning_rate": 9.534482444193416e-07,
      "loss": 0.0141,
      "step": 9175
    },
    {
      "epoch": 2.460713327969965,
      "grad_norm": 0.23894138948062665,
      "learning_rate": 9.525320167825386e-07,
      "loss": 0.0108,
      "step": 9176
    },
    {
      "epoch": 2.4609814963797265,
      "grad_norm": 0.2524481967325974,
      "learning_rate": 9.516161832369464e-07,
      "loss": 0.0134,
      "step": 9177
    },
    {
      "epoch": 2.461249664789488,
      "grad_norm": 0.2651150093550674,
      "learning_rate": 9.50700743871738e-07,
      "loss": 0.0117,
      "step": 9178
    },
    {
      "epoch": 2.461517833199249,
      "grad_norm": 0.28577626381068877,
      "learning_rate": 9.497856987760484e-07,
      "loss": 0.0138,
      "step": 9179
    },
    {
      "epoch": 2.4617860016090103,
      "grad_norm": 0.23908365808011442,
      "learning_rate": 9.488710480389718e-07,
      "loss": 0.0115,
      "step": 9180
    },
    {
      "epoch": 2.4620541700187717,
      "grad_norm": 0.2416333645910043,
      "learning_rate": 9.479567917495664e-07,
      "loss": 0.0169,
      "step": 9181
    },
    {
      "epoch": 2.462322338428533,
      "grad_norm": 0.2708626835498375,
      "learning_rate": 9.470429299968487e-07,
      "loss": 0.0166,
      "step": 9182
    },
    {
      "epoch": 2.4625905068382945,
      "grad_norm": 0.25508976425393765,
      "learning_rate": 9.46129462869802e-07,
      "loss": 0.0142,
      "step": 9183
    },
    {
      "epoch": 2.462858675248056,
      "grad_norm": 0.23599914868631805,
      "learning_rate": 9.452163904573658e-07,
      "loss": 0.0198,
      "step": 9184
    },
    {
      "epoch": 2.4631268436578173,
      "grad_norm": 0.25178989538739915,
      "learning_rate": 9.443037128484456e-07,
      "loss": 0.0107,
      "step": 9185
    },
    {
      "epoch": 2.4633950120675783,
      "grad_norm": 0.23814964649319523,
      "learning_rate": 9.433914301319047e-07,
      "loss": 0.0183,
      "step": 9186
    },
    {
      "epoch": 2.4636631804773397,
      "grad_norm": 0.24601129678752232,
      "learning_rate": 9.424795423965694e-07,
      "loss": 0.0145,
      "step": 9187
    },
    {
      "epoch": 2.463931348887101,
      "grad_norm": 0.28716767317147235,
      "learning_rate": 9.415680497312286e-07,
      "loss": 0.0167,
      "step": 9188
    },
    {
      "epoch": 2.4641995172968625,
      "grad_norm": 0.2866392488354706,
      "learning_rate": 9.406569522246328e-07,
      "loss": 0.018,
      "step": 9189
    },
    {
      "epoch": 2.464467685706624,
      "grad_norm": 0.197146445862933,
      "learning_rate": 9.397462499654907e-07,
      "loss": 0.0091,
      "step": 9190
    },
    {
      "epoch": 2.464735854116385,
      "grad_norm": 0.26896943471059503,
      "learning_rate": 9.388359430424781e-07,
      "loss": 0.011,
      "step": 9191
    },
    {
      "epoch": 2.4650040225261463,
      "grad_norm": 0.24824809911805282,
      "learning_rate": 9.379260315442262e-07,
      "loss": 0.0121,
      "step": 9192
    },
    {
      "epoch": 2.4652721909359077,
      "grad_norm": 0.21176101340558656,
      "learning_rate": 9.370165155593336e-07,
      "loss": 0.0104,
      "step": 9193
    },
    {
      "epoch": 2.465540359345669,
      "grad_norm": 0.2340141063259027,
      "learning_rate": 9.361073951763544e-07,
      "loss": 0.0146,
      "step": 9194
    },
    {
      "epoch": 2.4658085277554305,
      "grad_norm": 0.25249987942683655,
      "learning_rate": 9.351986704838084e-07,
      "loss": 0.015,
      "step": 9195
    },
    {
      "epoch": 2.466076696165192,
      "grad_norm": 0.3391748678436711,
      "learning_rate": 9.342903415701776e-07,
      "loss": 0.0311,
      "step": 9196
    },
    {
      "epoch": 2.4663448645749533,
      "grad_norm": 0.24056745241319946,
      "learning_rate": 9.333824085239018e-07,
      "loss": 0.0108,
      "step": 9197
    },
    {
      "epoch": 2.4666130329847142,
      "grad_norm": 0.2807332882703198,
      "learning_rate": 9.324748714333843e-07,
      "loss": 0.0177,
      "step": 9198
    },
    {
      "epoch": 2.4668812013944756,
      "grad_norm": 0.28351082786320636,
      "learning_rate": 9.315677303869885e-07,
      "loss": 0.0229,
      "step": 9199
    },
    {
      "epoch": 2.467149369804237,
      "grad_norm": 0.24293759022917852,
      "learning_rate": 9.306609854730409e-07,
      "loss": 0.0157,
      "step": 9200
    },
    {
      "epoch": 2.4674175382139985,
      "grad_norm": 0.20421934596844485,
      "learning_rate": 9.29754636779831e-07,
      "loss": 0.0142,
      "step": 9201
    },
    {
      "epoch": 2.46768570662376,
      "grad_norm": 0.20201085911995098,
      "learning_rate": 9.288486843956041e-07,
      "loss": 0.0127,
      "step": 9202
    },
    {
      "epoch": 2.467953875033521,
      "grad_norm": 0.24016654529758918,
      "learning_rate": 9.279431284085739e-07,
      "loss": 0.011,
      "step": 9203
    },
    {
      "epoch": 2.468222043443282,
      "grad_norm": 0.3278515204587565,
      "learning_rate": 9.270379689069081e-07,
      "loss": 0.014,
      "step": 9204
    },
    {
      "epoch": 2.4684902118530436,
      "grad_norm": 0.5102940585112304,
      "learning_rate": 9.261332059787425e-07,
      "loss": 0.0214,
      "step": 9205
    },
    {
      "epoch": 2.468758380262805,
      "grad_norm": 0.3695663672414119,
      "learning_rate": 9.252288397121717e-07,
      "loss": 0.0305,
      "step": 9206
    },
    {
      "epoch": 2.4690265486725664,
      "grad_norm": 0.21050008625065145,
      "learning_rate": 9.243248701952489e-07,
      "loss": 0.0107,
      "step": 9207
    },
    {
      "epoch": 2.469294717082328,
      "grad_norm": 0.23246668435878182,
      "learning_rate": 9.234212975159946e-07,
      "loss": 0.0134,
      "step": 9208
    },
    {
      "epoch": 2.4695628854920892,
      "grad_norm": 0.22608056272490695,
      "learning_rate": 9.225181217623852e-07,
      "loss": 0.0099,
      "step": 9209
    },
    {
      "epoch": 2.46983105390185,
      "grad_norm": 0.2893370296497248,
      "learning_rate": 9.216153430223595e-07,
      "loss": 0.0124,
      "step": 9210
    },
    {
      "epoch": 2.4700992223116116,
      "grad_norm": 0.1873890362573931,
      "learning_rate": 9.207129613838212e-07,
      "loss": 0.0098,
      "step": 9211
    },
    {
      "epoch": 2.470367390721373,
      "grad_norm": 0.24998210174784793,
      "learning_rate": 9.198109769346303e-07,
      "loss": 0.0113,
      "step": 9212
    },
    {
      "epoch": 2.4706355591311344,
      "grad_norm": 0.3439947767713896,
      "learning_rate": 9.189093897626134e-07,
      "loss": 0.013,
      "step": 9213
    },
    {
      "epoch": 2.470903727540896,
      "grad_norm": 0.21664700900986422,
      "learning_rate": 9.180081999555523e-07,
      "loss": 0.0115,
      "step": 9214
    },
    {
      "epoch": 2.471171895950657,
      "grad_norm": 0.23912641517332364,
      "learning_rate": 9.171074076011971e-07,
      "loss": 0.0096,
      "step": 9215
    },
    {
      "epoch": 2.471440064360418,
      "grad_norm": 0.32880356919331805,
      "learning_rate": 9.162070127872525e-07,
      "loss": 0.0236,
      "step": 9216
    },
    {
      "epoch": 2.4717082327701796,
      "grad_norm": 0.34252432156570595,
      "learning_rate": 9.153070156013883e-07,
      "loss": 0.0184,
      "step": 9217
    },
    {
      "epoch": 2.471976401179941,
      "grad_norm": 0.22622283317221914,
      "learning_rate": 9.144074161312372e-07,
      "loss": 0.0136,
      "step": 9218
    },
    {
      "epoch": 2.4722445695897024,
      "grad_norm": 0.24698397580815412,
      "learning_rate": 9.135082144643869e-07,
      "loss": 0.0114,
      "step": 9219
    },
    {
      "epoch": 2.472512737999464,
      "grad_norm": 0.25438089315939016,
      "learning_rate": 9.126094106883943e-07,
      "loss": 0.0118,
      "step": 9220
    },
    {
      "epoch": 2.472780906409225,
      "grad_norm": 0.24273703591156764,
      "learning_rate": 9.117110048907713e-07,
      "loss": 0.0154,
      "step": 9221
    },
    {
      "epoch": 2.473049074818986,
      "grad_norm": 0.327189171879408,
      "learning_rate": 9.108129971589924e-07,
      "loss": 0.0138,
      "step": 9222
    },
    {
      "epoch": 2.4733172432287476,
      "grad_norm": 0.2516938584466438,
      "learning_rate": 9.099153875804966e-07,
      "loss": 0.0167,
      "step": 9223
    },
    {
      "epoch": 2.473585411638509,
      "grad_norm": 0.16446938910255401,
      "learning_rate": 9.090181762426791e-07,
      "loss": 0.0074,
      "step": 9224
    },
    {
      "epoch": 2.4738535800482704,
      "grad_norm": 0.4072454825073082,
      "learning_rate": 9.081213632329022e-07,
      "loss": 0.0163,
      "step": 9225
    },
    {
      "epoch": 2.474121748458032,
      "grad_norm": 0.2856723753127558,
      "learning_rate": 9.072249486384827e-07,
      "loss": 0.0161,
      "step": 9226
    },
    {
      "epoch": 2.4743899168677927,
      "grad_norm": 0.2083062180906677,
      "learning_rate": 9.06328932546704e-07,
      "loss": 0.0092,
      "step": 9227
    },
    {
      "epoch": 2.474658085277554,
      "grad_norm": 0.20889370099236584,
      "learning_rate": 9.054333150448097e-07,
      "loss": 0.0115,
      "step": 9228
    },
    {
      "epoch": 2.4749262536873156,
      "grad_norm": 0.25967862634241773,
      "learning_rate": 9.045380962200018e-07,
      "loss": 0.012,
      "step": 9229
    },
    {
      "epoch": 2.475194422097077,
      "grad_norm": 0.19075519532576385,
      "learning_rate": 9.036432761594472e-07,
      "loss": 0.0129,
      "step": 9230
    },
    {
      "epoch": 2.4754625905068384,
      "grad_norm": 0.22766674300648002,
      "learning_rate": 9.0274885495027e-07,
      "loss": 0.0122,
      "step": 9231
    },
    {
      "epoch": 2.4757307589165998,
      "grad_norm": 0.2341301723430741,
      "learning_rate": 9.018548326795601e-07,
      "loss": 0.0159,
      "step": 9232
    },
    {
      "epoch": 2.475998927326361,
      "grad_norm": 0.23740926127419162,
      "learning_rate": 9.009612094343651e-07,
      "loss": 0.0134,
      "step": 9233
    },
    {
      "epoch": 2.476267095736122,
      "grad_norm": 0.2599515312830257,
      "learning_rate": 9.000679853016925e-07,
      "loss": 0.0158,
      "step": 9234
    },
    {
      "epoch": 2.4765352641458835,
      "grad_norm": 0.32379834884783953,
      "learning_rate": 8.991751603685167e-07,
      "loss": 0.0155,
      "step": 9235
    },
    {
      "epoch": 2.476803432555645,
      "grad_norm": 0.2880043763263693,
      "learning_rate": 8.982827347217665e-07,
      "loss": 0.0145,
      "step": 9236
    },
    {
      "epoch": 2.4770716009654064,
      "grad_norm": 0.3659978534304251,
      "learning_rate": 8.973907084483369e-07,
      "loss": 0.021,
      "step": 9237
    },
    {
      "epoch": 2.4773397693751678,
      "grad_norm": 0.3387240982597907,
      "learning_rate": 8.964990816350827e-07,
      "loss": 0.0193,
      "step": 9238
    },
    {
      "epoch": 2.4776079377849287,
      "grad_norm": 0.3092535010534687,
      "learning_rate": 8.956078543688174e-07,
      "loss": 0.0189,
      "step": 9239
    },
    {
      "epoch": 2.47787610619469,
      "grad_norm": 0.30021676895668187,
      "learning_rate": 8.947170267363197e-07,
      "loss": 0.0141,
      "step": 9240
    },
    {
      "epoch": 2.4781442746044515,
      "grad_norm": 0.24434914794978602,
      "learning_rate": 8.938265988243245e-07,
      "loss": 0.0118,
      "step": 9241
    },
    {
      "epoch": 2.478412443014213,
      "grad_norm": 0.2696306055270258,
      "learning_rate": 8.929365707195325e-07,
      "loss": 0.0149,
      "step": 9242
    },
    {
      "epoch": 2.4786806114239743,
      "grad_norm": 0.29524018147253095,
      "learning_rate": 8.92046942508602e-07,
      "loss": 0.016,
      "step": 9243
    },
    {
      "epoch": 2.4789487798337357,
      "grad_norm": 0.24036427029742488,
      "learning_rate": 8.911577142781536e-07,
      "loss": 0.0108,
      "step": 9244
    },
    {
      "epoch": 2.479216948243497,
      "grad_norm": 0.24997744630048055,
      "learning_rate": 8.902688861147729e-07,
      "loss": 0.0116,
      "step": 9245
    },
    {
      "epoch": 2.479485116653258,
      "grad_norm": 0.2661673124839509,
      "learning_rate": 8.89380458104997e-07,
      "loss": 0.0129,
      "step": 9246
    },
    {
      "epoch": 2.4797532850630195,
      "grad_norm": 0.22783306806154455,
      "learning_rate": 8.884924303353321e-07,
      "loss": 0.0131,
      "step": 9247
    },
    {
      "epoch": 2.480021453472781,
      "grad_norm": 0.25749362446005863,
      "learning_rate": 8.876048028922446e-07,
      "loss": 0.0147,
      "step": 9248
    },
    {
      "epoch": 2.4802896218825423,
      "grad_norm": 0.270717435769717,
      "learning_rate": 8.867175758621582e-07,
      "loss": 0.0132,
      "step": 9249
    },
    {
      "epoch": 2.4805577902923037,
      "grad_norm": 0.2869659336814672,
      "learning_rate": 8.858307493314622e-07,
      "loss": 0.0185,
      "step": 9250
    },
    {
      "epoch": 2.4808259587020647,
      "grad_norm": 0.24640016829378278,
      "learning_rate": 8.849443233865018e-07,
      "loss": 0.0093,
      "step": 9251
    },
    {
      "epoch": 2.481094127111826,
      "grad_norm": 0.22487455978537646,
      "learning_rate": 8.840582981135892e-07,
      "loss": 0.0119,
      "step": 9252
    },
    {
      "epoch": 2.4813622955215875,
      "grad_norm": 0.27228414469981654,
      "learning_rate": 8.831726735989909e-07,
      "loss": 0.0144,
      "step": 9253
    },
    {
      "epoch": 2.481630463931349,
      "grad_norm": 0.21755255587860597,
      "learning_rate": 8.822874499289391e-07,
      "loss": 0.0114,
      "step": 9254
    },
    {
      "epoch": 2.4818986323411103,
      "grad_norm": 0.4066055997903778,
      "learning_rate": 8.814026271896275e-07,
      "loss": 0.016,
      "step": 9255
    },
    {
      "epoch": 2.4821668007508717,
      "grad_norm": 0.24698757050547154,
      "learning_rate": 8.805182054672063e-07,
      "loss": 0.0142,
      "step": 9256
    },
    {
      "epoch": 2.482434969160633,
      "grad_norm": 0.3579199690811656,
      "learning_rate": 8.796341848477913e-07,
      "loss": 0.0218,
      "step": 9257
    },
    {
      "epoch": 2.482703137570394,
      "grad_norm": 0.31044684013343465,
      "learning_rate": 8.787505654174566e-07,
      "loss": 0.0124,
      "step": 9258
    },
    {
      "epoch": 2.4829713059801555,
      "grad_norm": 0.3383822012926626,
      "learning_rate": 8.778673472622362e-07,
      "loss": 0.0097,
      "step": 9259
    },
    {
      "epoch": 2.483239474389917,
      "grad_norm": 0.36223422557075685,
      "learning_rate": 8.769845304681296e-07,
      "loss": 0.0174,
      "step": 9260
    },
    {
      "epoch": 2.4835076427996783,
      "grad_norm": 0.25985621401960435,
      "learning_rate": 8.761021151210918e-07,
      "loss": 0.0145,
      "step": 9261
    },
    {
      "epoch": 2.4837758112094397,
      "grad_norm": 0.28789622563206746,
      "learning_rate": 8.752201013070433e-07,
      "loss": 0.0226,
      "step": 9262
    },
    {
      "epoch": 2.4840439796192006,
      "grad_norm": 0.2135333928111125,
      "learning_rate": 8.743384891118612e-07,
      "loss": 0.0112,
      "step": 9263
    },
    {
      "epoch": 2.484312148028962,
      "grad_norm": 0.2836690329154222,
      "learning_rate": 8.734572786213869e-07,
      "loss": 0.0179,
      "step": 9264
    },
    {
      "epoch": 2.4845803164387235,
      "grad_norm": 0.19701853884599754,
      "learning_rate": 8.725764699214234e-07,
      "loss": 0.0116,
      "step": 9265
    },
    {
      "epoch": 2.484848484848485,
      "grad_norm": 0.21933328660018325,
      "learning_rate": 8.716960630977289e-07,
      "loss": 0.0155,
      "step": 9266
    },
    {
      "epoch": 2.4851166532582463,
      "grad_norm": 0.2479011197687232,
      "learning_rate": 8.708160582360303e-07,
      "loss": 0.0134,
      "step": 9267
    },
    {
      "epoch": 2.4853848216680077,
      "grad_norm": 0.2787807893459746,
      "learning_rate": 8.699364554220074e-07,
      "loss": 0.0169,
      "step": 9268
    },
    {
      "epoch": 2.485652990077769,
      "grad_norm": 0.28692719094539865,
      "learning_rate": 8.690572547413084e-07,
      "loss": 0.0226,
      "step": 9269
    },
    {
      "epoch": 2.48592115848753,
      "grad_norm": 0.22605099456488412,
      "learning_rate": 8.68178456279537e-07,
      "loss": 0.0122,
      "step": 9270
    },
    {
      "epoch": 2.4861893268972914,
      "grad_norm": 0.24256574286868346,
      "learning_rate": 8.673000601222581e-07,
      "loss": 0.0169,
      "step": 9271
    },
    {
      "epoch": 2.486457495307053,
      "grad_norm": 0.1973340432413498,
      "learning_rate": 8.664220663550022e-07,
      "loss": 0.0096,
      "step": 9272
    },
    {
      "epoch": 2.4867256637168142,
      "grad_norm": 0.1851371195999561,
      "learning_rate": 8.65544475063253e-07,
      "loss": 0.009,
      "step": 9273
    },
    {
      "epoch": 2.4869938321265757,
      "grad_norm": 0.31312061246284895,
      "learning_rate": 8.646672863324623e-07,
      "loss": 0.0196,
      "step": 9274
    },
    {
      "epoch": 2.4872620005363366,
      "grad_norm": 0.3134208618134148,
      "learning_rate": 8.637905002480396e-07,
      "loss": 0.0127,
      "step": 9275
    },
    {
      "epoch": 2.487530168946098,
      "grad_norm": 0.24500803620328263,
      "learning_rate": 8.629141168953531e-07,
      "loss": 0.0117,
      "step": 9276
    },
    {
      "epoch": 2.4877983373558594,
      "grad_norm": 0.26815995287821287,
      "learning_rate": 8.620381363597369e-07,
      "loss": 0.0169,
      "step": 9277
    },
    {
      "epoch": 2.488066505765621,
      "grad_norm": 0.21817183086192776,
      "learning_rate": 8.611625587264799e-07,
      "loss": 0.0113,
      "step": 9278
    },
    {
      "epoch": 2.4883346741753822,
      "grad_norm": 0.20983831249169235,
      "learning_rate": 8.602873840808379e-07,
      "loss": 0.0138,
      "step": 9279
    },
    {
      "epoch": 2.4886028425851436,
      "grad_norm": 0.17894457823455856,
      "learning_rate": 8.594126125080204e-07,
      "loss": 0.0118,
      "step": 9280
    },
    {
      "epoch": 2.488871010994905,
      "grad_norm": 0.24750343311940268,
      "learning_rate": 8.585382440932044e-07,
      "loss": 0.011,
      "step": 9281
    },
    {
      "epoch": 2.489139179404666,
      "grad_norm": 0.20045463358672191,
      "learning_rate": 8.576642789215267e-07,
      "loss": 0.013,
      "step": 9282
    },
    {
      "epoch": 2.4894073478144274,
      "grad_norm": 0.28351680342072455,
      "learning_rate": 8.567907170780776e-07,
      "loss": 0.012,
      "step": 9283
    },
    {
      "epoch": 2.489675516224189,
      "grad_norm": 0.5307186204540704,
      "learning_rate": 8.559175586479168e-07,
      "loss": 0.0222,
      "step": 9284
    },
    {
      "epoch": 2.48994368463395,
      "grad_norm": 0.26590827075639856,
      "learning_rate": 8.550448037160619e-07,
      "loss": 0.0165,
      "step": 9285
    },
    {
      "epoch": 2.4902118530437116,
      "grad_norm": 0.1769189717905138,
      "learning_rate": 8.541724523674888e-07,
      "loss": 0.0089,
      "step": 9286
    },
    {
      "epoch": 2.4904800214534726,
      "grad_norm": 0.21917642471272678,
      "learning_rate": 8.533005046871384e-07,
      "loss": 0.0079,
      "step": 9287
    },
    {
      "epoch": 2.490748189863234,
      "grad_norm": 0.20868106367121747,
      "learning_rate": 8.524289607599068e-07,
      "loss": 0.0133,
      "step": 9288
    },
    {
      "epoch": 2.4910163582729954,
      "grad_norm": 0.23126503666288667,
      "learning_rate": 8.515578206706576e-07,
      "loss": 0.0063,
      "step": 9289
    },
    {
      "epoch": 2.491284526682757,
      "grad_norm": 0.2236184936481097,
      "learning_rate": 8.506870845042081e-07,
      "loss": 0.0135,
      "step": 9290
    },
    {
      "epoch": 2.491552695092518,
      "grad_norm": 0.18326375751710228,
      "learning_rate": 8.498167523453404e-07,
      "loss": 0.0115,
      "step": 9291
    },
    {
      "epoch": 2.4918208635022796,
      "grad_norm": 0.23731584688930557,
      "learning_rate": 8.48946824278799e-07,
      "loss": 0.0112,
      "step": 9292
    },
    {
      "epoch": 2.492089031912041,
      "grad_norm": 0.3549126874754087,
      "learning_rate": 8.480773003892828e-07,
      "loss": 0.0171,
      "step": 9293
    },
    {
      "epoch": 2.492357200321802,
      "grad_norm": 0.47497774591491004,
      "learning_rate": 8.47208180761459e-07,
      "loss": 0.0168,
      "step": 9294
    },
    {
      "epoch": 2.4926253687315634,
      "grad_norm": 0.17641447437470914,
      "learning_rate": 8.463394654799472e-07,
      "loss": 0.0075,
      "step": 9295
    },
    {
      "epoch": 2.4928935371413248,
      "grad_norm": 0.22724606909252956,
      "learning_rate": 8.454711546293337e-07,
      "loss": 0.0126,
      "step": 9296
    },
    {
      "epoch": 2.493161705551086,
      "grad_norm": 0.2295144701976218,
      "learning_rate": 8.446032482941652e-07,
      "loss": 0.0121,
      "step": 9297
    },
    {
      "epoch": 2.4934298739608476,
      "grad_norm": 0.2568841451754485,
      "learning_rate": 8.437357465589446e-07,
      "loss": 0.0135,
      "step": 9298
    },
    {
      "epoch": 2.4936980423706085,
      "grad_norm": 0.21031944016170767,
      "learning_rate": 8.428686495081412e-07,
      "loss": 0.0107,
      "step": 9299
    },
    {
      "epoch": 2.49396621078037,
      "grad_norm": 0.24915948410325814,
      "learning_rate": 8.420019572261789e-07,
      "loss": 0.0145,
      "step": 9300
    },
    {
      "epoch": 2.4942343791901314,
      "grad_norm": 0.24648888331787241,
      "learning_rate": 8.411356697974471e-07,
      "loss": 0.0132,
      "step": 9301
    },
    {
      "epoch": 2.4945025475998928,
      "grad_norm": 0.22145390383905986,
      "learning_rate": 8.402697873062942e-07,
      "loss": 0.0111,
      "step": 9302
    },
    {
      "epoch": 2.494770716009654,
      "grad_norm": 0.22356017805573006,
      "learning_rate": 8.394043098370275e-07,
      "loss": 0.0124,
      "step": 9303
    },
    {
      "epoch": 2.4950388844194156,
      "grad_norm": 0.4012839689174112,
      "learning_rate": 8.385392374739181e-07,
      "loss": 0.0164,
      "step": 9304
    },
    {
      "epoch": 2.4953070528291765,
      "grad_norm": 0.2434705140636726,
      "learning_rate": 8.376745703011935e-07,
      "loss": 0.0117,
      "step": 9305
    },
    {
      "epoch": 2.495575221238938,
      "grad_norm": 0.26533543321722064,
      "learning_rate": 8.36810308403046e-07,
      "loss": 0.0106,
      "step": 9306
    },
    {
      "epoch": 2.4958433896486993,
      "grad_norm": 0.23742776434874427,
      "learning_rate": 8.359464518636262e-07,
      "loss": 0.0116,
      "step": 9307
    },
    {
      "epoch": 2.4961115580584607,
      "grad_norm": 0.21951364812744953,
      "learning_rate": 8.350830007670435e-07,
      "loss": 0.0137,
      "step": 9308
    },
    {
      "epoch": 2.496379726468222,
      "grad_norm": 0.2317982391694231,
      "learning_rate": 8.342199551973728e-07,
      "loss": 0.0103,
      "step": 9309
    },
    {
      "epoch": 2.4966478948779836,
      "grad_norm": 0.20102024595668322,
      "learning_rate": 8.333573152386442e-07,
      "loss": 0.0086,
      "step": 9310
    },
    {
      "epoch": 2.4969160632877445,
      "grad_norm": 0.23566939560104486,
      "learning_rate": 8.324950809748528e-07,
      "loss": 0.0098,
      "step": 9311
    },
    {
      "epoch": 2.497184231697506,
      "grad_norm": 0.27893858250160447,
      "learning_rate": 8.31633252489949e-07,
      "loss": 0.0107,
      "step": 9312
    },
    {
      "epoch": 2.4974524001072673,
      "grad_norm": 0.36002322684128896,
      "learning_rate": 8.307718298678491e-07,
      "loss": 0.0172,
      "step": 9313
    },
    {
      "epoch": 2.4977205685170287,
      "grad_norm": 0.2926912626210707,
      "learning_rate": 8.299108131924282e-07,
      "loss": 0.014,
      "step": 9314
    },
    {
      "epoch": 2.49798873692679,
      "grad_norm": 0.26282003915860813,
      "learning_rate": 8.290502025475183e-07,
      "loss": 0.0126,
      "step": 9315
    },
    {
      "epoch": 2.4982569053365515,
      "grad_norm": 0.29328890442854455,
      "learning_rate": 8.281899980169183e-07,
      "loss": 0.0124,
      "step": 9316
    },
    {
      "epoch": 2.4985250737463125,
      "grad_norm": 0.2571295982446838,
      "learning_rate": 8.273301996843813e-07,
      "loss": 0.0112,
      "step": 9317
    },
    {
      "epoch": 2.498793242156074,
      "grad_norm": 0.2600465160491636,
      "learning_rate": 8.264708076336254e-07,
      "loss": 0.0162,
      "step": 9318
    },
    {
      "epoch": 2.4990614105658353,
      "grad_norm": 0.2609114514398319,
      "learning_rate": 8.256118219483261e-07,
      "loss": 0.0152,
      "step": 9319
    },
    {
      "epoch": 2.4993295789755967,
      "grad_norm": 0.26227824765205804,
      "learning_rate": 8.247532427121202e-07,
      "loss": 0.0132,
      "step": 9320
    },
    {
      "epoch": 2.499597747385358,
      "grad_norm": 0.190664796141924,
      "learning_rate": 8.238950700086068e-07,
      "loss": 0.0087,
      "step": 9321
    },
    {
      "epoch": 2.4998659157951195,
      "grad_norm": 0.20752542479626115,
      "learning_rate": 8.230373039213424e-07,
      "loss": 0.0135,
      "step": 9322
    },
    {
      "epoch": 2.5001340842048805,
      "grad_norm": 0.2609615414901576,
      "learning_rate": 8.221799445338457e-07,
      "loss": 0.0148,
      "step": 9323
    },
    {
      "epoch": 2.500402252614642,
      "grad_norm": 0.31585381704870674,
      "learning_rate": 8.213229919295973e-07,
      "loss": 0.0151,
      "step": 9324
    },
    {
      "epoch": 2.5006704210244033,
      "grad_norm": 0.19384376235332715,
      "learning_rate": 8.204664461920337e-07,
      "loss": 0.0098,
      "step": 9325
    },
    {
      "epoch": 2.5009385894341647,
      "grad_norm": 0.3019653416114766,
      "learning_rate": 8.19610307404557e-07,
      "loss": 0.018,
      "step": 9326
    },
    {
      "epoch": 2.501206757843926,
      "grad_norm": 0.24584219875330476,
      "learning_rate": 8.187545756505244e-07,
      "loss": 0.0125,
      "step": 9327
    },
    {
      "epoch": 2.501474926253687,
      "grad_norm": 0.32739421611320624,
      "learning_rate": 8.178992510132594e-07,
      "loss": 0.0211,
      "step": 9328
    },
    {
      "epoch": 2.501743094663449,
      "grad_norm": 0.3184702794939725,
      "learning_rate": 8.170443335760397e-07,
      "loss": 0.0147,
      "step": 9329
    },
    {
      "epoch": 2.50201126307321,
      "grad_norm": 0.3402417217515435,
      "learning_rate": 8.161898234221094e-07,
      "loss": 0.016,
      "step": 9330
    },
    {
      "epoch": 2.5022794314829713,
      "grad_norm": 0.2031215416719073,
      "learning_rate": 8.153357206346674e-07,
      "loss": 0.0113,
      "step": 9331
    },
    {
      "epoch": 2.5025475998927327,
      "grad_norm": 0.25098818373195,
      "learning_rate": 8.144820252968755e-07,
      "loss": 0.0116,
      "step": 9332
    },
    {
      "epoch": 2.502815768302494,
      "grad_norm": 0.31156939742557493,
      "learning_rate": 8.136287374918561e-07,
      "loss": 0.0133,
      "step": 9333
    },
    {
      "epoch": 2.5030839367122555,
      "grad_norm": 0.2578483384234698,
      "learning_rate": 8.12775857302694e-07,
      "loss": 0.0112,
      "step": 9334
    },
    {
      "epoch": 2.5033521051220164,
      "grad_norm": 0.3211914835844652,
      "learning_rate": 8.119233848124274e-07,
      "loss": 0.0116,
      "step": 9335
    },
    {
      "epoch": 2.503620273531778,
      "grad_norm": 0.32893121473117226,
      "learning_rate": 8.110713201040637e-07,
      "loss": 0.0223,
      "step": 9336
    },
    {
      "epoch": 2.5038884419415393,
      "grad_norm": 0.23734762545189533,
      "learning_rate": 8.10219663260563e-07,
      "loss": 0.0146,
      "step": 9337
    },
    {
      "epoch": 2.5041566103513007,
      "grad_norm": 0.2653282914067344,
      "learning_rate": 8.093684143648506e-07,
      "loss": 0.0138,
      "step": 9338
    },
    {
      "epoch": 2.504424778761062,
      "grad_norm": 0.20349547032176826,
      "learning_rate": 8.085175734998091e-07,
      "loss": 0.0094,
      "step": 9339
    },
    {
      "epoch": 2.504692947170823,
      "grad_norm": 0.17062330220112448,
      "learning_rate": 8.076671407482833e-07,
      "loss": 0.0082,
      "step": 9340
    },
    {
      "epoch": 2.504961115580585,
      "grad_norm": 0.2553387714102383,
      "learning_rate": 8.068171161930788e-07,
      "loss": 0.0134,
      "step": 9341
    },
    {
      "epoch": 2.505229283990346,
      "grad_norm": 0.23201569039276287,
      "learning_rate": 8.059674999169586e-07,
      "loss": 0.0146,
      "step": 9342
    },
    {
      "epoch": 2.5054974524001072,
      "grad_norm": 0.3484873723411901,
      "learning_rate": 8.051182920026474e-07,
      "loss": 0.0122,
      "step": 9343
    },
    {
      "epoch": 2.5057656208098686,
      "grad_norm": 0.23480805073852964,
      "learning_rate": 8.042694925328315e-07,
      "loss": 0.0112,
      "step": 9344
    },
    {
      "epoch": 2.50603378921963,
      "grad_norm": 0.2893715867129857,
      "learning_rate": 8.03421101590155e-07,
      "loss": 0.0152,
      "step": 9345
    },
    {
      "epoch": 2.5063019576293915,
      "grad_norm": 0.2644184628217205,
      "learning_rate": 8.02573119257225e-07,
      "loss": 0.0178,
      "step": 9346
    },
    {
      "epoch": 2.5065701260391524,
      "grad_norm": 0.258811159761651,
      "learning_rate": 8.01725545616605e-07,
      "loss": 0.0136,
      "step": 9347
    },
    {
      "epoch": 2.506838294448914,
      "grad_norm": 0.2776946895700495,
      "learning_rate": 8.008783807508241e-07,
      "loss": 0.0117,
      "step": 9348
    },
    {
      "epoch": 2.5071064628586752,
      "grad_norm": 0.19193524629280395,
      "learning_rate": 8.000316247423656e-07,
      "loss": 0.0126,
      "step": 9349
    },
    {
      "epoch": 2.5073746312684366,
      "grad_norm": 0.21740318014120313,
      "learning_rate": 7.991852776736769e-07,
      "loss": 0.0122,
      "step": 9350
    },
    {
      "epoch": 2.507642799678198,
      "grad_norm": 0.25071331519248574,
      "learning_rate": 7.98339339627166e-07,
      "loss": 0.0182,
      "step": 9351
    },
    {
      "epoch": 2.507910968087959,
      "grad_norm": 0.25425966438531805,
      "learning_rate": 7.97493810685197e-07,
      "loss": 0.0167,
      "step": 9352
    },
    {
      "epoch": 2.508179136497721,
      "grad_norm": 0.26998562688679706,
      "learning_rate": 7.966486909300997e-07,
      "loss": 0.0135,
      "step": 9353
    },
    {
      "epoch": 2.508447304907482,
      "grad_norm": 0.20898533458711333,
      "learning_rate": 7.958039804441597e-07,
      "loss": 0.0125,
      "step": 9354
    },
    {
      "epoch": 2.508715473317243,
      "grad_norm": 0.33041257815675745,
      "learning_rate": 7.949596793096226e-07,
      "loss": 0.0252,
      "step": 9355
    },
    {
      "epoch": 2.5089836417270046,
      "grad_norm": 0.28926202113459537,
      "learning_rate": 7.941157876086991e-07,
      "loss": 0.0161,
      "step": 9356
    },
    {
      "epoch": 2.509251810136766,
      "grad_norm": 0.23697618016860766,
      "learning_rate": 7.932723054235536e-07,
      "loss": 0.014,
      "step": 9357
    },
    {
      "epoch": 2.5095199785465274,
      "grad_norm": 0.3353980650390598,
      "learning_rate": 7.924292328363159e-07,
      "loss": 0.0167,
      "step": 9358
    },
    {
      "epoch": 2.5097881469562884,
      "grad_norm": 0.2148554768827169,
      "learning_rate": 7.915865699290721e-07,
      "loss": 0.0106,
      "step": 9359
    },
    {
      "epoch": 2.51005631536605,
      "grad_norm": 0.3206924056062531,
      "learning_rate": 7.907443167838708e-07,
      "loss": 0.0154,
      "step": 9360
    },
    {
      "epoch": 2.510324483775811,
      "grad_norm": 0.25481679779760774,
      "learning_rate": 7.89902473482721e-07,
      "loss": 0.0126,
      "step": 9361
    },
    {
      "epoch": 2.5105926521855726,
      "grad_norm": 0.25797670236067866,
      "learning_rate": 7.890610401075888e-07,
      "loss": 0.014,
      "step": 9362
    },
    {
      "epoch": 2.510860820595334,
      "grad_norm": 0.27637027720668916,
      "learning_rate": 7.882200167404047e-07,
      "loss": 0.0125,
      "step": 9363
    },
    {
      "epoch": 2.511128989005095,
      "grad_norm": 0.2440645985470306,
      "learning_rate": 7.873794034630539e-07,
      "loss": 0.0143,
      "step": 9364
    },
    {
      "epoch": 2.511397157414857,
      "grad_norm": 0.23538576769010774,
      "learning_rate": 7.865392003573874e-07,
      "loss": 0.0119,
      "step": 9365
    },
    {
      "epoch": 2.5116653258246178,
      "grad_norm": 0.27382243909041265,
      "learning_rate": 7.856994075052127e-07,
      "loss": 0.0188,
      "step": 9366
    },
    {
      "epoch": 2.511933494234379,
      "grad_norm": 0.28099861951984345,
      "learning_rate": 7.84860024988297e-07,
      "loss": 0.0116,
      "step": 9367
    },
    {
      "epoch": 2.5122016626441406,
      "grad_norm": 0.2560651309373312,
      "learning_rate": 7.840210528883708e-07,
      "loss": 0.0131,
      "step": 9368
    },
    {
      "epoch": 2.512469831053902,
      "grad_norm": 0.27970997941371256,
      "learning_rate": 7.8318249128712e-07,
      "loss": 0.0167,
      "step": 9369
    },
    {
      "epoch": 2.5127379994636634,
      "grad_norm": 0.31692455915245166,
      "learning_rate": 7.823443402661945e-07,
      "loss": 0.0165,
      "step": 9370
    },
    {
      "epoch": 2.5130061678734243,
      "grad_norm": 0.27795471323836,
      "learning_rate": 7.815065999072041e-07,
      "loss": 0.0132,
      "step": 9371
    },
    {
      "epoch": 2.5132743362831858,
      "grad_norm": 0.284347118224109,
      "learning_rate": 7.806692702917151e-07,
      "loss": 0.015,
      "step": 9372
    },
    {
      "epoch": 2.513542504692947,
      "grad_norm": 0.2949177890169353,
      "learning_rate": 7.798323515012585e-07,
      "loss": 0.0154,
      "step": 9373
    },
    {
      "epoch": 2.5138106731027086,
      "grad_norm": 0.23814751524051783,
      "learning_rate": 7.789958436173206e-07,
      "loss": 0.0129,
      "step": 9374
    },
    {
      "epoch": 2.51407884151247,
      "grad_norm": 0.20623561549637223,
      "learning_rate": 7.781597467213514e-07,
      "loss": 0.0101,
      "step": 9375
    },
    {
      "epoch": 2.514347009922231,
      "grad_norm": 0.3604045484507002,
      "learning_rate": 7.773240608947579e-07,
      "loss": 0.0196,
      "step": 9376
    },
    {
      "epoch": 2.5146151783319923,
      "grad_norm": 0.1923246582803911,
      "learning_rate": 7.764887862189097e-07,
      "loss": 0.0131,
      "step": 9377
    },
    {
      "epoch": 2.5148833467417537,
      "grad_norm": 0.24649372823669893,
      "learning_rate": 7.756539227751375e-07,
      "loss": 0.0101,
      "step": 9378
    },
    {
      "epoch": 2.515151515151515,
      "grad_norm": 0.26552433923361396,
      "learning_rate": 7.748194706447254e-07,
      "loss": 0.0128,
      "step": 9379
    },
    {
      "epoch": 2.5154196835612765,
      "grad_norm": 0.2718384037056347,
      "learning_rate": 7.739854299089233e-07,
      "loss": 0.0145,
      "step": 9380
    },
    {
      "epoch": 2.515687851971038,
      "grad_norm": 0.21685449209594948,
      "learning_rate": 7.731518006489414e-07,
      "loss": 0.012,
      "step": 9381
    },
    {
      "epoch": 2.5159560203807994,
      "grad_norm": 0.24099970576014063,
      "learning_rate": 7.723185829459456e-07,
      "loss": 0.0091,
      "step": 9382
    },
    {
      "epoch": 2.5162241887905603,
      "grad_norm": 0.4328528253769127,
      "learning_rate": 7.714857768810663e-07,
      "loss": 0.0169,
      "step": 9383
    },
    {
      "epoch": 2.5164923572003217,
      "grad_norm": 0.3075128121775694,
      "learning_rate": 7.706533825353891e-07,
      "loss": 0.0178,
      "step": 9384
    },
    {
      "epoch": 2.516760525610083,
      "grad_norm": 0.1879235576849328,
      "learning_rate": 7.698213999899645e-07,
      "loss": 0.0087,
      "step": 9385
    },
    {
      "epoch": 2.5170286940198445,
      "grad_norm": 0.31382938300899316,
      "learning_rate": 7.68989829325798e-07,
      "loss": 0.0237,
      "step": 9386
    },
    {
      "epoch": 2.517296862429606,
      "grad_norm": 0.2753799636498773,
      "learning_rate": 7.681586706238586e-07,
      "loss": 0.0139,
      "step": 9387
    },
    {
      "epoch": 2.517565030839367,
      "grad_norm": 0.2468139398245865,
      "learning_rate": 7.673279239650755e-07,
      "loss": 0.0145,
      "step": 9388
    },
    {
      "epoch": 2.5178331992491283,
      "grad_norm": 0.2799748470692375,
      "learning_rate": 7.66497589430334e-07,
      "loss": 0.0147,
      "step": 9389
    },
    {
      "epoch": 2.5181013676588897,
      "grad_norm": 0.22264963965353896,
      "learning_rate": 7.656676671004832e-07,
      "loss": 0.0137,
      "step": 9390
    },
    {
      "epoch": 2.518369536068651,
      "grad_norm": 0.26735823741414494,
      "learning_rate": 7.648381570563301e-07,
      "loss": 0.0149,
      "step": 9391
    },
    {
      "epoch": 2.5186377044784125,
      "grad_norm": 0.26516554334318954,
      "learning_rate": 7.640090593786398e-07,
      "loss": 0.0084,
      "step": 9392
    },
    {
      "epoch": 2.518905872888174,
      "grad_norm": 0.23098576365259924,
      "learning_rate": 7.631803741481425e-07,
      "loss": 0.0126,
      "step": 9393
    },
    {
      "epoch": 2.5191740412979353,
      "grad_norm": 0.28602044558464945,
      "learning_rate": 7.623521014455226e-07,
      "loss": 0.0198,
      "step": 9394
    },
    {
      "epoch": 2.5194422097076963,
      "grad_norm": 0.26324239958421525,
      "learning_rate": 7.61524241351429e-07,
      "loss": 0.0139,
      "step": 9395
    },
    {
      "epoch": 2.5197103781174577,
      "grad_norm": 0.23311793363772013,
      "learning_rate": 7.60696793946466e-07,
      "loss": 0.0133,
      "step": 9396
    },
    {
      "epoch": 2.519978546527219,
      "grad_norm": 0.21656357026413522,
      "learning_rate": 7.598697593112009e-07,
      "loss": 0.0113,
      "step": 9397
    },
    {
      "epoch": 2.5202467149369805,
      "grad_norm": 0.2075177922573687,
      "learning_rate": 7.590431375261615e-07,
      "loss": 0.01,
      "step": 9398
    },
    {
      "epoch": 2.520514883346742,
      "grad_norm": 0.21962244917120785,
      "learning_rate": 7.582169286718305e-07,
      "loss": 0.0119,
      "step": 9399
    },
    {
      "epoch": 2.520783051756503,
      "grad_norm": 0.28319058939515795,
      "learning_rate": 7.573911328286571e-07,
      "loss": 0.0231,
      "step": 9400
    },
    {
      "epoch": 2.5210512201662643,
      "grad_norm": 0.30751509893707457,
      "learning_rate": 7.565657500770435e-07,
      "loss": 0.0134,
      "step": 9401
    },
    {
      "epoch": 2.5213193885760257,
      "grad_norm": 0.1973453962088212,
      "learning_rate": 7.557407804973582e-07,
      "loss": 0.01,
      "step": 9402
    },
    {
      "epoch": 2.521587556985787,
      "grad_norm": 0.15641184525651589,
      "learning_rate": 7.549162241699248e-07,
      "loss": 0.0065,
      "step": 9403
    },
    {
      "epoch": 2.5218557253955485,
      "grad_norm": 0.16853363977514846,
      "learning_rate": 7.540920811750268e-07,
      "loss": 0.0088,
      "step": 9404
    },
    {
      "epoch": 2.52212389380531,
      "grad_norm": 0.31081775012674273,
      "learning_rate": 7.532683515929118e-07,
      "loss": 0.0179,
      "step": 9405
    },
    {
      "epoch": 2.5223920622150713,
      "grad_norm": 0.24935578779087167,
      "learning_rate": 7.52445035503781e-07,
      "loss": 0.0136,
      "step": 9406
    },
    {
      "epoch": 2.5226602306248322,
      "grad_norm": 0.24258004924281448,
      "learning_rate": 7.516221329878015e-07,
      "loss": 0.01,
      "step": 9407
    },
    {
      "epoch": 2.5229283990345936,
      "grad_norm": 0.2616340551514378,
      "learning_rate": 7.507996441250942e-07,
      "loss": 0.0163,
      "step": 9408
    },
    {
      "epoch": 2.523196567444355,
      "grad_norm": 0.4975836370875494,
      "learning_rate": 7.499775689957445e-07,
      "loss": 0.0231,
      "step": 9409
    },
    {
      "epoch": 2.5234647358541165,
      "grad_norm": 0.2680221311585629,
      "learning_rate": 7.491559076797967e-07,
      "loss": 0.0142,
      "step": 9410
    },
    {
      "epoch": 2.523732904263878,
      "grad_norm": 0.250994177820671,
      "learning_rate": 7.483346602572505e-07,
      "loss": 0.0125,
      "step": 9411
    },
    {
      "epoch": 2.524001072673639,
      "grad_norm": 0.22794402855872178,
      "learning_rate": 7.475138268080723e-07,
      "loss": 0.0118,
      "step": 9412
    },
    {
      "epoch": 2.5242692410834002,
      "grad_norm": 0.2976494447402904,
      "learning_rate": 7.466934074121806e-07,
      "loss": 0.017,
      "step": 9413
    },
    {
      "epoch": 2.5245374094931616,
      "grad_norm": 0.22547765459773292,
      "learning_rate": 7.458734021494601e-07,
      "loss": 0.0162,
      "step": 9414
    },
    {
      "epoch": 2.524805577902923,
      "grad_norm": 0.2624689756441355,
      "learning_rate": 7.450538110997546e-07,
      "loss": 0.0155,
      "step": 9415
    },
    {
      "epoch": 2.5250737463126844,
      "grad_norm": 0.24479507072800955,
      "learning_rate": 7.442346343428603e-07,
      "loss": 0.014,
      "step": 9416
    },
    {
      "epoch": 2.525341914722446,
      "grad_norm": 0.2583250532445407,
      "learning_rate": 7.434158719585416e-07,
      "loss": 0.0128,
      "step": 9417
    },
    {
      "epoch": 2.5256100831322073,
      "grad_norm": 0.3927372753334839,
      "learning_rate": 7.425975240265176e-07,
      "loss": 0.0144,
      "step": 9418
    },
    {
      "epoch": 2.525878251541968,
      "grad_norm": 0.27421540196886646,
      "learning_rate": 7.417795906264696e-07,
      "loss": 0.0117,
      "step": 9419
    },
    {
      "epoch": 2.5261464199517296,
      "grad_norm": 0.26677716246507216,
      "learning_rate": 7.409620718380383e-07,
      "loss": 0.0145,
      "step": 9420
    },
    {
      "epoch": 2.526414588361491,
      "grad_norm": 0.2321151408605178,
      "learning_rate": 7.401449677408218e-07,
      "loss": 0.0132,
      "step": 9421
    },
    {
      "epoch": 2.5266827567712524,
      "grad_norm": 0.25042192965746063,
      "learning_rate": 7.393282784143807e-07,
      "loss": 0.011,
      "step": 9422
    },
    {
      "epoch": 2.526950925181014,
      "grad_norm": 0.22564207907228678,
      "learning_rate": 7.385120039382326e-07,
      "loss": 0.014,
      "step": 9423
    },
    {
      "epoch": 2.527219093590775,
      "grad_norm": 0.1910721005379831,
      "learning_rate": 7.376961443918579e-07,
      "loss": 0.0111,
      "step": 9424
    },
    {
      "epoch": 2.527487262000536,
      "grad_norm": 0.2335135646519822,
      "learning_rate": 7.368806998546918e-07,
      "loss": 0.0142,
      "step": 9425
    },
    {
      "epoch": 2.5277554304102976,
      "grad_norm": 0.21520466697185064,
      "learning_rate": 7.360656704061337e-07,
      "loss": 0.0117,
      "step": 9426
    },
    {
      "epoch": 2.528023598820059,
      "grad_norm": 0.23402909635821295,
      "learning_rate": 7.352510561255433e-07,
      "loss": 0.0116,
      "step": 9427
    },
    {
      "epoch": 2.5282917672298204,
      "grad_norm": 0.3832516148622252,
      "learning_rate": 7.344368570922322e-07,
      "loss": 0.0201,
      "step": 9428
    },
    {
      "epoch": 2.528559935639582,
      "grad_norm": 0.21102974373105857,
      "learning_rate": 7.336230733854799e-07,
      "loss": 0.0085,
      "step": 9429
    },
    {
      "epoch": 2.528828104049343,
      "grad_norm": 0.2529114956554202,
      "learning_rate": 7.328097050845228e-07,
      "loss": 0.0088,
      "step": 9430
    },
    {
      "epoch": 2.529096272459104,
      "grad_norm": 0.27659341176767804,
      "learning_rate": 7.319967522685545e-07,
      "loss": 0.0175,
      "step": 9431
    },
    {
      "epoch": 2.5293644408688656,
      "grad_norm": 0.23515939399462932,
      "learning_rate": 7.31184215016732e-07,
      "loss": 0.0198,
      "step": 9432
    },
    {
      "epoch": 2.529632609278627,
      "grad_norm": 0.23230097262032512,
      "learning_rate": 7.303720934081681e-07,
      "loss": 0.009,
      "step": 9433
    },
    {
      "epoch": 2.5299007776883884,
      "grad_norm": 0.26441923470861695,
      "learning_rate": 7.295603875219392e-07,
      "loss": 0.0149,
      "step": 9434
    },
    {
      "epoch": 2.53016894609815,
      "grad_norm": 0.26770246842684875,
      "learning_rate": 7.287490974370759e-07,
      "loss": 0.0133,
      "step": 9435
    },
    {
      "epoch": 2.5304371145079108,
      "grad_norm": 0.33799200115558714,
      "learning_rate": 7.279382232325732e-07,
      "loss": 0.0202,
      "step": 9436
    },
    {
      "epoch": 2.530705282917672,
      "grad_norm": 0.23124341573191828,
      "learning_rate": 7.271277649873854e-07,
      "loss": 0.0131,
      "step": 9437
    },
    {
      "epoch": 2.5309734513274336,
      "grad_norm": 0.2536148812618344,
      "learning_rate": 7.263177227804207e-07,
      "loss": 0.0119,
      "step": 9438
    },
    {
      "epoch": 2.531241619737195,
      "grad_norm": 0.2470937402080855,
      "learning_rate": 7.255080966905547e-07,
      "loss": 0.0111,
      "step": 9439
    },
    {
      "epoch": 2.5315097881469564,
      "grad_norm": 0.37213692936665455,
      "learning_rate": 7.246988867966165e-07,
      "loss": 0.0169,
      "step": 9440
    },
    {
      "epoch": 2.531777956556718,
      "grad_norm": 0.24438838159388515,
      "learning_rate": 7.238900931773963e-07,
      "loss": 0.0188,
      "step": 9441
    },
    {
      "epoch": 2.532046124966479,
      "grad_norm": 0.31372434197498117,
      "learning_rate": 7.230817159116454e-07,
      "loss": 0.0185,
      "step": 9442
    },
    {
      "epoch": 2.53231429337624,
      "grad_norm": 0.2341786170493035,
      "learning_rate": 7.222737550780717e-07,
      "loss": 0.0129,
      "step": 9443
    },
    {
      "epoch": 2.5325824617860015,
      "grad_norm": 0.21083641700181532,
      "learning_rate": 7.214662107553472e-07,
      "loss": 0.0134,
      "step": 9444
    },
    {
      "epoch": 2.532850630195763,
      "grad_norm": 0.21339680792678886,
      "learning_rate": 7.206590830220966e-07,
      "loss": 0.0105,
      "step": 9445
    },
    {
      "epoch": 2.5331187986055244,
      "grad_norm": 0.22705720005814198,
      "learning_rate": 7.198523719569101e-07,
      "loss": 0.0114,
      "step": 9446
    },
    {
      "epoch": 2.5333869670152858,
      "grad_norm": 0.23861263153662646,
      "learning_rate": 7.190460776383351e-07,
      "loss": 0.0096,
      "step": 9447
    },
    {
      "epoch": 2.5336551354250467,
      "grad_norm": 0.2606539697455914,
      "learning_rate": 7.182402001448774e-07,
      "loss": 0.0118,
      "step": 9448
    },
    {
      "epoch": 2.533923303834808,
      "grad_norm": 0.24296670682648355,
      "learning_rate": 7.174347395550041e-07,
      "loss": 0.019,
      "step": 9449
    },
    {
      "epoch": 2.5341914722445695,
      "grad_norm": 0.19807502987992218,
      "learning_rate": 7.166296959471397e-07,
      "loss": 0.0131,
      "step": 9450
    },
    {
      "epoch": 2.534459640654331,
      "grad_norm": 0.3362818930677023,
      "learning_rate": 7.158250693996705e-07,
      "loss": 0.0252,
      "step": 9451
    },
    {
      "epoch": 2.5347278090640923,
      "grad_norm": 0.2311696274986188,
      "learning_rate": 7.150208599909397e-07,
      "loss": 0.0138,
      "step": 9452
    },
    {
      "epoch": 2.5349959774738537,
      "grad_norm": 0.17533233544582852,
      "learning_rate": 7.142170677992505e-07,
      "loss": 0.0078,
      "step": 9453
    },
    {
      "epoch": 2.535264145883615,
      "grad_norm": 0.2572256906544112,
      "learning_rate": 7.134136929028679e-07,
      "loss": 0.0174,
      "step": 9454
    },
    {
      "epoch": 2.535532314293376,
      "grad_norm": 0.2885086991592223,
      "learning_rate": 7.126107353800127e-07,
      "loss": 0.0116,
      "step": 9455
    },
    {
      "epoch": 2.5358004827031375,
      "grad_norm": 0.24484515519660538,
      "learning_rate": 7.118081953088668e-07,
      "loss": 0.0173,
      "step": 9456
    },
    {
      "epoch": 2.536068651112899,
      "grad_norm": 0.18998392629644956,
      "learning_rate": 7.110060727675733e-07,
      "loss": 0.013,
      "step": 9457
    },
    {
      "epoch": 2.5363368195226603,
      "grad_norm": 0.36590432450628585,
      "learning_rate": 7.102043678342307e-07,
      "loss": 0.0124,
      "step": 9458
    },
    {
      "epoch": 2.5366049879324217,
      "grad_norm": 0.28302041295818803,
      "learning_rate": 7.094030805869001e-07,
      "loss": 0.011,
      "step": 9459
    },
    {
      "epoch": 2.5368731563421827,
      "grad_norm": 0.35273529280911353,
      "learning_rate": 7.086022111035995e-07,
      "loss": 0.0317,
      "step": 9460
    },
    {
      "epoch": 2.537141324751944,
      "grad_norm": 0.20147569637295443,
      "learning_rate": 7.078017594623094e-07,
      "loss": 0.01,
      "step": 9461
    },
    {
      "epoch": 2.5374094931617055,
      "grad_norm": 0.20480887869424974,
      "learning_rate": 7.07001725740965e-07,
      "loss": 0.0106,
      "step": 9462
    },
    {
      "epoch": 2.537677661571467,
      "grad_norm": 0.22712988068062642,
      "learning_rate": 7.06202110017466e-07,
      "loss": 0.0102,
      "step": 9463
    },
    {
      "epoch": 2.5379458299812283,
      "grad_norm": 0.34864594075019645,
      "learning_rate": 7.054029123696682e-07,
      "loss": 0.0147,
      "step": 9464
    },
    {
      "epoch": 2.5382139983909893,
      "grad_norm": 0.417098312524783,
      "learning_rate": 7.04604132875385e-07,
      "loss": 0.012,
      "step": 9465
    },
    {
      "epoch": 2.538482166800751,
      "grad_norm": 0.18970005179386235,
      "learning_rate": 7.038057716123941e-07,
      "loss": 0.0096,
      "step": 9466
    },
    {
      "epoch": 2.538750335210512,
      "grad_norm": 0.17280985877681193,
      "learning_rate": 7.030078286584296e-07,
      "loss": 0.0076,
      "step": 9467
    },
    {
      "epoch": 2.5390185036202735,
      "grad_norm": 0.29175975028967266,
      "learning_rate": 7.022103040911838e-07,
      "loss": 0.0195,
      "step": 9468
    },
    {
      "epoch": 2.539286672030035,
      "grad_norm": 0.2586930911462124,
      "learning_rate": 7.014131979883109e-07,
      "loss": 0.0161,
      "step": 9469
    },
    {
      "epoch": 2.5395548404397963,
      "grad_norm": 0.19148936002369674,
      "learning_rate": 7.006165104274215e-07,
      "loss": 0.0102,
      "step": 9470
    },
    {
      "epoch": 2.5398230088495577,
      "grad_norm": 0.26466346090085807,
      "learning_rate": 6.998202414860894e-07,
      "loss": 0.022,
      "step": 9471
    },
    {
      "epoch": 2.5400911772593187,
      "grad_norm": 0.20814060088270436,
      "learning_rate": 6.990243912418421e-07,
      "loss": 0.0109,
      "step": 9472
    },
    {
      "epoch": 2.54035934566908,
      "grad_norm": 0.24725164465675215,
      "learning_rate": 6.982289597721709e-07,
      "loss": 0.0118,
      "step": 9473
    },
    {
      "epoch": 2.5406275140788415,
      "grad_norm": 0.20601353897899788,
      "learning_rate": 6.97433947154526e-07,
      "loss": 0.0089,
      "step": 9474
    },
    {
      "epoch": 2.540895682488603,
      "grad_norm": 0.2312719764486107,
      "learning_rate": 6.966393534663146e-07,
      "loss": 0.0142,
      "step": 9475
    },
    {
      "epoch": 2.5411638508983643,
      "grad_norm": 0.21179140604480123,
      "learning_rate": 6.958451787849046e-07,
      "loss": 0.0097,
      "step": 9476
    },
    {
      "epoch": 2.5414320193081252,
      "grad_norm": 0.22056481497299052,
      "learning_rate": 6.950514231876204e-07,
      "loss": 0.0113,
      "step": 9477
    },
    {
      "epoch": 2.541700187717887,
      "grad_norm": 0.21054747415008213,
      "learning_rate": 6.942580867517501e-07,
      "loss": 0.0123,
      "step": 9478
    },
    {
      "epoch": 2.541968356127648,
      "grad_norm": 0.2650449733202439,
      "learning_rate": 6.934651695545396e-07,
      "loss": 0.016,
      "step": 9479
    },
    {
      "epoch": 2.5422365245374094,
      "grad_norm": 0.23960924225287497,
      "learning_rate": 6.926726716731908e-07,
      "loss": 0.0104,
      "step": 9480
    },
    {
      "epoch": 2.542504692947171,
      "grad_norm": 0.2529858288042599,
      "learning_rate": 6.918805931848693e-07,
      "loss": 0.0155,
      "step": 9481
    },
    {
      "epoch": 2.5427728613569323,
      "grad_norm": 0.22624243306150238,
      "learning_rate": 6.910889341666955e-07,
      "loss": 0.0138,
      "step": 9482
    },
    {
      "epoch": 2.5430410297666937,
      "grad_norm": 0.2525850824082361,
      "learning_rate": 6.902976946957518e-07,
      "loss": 0.0161,
      "step": 9483
    },
    {
      "epoch": 2.5433091981764546,
      "grad_norm": 0.30318854664762496,
      "learning_rate": 6.895068748490807e-07,
      "loss": 0.0136,
      "step": 9484
    },
    {
      "epoch": 2.543577366586216,
      "grad_norm": 0.3446667585344443,
      "learning_rate": 6.887164747036807e-07,
      "loss": 0.0203,
      "step": 9485
    },
    {
      "epoch": 2.5438455349959774,
      "grad_norm": 0.21134220208256024,
      "learning_rate": 6.879264943365116e-07,
      "loss": 0.0137,
      "step": 9486
    },
    {
      "epoch": 2.544113703405739,
      "grad_norm": 0.24559824053085458,
      "learning_rate": 6.871369338244921e-07,
      "loss": 0.0161,
      "step": 9487
    },
    {
      "epoch": 2.5443818718155002,
      "grad_norm": 0.26401556026005163,
      "learning_rate": 6.863477932444973e-07,
      "loss": 0.012,
      "step": 9488
    },
    {
      "epoch": 2.544650040225261,
      "grad_norm": 0.2725076540183263,
      "learning_rate": 6.855590726733669e-07,
      "loss": 0.0104,
      "step": 9489
    },
    {
      "epoch": 2.544918208635023,
      "grad_norm": 0.2607824114444577,
      "learning_rate": 6.847707721878932e-07,
      "loss": 0.014,
      "step": 9490
    },
    {
      "epoch": 2.545186377044784,
      "grad_norm": 0.2516989925210422,
      "learning_rate": 6.839828918648344e-07,
      "loss": 0.0162,
      "step": 9491
    },
    {
      "epoch": 2.5454545454545454,
      "grad_norm": 0.19844603010961745,
      "learning_rate": 6.831954317809014e-07,
      "loss": 0.0105,
      "step": 9492
    },
    {
      "epoch": 2.545722713864307,
      "grad_norm": 0.20831928752433285,
      "learning_rate": 6.824083920127684e-07,
      "loss": 0.0124,
      "step": 9493
    },
    {
      "epoch": 2.5459908822740682,
      "grad_norm": 0.22225699525235018,
      "learning_rate": 6.816217726370677e-07,
      "loss": 0.0182,
      "step": 9494
    },
    {
      "epoch": 2.5462590506838296,
      "grad_norm": 0.22338885344206338,
      "learning_rate": 6.808355737303895e-07,
      "loss": 0.0134,
      "step": 9495
    },
    {
      "epoch": 2.5465272190935906,
      "grad_norm": 0.23310416661245775,
      "learning_rate": 6.800497953692852e-07,
      "loss": 0.0097,
      "step": 9496
    },
    {
      "epoch": 2.546795387503352,
      "grad_norm": 0.15456995314946473,
      "learning_rate": 6.79264437630262e-07,
      "loss": 0.007,
      "step": 9497
    },
    {
      "epoch": 2.5470635559131134,
      "grad_norm": 0.2588603927048196,
      "learning_rate": 6.784795005897904e-07,
      "loss": 0.0243,
      "step": 9498
    },
    {
      "epoch": 2.547331724322875,
      "grad_norm": 0.35351844221798034,
      "learning_rate": 6.776949843242964e-07,
      "loss": 0.0181,
      "step": 9499
    },
    {
      "epoch": 2.547599892732636,
      "grad_norm": 0.23335516098660394,
      "learning_rate": 6.76910888910165e-07,
      "loss": 0.0136,
      "step": 9500
    },
    {
      "epoch": 2.547868061142397,
      "grad_norm": 0.27165045123093073,
      "learning_rate": 6.76127214423744e-07,
      "loss": 0.0146,
      "step": 9501
    },
    {
      "epoch": 2.548136229552159,
      "grad_norm": 0.2554521551155583,
      "learning_rate": 6.753439609413354e-07,
      "loss": 0.0116,
      "step": 9502
    },
    {
      "epoch": 2.54840439796192,
      "grad_norm": 0.2906989885042511,
      "learning_rate": 6.745611285392045e-07,
      "loss": 0.013,
      "step": 9503
    },
    {
      "epoch": 2.5486725663716814,
      "grad_norm": 0.21810406691905312,
      "learning_rate": 6.737787172935717e-07,
      "loss": 0.0172,
      "step": 9504
    },
    {
      "epoch": 2.548940734781443,
      "grad_norm": 0.27675763821843874,
      "learning_rate": 6.729967272806193e-07,
      "loss": 0.0133,
      "step": 9505
    },
    {
      "epoch": 2.549208903191204,
      "grad_norm": 0.23657425124578915,
      "learning_rate": 6.722151585764891e-07,
      "loss": 0.0157,
      "step": 9506
    },
    {
      "epoch": 2.5494770716009656,
      "grad_norm": 0.19730935926404106,
      "learning_rate": 6.71434011257277e-07,
      "loss": 0.0098,
      "step": 9507
    },
    {
      "epoch": 2.5497452400107266,
      "grad_norm": 0.24848905590579023,
      "learning_rate": 6.70653285399045e-07,
      "loss": 0.0125,
      "step": 9508
    },
    {
      "epoch": 2.550013408420488,
      "grad_norm": 0.35235564818107734,
      "learning_rate": 6.698729810778065e-07,
      "loss": 0.0219,
      "step": 9509
    },
    {
      "epoch": 2.5502815768302494,
      "grad_norm": 0.30984199488128766,
      "learning_rate": 6.690930983695404e-07,
      "loss": 0.0198,
      "step": 9510
    },
    {
      "epoch": 2.5505497452400108,
      "grad_norm": 0.22742442133669358,
      "learning_rate": 6.683136373501814e-07,
      "loss": 0.0138,
      "step": 9511
    },
    {
      "epoch": 2.550817913649772,
      "grad_norm": 0.2378994277247094,
      "learning_rate": 6.675345980956232e-07,
      "loss": 0.0146,
      "step": 9512
    },
    {
      "epoch": 2.551086082059533,
      "grad_norm": 0.21513504156454763,
      "learning_rate": 6.667559806817192e-07,
      "loss": 0.0125,
      "step": 9513
    },
    {
      "epoch": 2.551354250469295,
      "grad_norm": 0.24894058685684417,
      "learning_rate": 6.659777851842792e-07,
      "loss": 0.0122,
      "step": 9514
    },
    {
      "epoch": 2.551622418879056,
      "grad_norm": 0.28221802343009905,
      "learning_rate": 6.652000116790758e-07,
      "loss": 0.0195,
      "step": 9515
    },
    {
      "epoch": 2.5518905872888173,
      "grad_norm": 0.22316113318038344,
      "learning_rate": 6.644226602418396e-07,
      "loss": 0.0122,
      "step": 9516
    },
    {
      "epoch": 2.5521587556985788,
      "grad_norm": 0.252378479072891,
      "learning_rate": 6.636457309482575e-07,
      "loss": 0.0102,
      "step": 9517
    },
    {
      "epoch": 2.55242692410834,
      "grad_norm": 0.27991927000859523,
      "learning_rate": 6.628692238739786e-07,
      "loss": 0.0092,
      "step": 9518
    },
    {
      "epoch": 2.5526950925181016,
      "grad_norm": 0.20249451341450286,
      "learning_rate": 6.620931390946078e-07,
      "loss": 0.0142,
      "step": 9519
    },
    {
      "epoch": 2.5529632609278625,
      "grad_norm": 0.2026432694820056,
      "learning_rate": 6.613174766857117e-07,
      "loss": 0.0144,
      "step": 9520
    },
    {
      "epoch": 2.553231429337624,
      "grad_norm": 0.25899302965105037,
      "learning_rate": 6.605422367228131e-07,
      "loss": 0.0151,
      "step": 9521
    },
    {
      "epoch": 2.5534995977473853,
      "grad_norm": 0.2653073779569596,
      "learning_rate": 6.597674192813958e-07,
      "loss": 0.0145,
      "step": 9522
    },
    {
      "epoch": 2.5537677661571467,
      "grad_norm": 0.35344767780103065,
      "learning_rate": 6.589930244369025e-07,
      "loss": 0.0125,
      "step": 9523
    },
    {
      "epoch": 2.554035934566908,
      "grad_norm": 0.326646743422972,
      "learning_rate": 6.582190522647336e-07,
      "loss": 0.0176,
      "step": 9524
    },
    {
      "epoch": 2.554304102976669,
      "grad_norm": 0.24722480993901944,
      "learning_rate": 6.574455028402476e-07,
      "loss": 0.0132,
      "step": 9525
    },
    {
      "epoch": 2.554572271386431,
      "grad_norm": 0.2193019539577276,
      "learning_rate": 6.566723762387645e-07,
      "loss": 0.0131,
      "step": 9526
    },
    {
      "epoch": 2.554840439796192,
      "grad_norm": 0.30719213146715,
      "learning_rate": 6.558996725355599e-07,
      "loss": 0.0217,
      "step": 9527
    },
    {
      "epoch": 2.5551086082059533,
      "grad_norm": 0.2581147632892664,
      "learning_rate": 6.551273918058715e-07,
      "loss": 0.0086,
      "step": 9528
    },
    {
      "epoch": 2.5553767766157147,
      "grad_norm": 0.23906684604652964,
      "learning_rate": 6.543555341248931e-07,
      "loss": 0.0115,
      "step": 9529
    },
    {
      "epoch": 2.555644945025476,
      "grad_norm": 0.2372846652298987,
      "learning_rate": 6.535840995677794e-07,
      "loss": 0.0126,
      "step": 9530
    },
    {
      "epoch": 2.5559131134352375,
      "grad_norm": 0.21851502494245809,
      "learning_rate": 6.528130882096418e-07,
      "loss": 0.0128,
      "step": 9531
    },
    {
      "epoch": 2.5561812818449985,
      "grad_norm": 0.2700136571282098,
      "learning_rate": 6.520425001255514e-07,
      "loss": 0.0091,
      "step": 9532
    },
    {
      "epoch": 2.55644945025476,
      "grad_norm": 0.2420919389649872,
      "learning_rate": 6.512723353905409e-07,
      "loss": 0.0109,
      "step": 9533
    },
    {
      "epoch": 2.5567176186645213,
      "grad_norm": 0.3193657751309933,
      "learning_rate": 6.505025940795962e-07,
      "loss": 0.0152,
      "step": 9534
    },
    {
      "epoch": 2.5569857870742827,
      "grad_norm": 0.25078611699879266,
      "learning_rate": 6.497332762676678e-07,
      "loss": 0.0135,
      "step": 9535
    },
    {
      "epoch": 2.557253955484044,
      "grad_norm": 0.3148107452211699,
      "learning_rate": 6.489643820296598e-07,
      "loss": 0.0174,
      "step": 9536
    },
    {
      "epoch": 2.557522123893805,
      "grad_norm": 0.2147691272104118,
      "learning_rate": 6.481959114404373e-07,
      "loss": 0.0096,
      "step": 9537
    },
    {
      "epoch": 2.557790292303567,
      "grad_norm": 0.1916516684301197,
      "learning_rate": 6.474278645748266e-07,
      "loss": 0.0091,
      "step": 9538
    },
    {
      "epoch": 2.558058460713328,
      "grad_norm": 0.27664828487899623,
      "learning_rate": 6.466602415076073e-07,
      "loss": 0.0194,
      "step": 9539
    },
    {
      "epoch": 2.5583266291230893,
      "grad_norm": 0.2632584254404309,
      "learning_rate": 6.458930423135234e-07,
      "loss": 0.0174,
      "step": 9540
    },
    {
      "epoch": 2.5585947975328507,
      "grad_norm": 0.1999741147358491,
      "learning_rate": 6.451262670672737e-07,
      "loss": 0.0095,
      "step": 9541
    },
    {
      "epoch": 2.558862965942612,
      "grad_norm": 0.23631347840402372,
      "learning_rate": 6.443599158435166e-07,
      "loss": 0.0171,
      "step": 9542
    },
    {
      "epoch": 2.5591311343523735,
      "grad_norm": 0.20368476714047465,
      "learning_rate": 6.435939887168718e-07,
      "loss": 0.0094,
      "step": 9543
    },
    {
      "epoch": 2.5593993027621345,
      "grad_norm": 0.24072949296536492,
      "learning_rate": 6.428284857619132e-07,
      "loss": 0.0103,
      "step": 9544
    },
    {
      "epoch": 2.559667471171896,
      "grad_norm": 0.24252462716023684,
      "learning_rate": 6.420634070531784e-07,
      "loss": 0.0172,
      "step": 9545
    },
    {
      "epoch": 2.5599356395816573,
      "grad_norm": 0.29879442948606977,
      "learning_rate": 6.412987526651582e-07,
      "loss": 0.0162,
      "step": 9546
    },
    {
      "epoch": 2.5602038079914187,
      "grad_norm": 0.2620983020359227,
      "learning_rate": 6.40534522672307e-07,
      "loss": 0.0159,
      "step": 9547
    },
    {
      "epoch": 2.56047197640118,
      "grad_norm": 0.29115664494466725,
      "learning_rate": 6.397707171490358e-07,
      "loss": 0.0109,
      "step": 9548
    },
    {
      "epoch": 2.560740144810941,
      "grad_norm": 0.2127764916180148,
      "learning_rate": 6.390073361697119e-07,
      "loss": 0.0134,
      "step": 9549
    },
    {
      "epoch": 2.5610083132207024,
      "grad_norm": 0.3353091940086364,
      "learning_rate": 6.382443798086668e-07,
      "loss": 0.0186,
      "step": 9550
    },
    {
      "epoch": 2.561276481630464,
      "grad_norm": 0.2809564606216137,
      "learning_rate": 6.374818481401857e-07,
      "loss": 0.0162,
      "step": 9551
    },
    {
      "epoch": 2.5615446500402252,
      "grad_norm": 0.1879216351288949,
      "learning_rate": 6.36719741238514e-07,
      "loss": 0.01,
      "step": 9552
    },
    {
      "epoch": 2.5618128184499867,
      "grad_norm": 0.24220223472299166,
      "learning_rate": 6.359580591778585e-07,
      "loss": 0.0127,
      "step": 9553
    },
    {
      "epoch": 2.562080986859748,
      "grad_norm": 0.2540490223487067,
      "learning_rate": 6.35196802032379e-07,
      "loss": 0.0138,
      "step": 9554
    },
    {
      "epoch": 2.5623491552695095,
      "grad_norm": 0.24445472446380354,
      "learning_rate": 6.344359698761998e-07,
      "loss": 0.0128,
      "step": 9555
    },
    {
      "epoch": 2.5626173236792704,
      "grad_norm": 0.251579212033659,
      "learning_rate": 6.33675562783399e-07,
      "loss": 0.0143,
      "step": 9556
    },
    {
      "epoch": 2.562885492089032,
      "grad_norm": 0.27640563444558186,
      "learning_rate": 6.329155808280168e-07,
      "loss": 0.0155,
      "step": 9557
    },
    {
      "epoch": 2.5631536604987932,
      "grad_norm": 0.3559963231355134,
      "learning_rate": 6.321560240840496e-07,
      "loss": 0.0138,
      "step": 9558
    },
    {
      "epoch": 2.5634218289085546,
      "grad_norm": 0.19415935577380275,
      "learning_rate": 6.313968926254532e-07,
      "loss": 0.0132,
      "step": 9559
    },
    {
      "epoch": 2.563689997318316,
      "grad_norm": 0.27803548424007435,
      "learning_rate": 6.306381865261457e-07,
      "loss": 0.0135,
      "step": 9560
    },
    {
      "epoch": 2.563958165728077,
      "grad_norm": 0.26510429029863436,
      "learning_rate": 6.298799058599953e-07,
      "loss": 0.0127,
      "step": 9561
    },
    {
      "epoch": 2.5642263341378384,
      "grad_norm": 0.23846288009430872,
      "learning_rate": 6.291220507008361e-07,
      "loss": 0.0116,
      "step": 9562
    },
    {
      "epoch": 2.5644945025476,
      "grad_norm": 0.2606124495259926,
      "learning_rate": 6.283646211224592e-07,
      "loss": 0.0173,
      "step": 9563
    },
    {
      "epoch": 2.564762670957361,
      "grad_norm": 0.21843678254155185,
      "learning_rate": 6.276076171986118e-07,
      "loss": 0.0096,
      "step": 9564
    },
    {
      "epoch": 2.5650308393671226,
      "grad_norm": 0.22313136152370336,
      "learning_rate": 6.268510390030036e-07,
      "loss": 0.0113,
      "step": 9565
    },
    {
      "epoch": 2.565299007776884,
      "grad_norm": 0.3313233812192762,
      "learning_rate": 6.260948866092975e-07,
      "loss": 0.0125,
      "step": 9566
    },
    {
      "epoch": 2.5655671761866454,
      "grad_norm": 0.31481594902030086,
      "learning_rate": 6.253391600911213e-07,
      "loss": 0.0114,
      "step": 9567
    },
    {
      "epoch": 2.5658353445964064,
      "grad_norm": 0.24279749017298913,
      "learning_rate": 6.245838595220554e-07,
      "loss": 0.0112,
      "step": 9568
    },
    {
      "epoch": 2.566103513006168,
      "grad_norm": 0.2111661157942914,
      "learning_rate": 6.238289849756424e-07,
      "loss": 0.0155,
      "step": 9569
    },
    {
      "epoch": 2.566371681415929,
      "grad_norm": 0.23318030760868128,
      "learning_rate": 6.230745365253843e-07,
      "loss": 0.0125,
      "step": 9570
    },
    {
      "epoch": 2.5666398498256906,
      "grad_norm": 0.6204347673957621,
      "learning_rate": 6.223205142447363e-07,
      "loss": 0.0107,
      "step": 9571
    },
    {
      "epoch": 2.566908018235452,
      "grad_norm": 0.19156784436037003,
      "learning_rate": 6.215669182071198e-07,
      "loss": 0.0105,
      "step": 9572
    },
    {
      "epoch": 2.567176186645213,
      "grad_norm": 0.33584013822487757,
      "learning_rate": 6.208137484859056e-07,
      "loss": 0.0138,
      "step": 9573
    },
    {
      "epoch": 2.5674443550549744,
      "grad_norm": 0.24788162282324636,
      "learning_rate": 6.200610051544303e-07,
      "loss": 0.012,
      "step": 9574
    },
    {
      "epoch": 2.5677125234647358,
      "grad_norm": 0.2138360215253765,
      "learning_rate": 6.193086882859878e-07,
      "loss": 0.0116,
      "step": 9575
    },
    {
      "epoch": 2.567980691874497,
      "grad_norm": 0.2635645696162309,
      "learning_rate": 6.185567979538265e-07,
      "loss": 0.0136,
      "step": 9576
    },
    {
      "epoch": 2.5682488602842586,
      "grad_norm": 0.27091264336285575,
      "learning_rate": 6.17805334231158e-07,
      "loss": 0.0195,
      "step": 9577
    },
    {
      "epoch": 2.56851702869402,
      "grad_norm": 0.3002674157313949,
      "learning_rate": 6.170542971911492e-07,
      "loss": 0.0183,
      "step": 9578
    },
    {
      "epoch": 2.5687851971037814,
      "grad_norm": 0.21573304830102366,
      "learning_rate": 6.163036869069267e-07,
      "loss": 0.0102,
      "step": 9579
    },
    {
      "epoch": 2.5690533655135424,
      "grad_norm": 0.22981437936207094,
      "learning_rate": 6.155535034515764e-07,
      "loss": 0.0144,
      "step": 9580
    },
    {
      "epoch": 2.5693215339233038,
      "grad_norm": 0.23787000209345202,
      "learning_rate": 6.148037468981404e-07,
      "loss": 0.0112,
      "step": 9581
    },
    {
      "epoch": 2.569589702333065,
      "grad_norm": 0.1886392946651145,
      "learning_rate": 6.140544173196223e-07,
      "loss": 0.0121,
      "step": 9582
    },
    {
      "epoch": 2.5698578707428266,
      "grad_norm": 0.2633728951485469,
      "learning_rate": 6.133055147889793e-07,
      "loss": 0.0179,
      "step": 9583
    },
    {
      "epoch": 2.570126039152588,
      "grad_norm": 0.31318794485473944,
      "learning_rate": 6.125570393791336e-07,
      "loss": 0.0174,
      "step": 9584
    },
    {
      "epoch": 2.570394207562349,
      "grad_norm": 0.31138280012968056,
      "learning_rate": 6.118089911629605e-07,
      "loss": 0.0184,
      "step": 9585
    },
    {
      "epoch": 2.5706623759721103,
      "grad_norm": 0.3231881498323299,
      "learning_rate": 6.110613702132939e-07,
      "loss": 0.0149,
      "step": 9586
    },
    {
      "epoch": 2.5709305443818717,
      "grad_norm": 0.20943652379313257,
      "learning_rate": 6.103141766029308e-07,
      "loss": 0.0109,
      "step": 9587
    },
    {
      "epoch": 2.571198712791633,
      "grad_norm": 0.27374095689920713,
      "learning_rate": 6.095674104046212e-07,
      "loss": 0.0147,
      "step": 9588
    },
    {
      "epoch": 2.5714668812013945,
      "grad_norm": 0.2889686610070686,
      "learning_rate": 6.088210716910769e-07,
      "loss": 0.0138,
      "step": 9589
    },
    {
      "epoch": 2.571735049611156,
      "grad_norm": 0.2273466533306436,
      "learning_rate": 6.080751605349655e-07,
      "loss": 0.0111,
      "step": 9590
    },
    {
      "epoch": 2.5720032180209174,
      "grad_norm": 0.2842319237140273,
      "learning_rate": 6.073296770089159e-07,
      "loss": 0.0146,
      "step": 9591
    },
    {
      "epoch": 2.5722713864306783,
      "grad_norm": 0.40721174345267563,
      "learning_rate": 6.065846211855136e-07,
      "loss": 0.0283,
      "step": 9592
    },
    {
      "epoch": 2.5725395548404397,
      "grad_norm": 0.3877247192794935,
      "learning_rate": 6.058399931373021e-07,
      "loss": 0.029,
      "step": 9593
    },
    {
      "epoch": 2.572807723250201,
      "grad_norm": 0.21692353197004266,
      "learning_rate": 6.050957929367851e-07,
      "loss": 0.0121,
      "step": 9594
    },
    {
      "epoch": 2.5730758916599625,
      "grad_norm": 0.22672274884597615,
      "learning_rate": 6.043520206564213e-07,
      "loss": 0.01,
      "step": 9595
    },
    {
      "epoch": 2.573344060069724,
      "grad_norm": 0.24027634185062474,
      "learning_rate": 6.036086763686322e-07,
      "loss": 0.0102,
      "step": 9596
    },
    {
      "epoch": 2.573612228479485,
      "grad_norm": 0.21531197569606886,
      "learning_rate": 6.028657601457938e-07,
      "loss": 0.0099,
      "step": 9597
    },
    {
      "epoch": 2.5738803968892463,
      "grad_norm": 0.29345009192828747,
      "learning_rate": 6.021232720602416e-07,
      "loss": 0.0157,
      "step": 9598
    },
    {
      "epoch": 2.5741485652990077,
      "grad_norm": 0.2323772356443803,
      "learning_rate": 6.013812121842711e-07,
      "loss": 0.0128,
      "step": 9599
    },
    {
      "epoch": 2.574416733708769,
      "grad_norm": 0.23008210507945906,
      "learning_rate": 6.006395805901327e-07,
      "loss": 0.0128,
      "step": 9600
    },
    {
      "epoch": 2.5746849021185305,
      "grad_norm": 0.24257049294277327,
      "learning_rate": 5.998983773500383e-07,
      "loss": 0.012,
      "step": 9601
    },
    {
      "epoch": 2.574953070528292,
      "grad_norm": 0.2896981451613895,
      "learning_rate": 5.991576025361578e-07,
      "loss": 0.0145,
      "step": 9602
    },
    {
      "epoch": 2.5752212389380533,
      "grad_norm": 0.26838698337079375,
      "learning_rate": 5.984172562206164e-07,
      "loss": 0.02,
      "step": 9603
    },
    {
      "epoch": 2.5754894073478143,
      "grad_norm": 0.3490655922217611,
      "learning_rate": 5.976773384755019e-07,
      "loss": 0.0147,
      "step": 9604
    },
    {
      "epoch": 2.5757575757575757,
      "grad_norm": 0.28232549831605586,
      "learning_rate": 5.969378493728561e-07,
      "loss": 0.0134,
      "step": 9605
    },
    {
      "epoch": 2.576025744167337,
      "grad_norm": 0.24480418125553421,
      "learning_rate": 5.961987889846821e-07,
      "loss": 0.0103,
      "step": 9606
    },
    {
      "epoch": 2.5762939125770985,
      "grad_norm": 0.22141500648385665,
      "learning_rate": 5.95460157382941e-07,
      "loss": 0.0095,
      "step": 9607
    },
    {
      "epoch": 2.57656208098686,
      "grad_norm": 0.2329942240550676,
      "learning_rate": 5.947219546395511e-07,
      "loss": 0.0098,
      "step": 9608
    },
    {
      "epoch": 2.576830249396621,
      "grad_norm": 0.21964374022417452,
      "learning_rate": 5.939841808263879e-07,
      "loss": 0.0152,
      "step": 9609
    },
    {
      "epoch": 2.5770984178063823,
      "grad_norm": 0.26982686832705055,
      "learning_rate": 5.932468360152866e-07,
      "loss": 0.0175,
      "step": 9610
    },
    {
      "epoch": 2.5773665862161437,
      "grad_norm": 0.20154045021809255,
      "learning_rate": 5.92509920278041e-07,
      "loss": 0.0109,
      "step": 9611
    },
    {
      "epoch": 2.577634754625905,
      "grad_norm": 0.2672357147931063,
      "learning_rate": 5.917734336864039e-07,
      "loss": 0.0125,
      "step": 9612
    },
    {
      "epoch": 2.5779029230356665,
      "grad_norm": 0.2083126523550803,
      "learning_rate": 5.91037376312083e-07,
      "loss": 0.0165,
      "step": 9613
    },
    {
      "epoch": 2.578171091445428,
      "grad_norm": 0.28786657670822385,
      "learning_rate": 5.903017482267487e-07,
      "loss": 0.0139,
      "step": 9614
    },
    {
      "epoch": 2.5784392598551893,
      "grad_norm": 0.32505621117056294,
      "learning_rate": 5.895665495020242e-07,
      "loss": 0.0165,
      "step": 9615
    },
    {
      "epoch": 2.5787074282649503,
      "grad_norm": 0.3021716616317763,
      "learning_rate": 5.888317802094962e-07,
      "loss": 0.0152,
      "step": 9616
    },
    {
      "epoch": 2.5789755966747117,
      "grad_norm": 0.2651405060361694,
      "learning_rate": 5.880974404207051e-07,
      "loss": 0.0139,
      "step": 9617
    },
    {
      "epoch": 2.579243765084473,
      "grad_norm": 0.2030589685361221,
      "learning_rate": 5.873635302071534e-07,
      "loss": 0.0109,
      "step": 9618
    },
    {
      "epoch": 2.5795119334942345,
      "grad_norm": 0.21084956334317448,
      "learning_rate": 5.866300496402999e-07,
      "loss": 0.0145,
      "step": 9619
    },
    {
      "epoch": 2.579780101903996,
      "grad_norm": 0.2919547969973172,
      "learning_rate": 5.858969987915614e-07,
      "loss": 0.0096,
      "step": 9620
    },
    {
      "epoch": 2.580048270313757,
      "grad_norm": 0.22073103686810105,
      "learning_rate": 5.851643777323124e-07,
      "loss": 0.0138,
      "step": 9621
    },
    {
      "epoch": 2.5803164387235182,
      "grad_norm": 0.2527187893110945,
      "learning_rate": 5.84432186533887e-07,
      "loss": 0.022,
      "step": 9622
    },
    {
      "epoch": 2.5805846071332796,
      "grad_norm": 0.25948060660323835,
      "learning_rate": 5.83700425267576e-07,
      "loss": 0.0129,
      "step": 9623
    },
    {
      "epoch": 2.580852775543041,
      "grad_norm": 0.21376553059459844,
      "learning_rate": 5.829690940046306e-07,
      "loss": 0.0113,
      "step": 9624
    },
    {
      "epoch": 2.5811209439528024,
      "grad_norm": 0.27229237607580975,
      "learning_rate": 5.822381928162563e-07,
      "loss": 0.0145,
      "step": 9625
    },
    {
      "epoch": 2.581389112362564,
      "grad_norm": 0.30076006379980436,
      "learning_rate": 5.815077217736209e-07,
      "loss": 0.0175,
      "step": 9626
    },
    {
      "epoch": 2.5816572807723253,
      "grad_norm": 0.20260742850851007,
      "learning_rate": 5.807776809478472e-07,
      "loss": 0.0104,
      "step": 9627
    },
    {
      "epoch": 2.581925449182086,
      "grad_norm": 0.2645360511042723,
      "learning_rate": 5.800480704100181e-07,
      "loss": 0.0147,
      "step": 9628
    },
    {
      "epoch": 2.5821936175918476,
      "grad_norm": 0.2295458897162766,
      "learning_rate": 5.793188902311742e-07,
      "loss": 0.0101,
      "step": 9629
    },
    {
      "epoch": 2.582461786001609,
      "grad_norm": 0.20310056674740432,
      "learning_rate": 5.785901404823124e-07,
      "loss": 0.0086,
      "step": 9630
    },
    {
      "epoch": 2.5827299544113704,
      "grad_norm": 0.3042658023438824,
      "learning_rate": 5.778618212343912e-07,
      "loss": 0.0196,
      "step": 9631
    },
    {
      "epoch": 2.582998122821132,
      "grad_norm": 0.27037584734066117,
      "learning_rate": 5.771339325583225e-07,
      "loss": 0.0127,
      "step": 9632
    },
    {
      "epoch": 2.583266291230893,
      "grad_norm": 0.24557460540419615,
      "learning_rate": 5.764064745249815e-07,
      "loss": 0.013,
      "step": 9633
    },
    {
      "epoch": 2.583534459640654,
      "grad_norm": 0.2531419830928739,
      "learning_rate": 5.75679447205198e-07,
      "loss": 0.013,
      "step": 9634
    },
    {
      "epoch": 2.5838026280504156,
      "grad_norm": 0.2676886021499712,
      "learning_rate": 5.749528506697589e-07,
      "loss": 0.0127,
      "step": 9635
    },
    {
      "epoch": 2.584070796460177,
      "grad_norm": 0.1817620486900637,
      "learning_rate": 5.742266849894141e-07,
      "loss": 0.0084,
      "step": 9636
    },
    {
      "epoch": 2.5843389648699384,
      "grad_norm": 0.29775432667213264,
      "learning_rate": 5.735009502348648e-07,
      "loss": 0.016,
      "step": 9637
    },
    {
      "epoch": 2.5846071332796994,
      "grad_norm": 0.25247249985868303,
      "learning_rate": 5.727756464767764e-07,
      "loss": 0.0109,
      "step": 9638
    },
    {
      "epoch": 2.5848753016894612,
      "grad_norm": 0.3198382469094153,
      "learning_rate": 5.720507737857706e-07,
      "loss": 0.012,
      "step": 9639
    },
    {
      "epoch": 2.585143470099222,
      "grad_norm": 0.22017368222906075,
      "learning_rate": 5.713263322324236e-07,
      "loss": 0.0117,
      "step": 9640
    },
    {
      "epoch": 2.5854116385089836,
      "grad_norm": 0.27690764653361155,
      "learning_rate": 5.706023218872747e-07,
      "loss": 0.0134,
      "step": 9641
    },
    {
      "epoch": 2.585679806918745,
      "grad_norm": 0.2509699069697176,
      "learning_rate": 5.698787428208169e-07,
      "loss": 0.0122,
      "step": 9642
    },
    {
      "epoch": 2.5859479753285064,
      "grad_norm": 0.20979573883287267,
      "learning_rate": 5.691555951035049e-07,
      "loss": 0.0104,
      "step": 9643
    },
    {
      "epoch": 2.586216143738268,
      "grad_norm": 0.2323767133558872,
      "learning_rate": 5.684328788057486e-07,
      "loss": 0.0089,
      "step": 9644
    },
    {
      "epoch": 2.5864843121480288,
      "grad_norm": 0.3283898502354258,
      "learning_rate": 5.677105939979178e-07,
      "loss": 0.0247,
      "step": 9645
    },
    {
      "epoch": 2.58675248055779,
      "grad_norm": 1.0638469810586986,
      "learning_rate": 5.669887407503394e-07,
      "loss": 0.0293,
      "step": 9646
    },
    {
      "epoch": 2.5870206489675516,
      "grad_norm": 0.2308026553970437,
      "learning_rate": 5.662673191332962e-07,
      "loss": 0.011,
      "step": 9647
    },
    {
      "epoch": 2.587288817377313,
      "grad_norm": 0.24368376355800547,
      "learning_rate": 5.655463292170332e-07,
      "loss": 0.0101,
      "step": 9648
    },
    {
      "epoch": 2.5875569857870744,
      "grad_norm": 0.30766176908353654,
      "learning_rate": 5.648257710717519e-07,
      "loss": 0.0152,
      "step": 9649
    },
    {
      "epoch": 2.5878251541968353,
      "grad_norm": 0.26915879885965305,
      "learning_rate": 5.641056447676091e-07,
      "loss": 0.0127,
      "step": 9650
    },
    {
      "epoch": 2.588093322606597,
      "grad_norm": 0.25638270898165233,
      "learning_rate": 5.633859503747241e-07,
      "loss": 0.0157,
      "step": 9651
    },
    {
      "epoch": 2.588361491016358,
      "grad_norm": 0.2558605318508441,
      "learning_rate": 5.626666879631681e-07,
      "loss": 0.0111,
      "step": 9652
    },
    {
      "epoch": 2.5886296594261196,
      "grad_norm": 0.26018949190115487,
      "learning_rate": 5.619478576029774e-07,
      "loss": 0.0114,
      "step": 9653
    },
    {
      "epoch": 2.588897827835881,
      "grad_norm": 0.21605831991327562,
      "learning_rate": 5.612294593641404e-07,
      "loss": 0.0139,
      "step": 9654
    },
    {
      "epoch": 2.5891659962456424,
      "grad_norm": 0.20357054983985248,
      "learning_rate": 5.605114933166056e-07,
      "loss": 0.012,
      "step": 9655
    },
    {
      "epoch": 2.5894341646554038,
      "grad_norm": 0.24555160558933845,
      "learning_rate": 5.597939595302815e-07,
      "loss": 0.0107,
      "step": 9656
    },
    {
      "epoch": 2.5897023330651647,
      "grad_norm": 0.2336453408079745,
      "learning_rate": 5.590768580750306e-07,
      "loss": 0.0116,
      "step": 9657
    },
    {
      "epoch": 2.589970501474926,
      "grad_norm": 0.24621125913539527,
      "learning_rate": 5.583601890206747e-07,
      "loss": 0.0136,
      "step": 9658
    },
    {
      "epoch": 2.5902386698846875,
      "grad_norm": 0.2437413621318941,
      "learning_rate": 5.576439524369959e-07,
      "loss": 0.0107,
      "step": 9659
    },
    {
      "epoch": 2.590506838294449,
      "grad_norm": 0.2704377603389716,
      "learning_rate": 5.569281483937305e-07,
      "loss": 0.0136,
      "step": 9660
    },
    {
      "epoch": 2.5907750067042103,
      "grad_norm": 0.27356030927549896,
      "learning_rate": 5.562127769605757e-07,
      "loss": 0.0155,
      "step": 9661
    },
    {
      "epoch": 2.5910431751139713,
      "grad_norm": 0.24653301104326364,
      "learning_rate": 5.554978382071841e-07,
      "loss": 0.0115,
      "step": 9662
    },
    {
      "epoch": 2.591311343523733,
      "grad_norm": 0.28260482023118916,
      "learning_rate": 5.547833322031693e-07,
      "loss": 0.0133,
      "step": 9663
    },
    {
      "epoch": 2.591579511933494,
      "grad_norm": 0.24103360051292708,
      "learning_rate": 5.540692590180979e-07,
      "loss": 0.0114,
      "step": 9664
    },
    {
      "epoch": 2.5918476803432555,
      "grad_norm": 0.25460054495610895,
      "learning_rate": 5.533556187214995e-07,
      "loss": 0.0124,
      "step": 9665
    },
    {
      "epoch": 2.592115848753017,
      "grad_norm": 0.262079400128003,
      "learning_rate": 5.526424113828599e-07,
      "loss": 0.0145,
      "step": 9666
    },
    {
      "epoch": 2.5923840171627783,
      "grad_norm": 0.27859602085568136,
      "learning_rate": 5.519296370716204e-07,
      "loss": 0.0132,
      "step": 9667
    },
    {
      "epoch": 2.5926521855725397,
      "grad_norm": 0.29989967859772665,
      "learning_rate": 5.512172958571832e-07,
      "loss": 0.0189,
      "step": 9668
    },
    {
      "epoch": 2.5929203539823007,
      "grad_norm": 0.2887292198785437,
      "learning_rate": 5.505053878089073e-07,
      "loss": 0.017,
      "step": 9669
    },
    {
      "epoch": 2.593188522392062,
      "grad_norm": 0.3029981092500451,
      "learning_rate": 5.497939129961072e-07,
      "loss": 0.0174,
      "step": 9670
    },
    {
      "epoch": 2.5934566908018235,
      "grad_norm": 0.2727914934286261,
      "learning_rate": 5.490828714880597e-07,
      "loss": 0.0125,
      "step": 9671
    },
    {
      "epoch": 2.593724859211585,
      "grad_norm": 0.23084753037087935,
      "learning_rate": 5.483722633539951e-07,
      "loss": 0.0112,
      "step": 9672
    },
    {
      "epoch": 2.5939930276213463,
      "grad_norm": 0.22356395308273508,
      "learning_rate": 5.476620886631051e-07,
      "loss": 0.0116,
      "step": 9673
    },
    {
      "epoch": 2.5942611960311073,
      "grad_norm": 0.24075710144226334,
      "learning_rate": 5.469523474845361e-07,
      "loss": 0.0111,
      "step": 9674
    },
    {
      "epoch": 2.594529364440869,
      "grad_norm": 0.22660721133326864,
      "learning_rate": 5.462430398873947e-07,
      "loss": 0.009,
      "step": 9675
    },
    {
      "epoch": 2.59479753285063,
      "grad_norm": 0.21824044277904792,
      "learning_rate": 5.455341659407443e-07,
      "loss": 0.0128,
      "step": 9676
    },
    {
      "epoch": 2.5950657012603915,
      "grad_norm": 0.23477599301938293,
      "learning_rate": 5.448257257136052e-07,
      "loss": 0.0134,
      "step": 9677
    },
    {
      "epoch": 2.595333869670153,
      "grad_norm": 0.20682349158116947,
      "learning_rate": 5.44117719274958e-07,
      "loss": 0.0102,
      "step": 9678
    },
    {
      "epoch": 2.5956020380799143,
      "grad_norm": 0.23453939734887239,
      "learning_rate": 5.434101466937375e-07,
      "loss": 0.0106,
      "step": 9679
    },
    {
      "epoch": 2.5958702064896757,
      "grad_norm": 0.23646830457939813,
      "learning_rate": 5.427030080388395e-07,
      "loss": 0.0112,
      "step": 9680
    },
    {
      "epoch": 2.5961383748994367,
      "grad_norm": 0.2426102653629967,
      "learning_rate": 5.41996303379116e-07,
      "loss": 0.012,
      "step": 9681
    },
    {
      "epoch": 2.596406543309198,
      "grad_norm": 0.3552421422788036,
      "learning_rate": 5.412900327833753e-07,
      "loss": 0.0099,
      "step": 9682
    },
    {
      "epoch": 2.5966747117189595,
      "grad_norm": 0.2189658202513733,
      "learning_rate": 5.405841963203878e-07,
      "loss": 0.0124,
      "step": 9683
    },
    {
      "epoch": 2.596942880128721,
      "grad_norm": 0.2683571119736287,
      "learning_rate": 5.398787940588768e-07,
      "loss": 0.018,
      "step": 9684
    },
    {
      "epoch": 2.5972110485384823,
      "grad_norm": 0.3097451674196923,
      "learning_rate": 5.391738260675266e-07,
      "loss": 0.0159,
      "step": 9685
    },
    {
      "epoch": 2.5974792169482432,
      "grad_norm": 0.2690517626914785,
      "learning_rate": 5.384692924149775e-07,
      "loss": 0.0093,
      "step": 9686
    },
    {
      "epoch": 2.597747385358005,
      "grad_norm": 0.2809748167695715,
      "learning_rate": 5.377651931698275e-07,
      "loss": 0.0212,
      "step": 9687
    },
    {
      "epoch": 2.598015553767766,
      "grad_norm": 0.291855495196349,
      "learning_rate": 5.370615284006353e-07,
      "loss": 0.0174,
      "step": 9688
    },
    {
      "epoch": 2.5982837221775275,
      "grad_norm": 0.27157849635993314,
      "learning_rate": 5.363582981759119e-07,
      "loss": 0.0133,
      "step": 9689
    },
    {
      "epoch": 2.598551890587289,
      "grad_norm": 0.2089887409921479,
      "learning_rate": 5.356555025641313e-07,
      "loss": 0.0131,
      "step": 9690
    },
    {
      "epoch": 2.5988200589970503,
      "grad_norm": 0.18097505551600332,
      "learning_rate": 5.349531416337206e-07,
      "loss": 0.0087,
      "step": 9691
    },
    {
      "epoch": 2.5990882274068117,
      "grad_norm": 0.30237609791212455,
      "learning_rate": 5.342512154530683e-07,
      "loss": 0.0139,
      "step": 9692
    },
    {
      "epoch": 2.5993563958165726,
      "grad_norm": 0.29363668357562717,
      "learning_rate": 5.335497240905207e-07,
      "loss": 0.0148,
      "step": 9693
    },
    {
      "epoch": 2.599624564226334,
      "grad_norm": 0.21664268671829232,
      "learning_rate": 5.328486676143762e-07,
      "loss": 0.0106,
      "step": 9694
    },
    {
      "epoch": 2.5998927326360954,
      "grad_norm": 0.2734728020378876,
      "learning_rate": 5.32148046092898e-07,
      "loss": 0.0113,
      "step": 9695
    },
    {
      "epoch": 2.600160901045857,
      "grad_norm": 0.2720700245107791,
      "learning_rate": 5.314478595943013e-07,
      "loss": 0.0134,
      "step": 9696
    },
    {
      "epoch": 2.6004290694556182,
      "grad_norm": 0.23089031090501963,
      "learning_rate": 5.307481081867632e-07,
      "loss": 0.0115,
      "step": 9697
    },
    {
      "epoch": 2.600697237865379,
      "grad_norm": 0.5793820527992269,
      "learning_rate": 5.300487919384168e-07,
      "loss": 0.0206,
      "step": 9698
    },
    {
      "epoch": 2.600965406275141,
      "grad_norm": 0.24691556390373193,
      "learning_rate": 5.293499109173517e-07,
      "loss": 0.0137,
      "step": 9699
    },
    {
      "epoch": 2.601233574684902,
      "grad_norm": 0.23243860541206107,
      "learning_rate": 5.286514651916169e-07,
      "loss": 0.011,
      "step": 9700
    },
    {
      "epoch": 2.6015017430946634,
      "grad_norm": 0.25151078223917833,
      "learning_rate": 5.279534548292165e-07,
      "loss": 0.0102,
      "step": 9701
    },
    {
      "epoch": 2.601769911504425,
      "grad_norm": 0.21992478956209324,
      "learning_rate": 5.272558798981164e-07,
      "loss": 0.0087,
      "step": 9702
    },
    {
      "epoch": 2.6020380799141862,
      "grad_norm": 0.30193752471653157,
      "learning_rate": 5.265587404662353e-07,
      "loss": 0.0223,
      "step": 9703
    },
    {
      "epoch": 2.6023062483239476,
      "grad_norm": 0.24956749217169652,
      "learning_rate": 5.258620366014533e-07,
      "loss": 0.0134,
      "step": 9704
    },
    {
      "epoch": 2.6025744167337086,
      "grad_norm": 0.18258629712663962,
      "learning_rate": 5.251657683716077e-07,
      "loss": 0.0095,
      "step": 9705
    },
    {
      "epoch": 2.60284258514347,
      "grad_norm": 0.30160196510325177,
      "learning_rate": 5.244699358444893e-07,
      "loss": 0.0109,
      "step": 9706
    },
    {
      "epoch": 2.6031107535532314,
      "grad_norm": 0.3065779431922009,
      "learning_rate": 5.237745390878502e-07,
      "loss": 0.0151,
      "step": 9707
    },
    {
      "epoch": 2.603378921962993,
      "grad_norm": 0.2521336028682725,
      "learning_rate": 5.230795781694015e-07,
      "loss": 0.0112,
      "step": 9708
    },
    {
      "epoch": 2.603647090372754,
      "grad_norm": 0.2269350614112126,
      "learning_rate": 5.223850531568076e-07,
      "loss": 0.0144,
      "step": 9709
    },
    {
      "epoch": 2.603915258782515,
      "grad_norm": 0.26521534861182117,
      "learning_rate": 5.216909641176937e-07,
      "loss": 0.0142,
      "step": 9710
    },
    {
      "epoch": 2.604183427192277,
      "grad_norm": 0.2341468526713637,
      "learning_rate": 5.209973111196404e-07,
      "loss": 0.0116,
      "step": 9711
    },
    {
      "epoch": 2.604451595602038,
      "grad_norm": 0.21313511438846597,
      "learning_rate": 5.203040942301879e-07,
      "loss": 0.0104,
      "step": 9712
    },
    {
      "epoch": 2.6047197640117994,
      "grad_norm": 0.31020807837126285,
      "learning_rate": 5.196113135168318e-07,
      "loss": 0.0173,
      "step": 9713
    },
    {
      "epoch": 2.604987932421561,
      "grad_norm": 0.2648750758505074,
      "learning_rate": 5.189189690470264e-07,
      "loss": 0.0125,
      "step": 9714
    },
    {
      "epoch": 2.605256100831322,
      "grad_norm": 0.27298593211255817,
      "learning_rate": 5.182270608881856e-07,
      "loss": 0.0143,
      "step": 9715
    },
    {
      "epoch": 2.6055242692410836,
      "grad_norm": 0.29689128774433354,
      "learning_rate": 5.175355891076755e-07,
      "loss": 0.0184,
      "step": 9716
    },
    {
      "epoch": 2.6057924376508446,
      "grad_norm": 0.29084685734435906,
      "learning_rate": 5.168445537728256e-07,
      "loss": 0.016,
      "step": 9717
    },
    {
      "epoch": 2.606060606060606,
      "grad_norm": 0.2365977274093258,
      "learning_rate": 5.161539549509186e-07,
      "loss": 0.0122,
      "step": 9718
    },
    {
      "epoch": 2.6063287744703674,
      "grad_norm": 0.35227329569852367,
      "learning_rate": 5.15463792709196e-07,
      "loss": 0.0258,
      "step": 9719
    },
    {
      "epoch": 2.6065969428801288,
      "grad_norm": 0.2518926627281299,
      "learning_rate": 5.147740671148588e-07,
      "loss": 0.0145,
      "step": 9720
    },
    {
      "epoch": 2.60686511128989,
      "grad_norm": 0.2557949679326179,
      "learning_rate": 5.140847782350616e-07,
      "loss": 0.0153,
      "step": 9721
    },
    {
      "epoch": 2.607133279699651,
      "grad_norm": 0.23683905531626048,
      "learning_rate": 5.133959261369203e-07,
      "loss": 0.0107,
      "step": 9722
    },
    {
      "epoch": 2.6074014481094125,
      "grad_norm": 0.2893682387866785,
      "learning_rate": 5.127075108875051e-07,
      "loss": 0.0181,
      "step": 9723
    },
    {
      "epoch": 2.607669616519174,
      "grad_norm": 0.3407715789685843,
      "learning_rate": 5.120195325538463e-07,
      "loss": 0.0166,
      "step": 9724
    },
    {
      "epoch": 2.6079377849289354,
      "grad_norm": 0.3188956049957169,
      "learning_rate": 5.113319912029313e-07,
      "loss": 0.0147,
      "step": 9725
    },
    {
      "epoch": 2.6082059533386968,
      "grad_norm": 0.20626689403722412,
      "learning_rate": 5.106448869017016e-07,
      "loss": 0.0111,
      "step": 9726
    },
    {
      "epoch": 2.608474121748458,
      "grad_norm": 0.20975218998156636,
      "learning_rate": 5.099582197170616e-07,
      "loss": 0.0095,
      "step": 9727
    },
    {
      "epoch": 2.6087422901582196,
      "grad_norm": 0.20807072565918291,
      "learning_rate": 5.092719897158682e-07,
      "loss": 0.0114,
      "step": 9728
    },
    {
      "epoch": 2.6090104585679805,
      "grad_norm": 0.23385151501162746,
      "learning_rate": 5.085861969649392e-07,
      "loss": 0.0122,
      "step": 9729
    },
    {
      "epoch": 2.609278626977742,
      "grad_norm": 0.18421212266994,
      "learning_rate": 5.079008415310476e-07,
      "loss": 0.0138,
      "step": 9730
    },
    {
      "epoch": 2.6095467953875033,
      "grad_norm": 0.22885200669836125,
      "learning_rate": 5.072159234809238e-07,
      "loss": 0.0117,
      "step": 9731
    },
    {
      "epoch": 2.6098149637972647,
      "grad_norm": 0.1938808652433685,
      "learning_rate": 5.065314428812584e-07,
      "loss": 0.0134,
      "step": 9732
    },
    {
      "epoch": 2.610083132207026,
      "grad_norm": 0.26879376980762076,
      "learning_rate": 5.058473997986957e-07,
      "loss": 0.0148,
      "step": 9733
    },
    {
      "epoch": 2.610351300616787,
      "grad_norm": 0.275230619608116,
      "learning_rate": 5.051637942998399e-07,
      "loss": 0.0131,
      "step": 9734
    },
    {
      "epoch": 2.6106194690265485,
      "grad_norm": 0.26023515323108526,
      "learning_rate": 5.044806264512525e-07,
      "loss": 0.0121,
      "step": 9735
    },
    {
      "epoch": 2.61088763743631,
      "grad_norm": 0.29313993264115074,
      "learning_rate": 5.037978963194507e-07,
      "loss": 0.0123,
      "step": 9736
    },
    {
      "epoch": 2.6111558058460713,
      "grad_norm": 0.22652109434881057,
      "learning_rate": 5.031156039709112e-07,
      "loss": 0.0127,
      "step": 9737
    },
    {
      "epoch": 2.6114239742558327,
      "grad_norm": 0.21490142795791686,
      "learning_rate": 5.024337494720655e-07,
      "loss": 0.0154,
      "step": 9738
    },
    {
      "epoch": 2.611692142665594,
      "grad_norm": 0.20401144903952398,
      "learning_rate": 5.017523328893059e-07,
      "loss": 0.0109,
      "step": 9739
    },
    {
      "epoch": 2.6119603110753555,
      "grad_norm": 0.2383892539758945,
      "learning_rate": 5.010713542889778e-07,
      "loss": 0.011,
      "step": 9740
    },
    {
      "epoch": 2.6122284794851165,
      "grad_norm": 0.26005931513439423,
      "learning_rate": 5.003908137373886e-07,
      "loss": 0.0117,
      "step": 9741
    },
    {
      "epoch": 2.612496647894878,
      "grad_norm": 0.197820296499835,
      "learning_rate": 4.997107113007992e-07,
      "loss": 0.0093,
      "step": 9742
    },
    {
      "epoch": 2.6127648163046393,
      "grad_norm": 0.21914521514595675,
      "learning_rate": 4.990310470454296e-07,
      "loss": 0.0141,
      "step": 9743
    },
    {
      "epoch": 2.6130329847144007,
      "grad_norm": 0.23043093774937187,
      "learning_rate": 4.983518210374566e-07,
      "loss": 0.0113,
      "step": 9744
    },
    {
      "epoch": 2.613301153124162,
      "grad_norm": 0.22347172797773107,
      "learning_rate": 4.976730333430163e-07,
      "loss": 0.0103,
      "step": 9745
    },
    {
      "epoch": 2.613569321533923,
      "grad_norm": 0.23695588244843113,
      "learning_rate": 4.969946840281986e-07,
      "loss": 0.0132,
      "step": 9746
    },
    {
      "epoch": 2.6138374899436845,
      "grad_norm": 0.26457191869328045,
      "learning_rate": 4.963167731590535e-07,
      "loss": 0.016,
      "step": 9747
    },
    {
      "epoch": 2.614105658353446,
      "grad_norm": 0.33477402094724973,
      "learning_rate": 4.95639300801587e-07,
      "loss": 0.0129,
      "step": 9748
    },
    {
      "epoch": 2.6143738267632073,
      "grad_norm": 0.27179647606986823,
      "learning_rate": 4.949622670217635e-07,
      "loss": 0.0147,
      "step": 9749
    },
    {
      "epoch": 2.6146419951729687,
      "grad_norm": 0.21369291600714949,
      "learning_rate": 4.942856718855021e-07,
      "loss": 0.0096,
      "step": 9750
    },
    {
      "epoch": 2.61491016358273,
      "grad_norm": 0.25316670681347625,
      "learning_rate": 4.936095154586829e-07,
      "loss": 0.0195,
      "step": 9751
    },
    {
      "epoch": 2.6151783319924915,
      "grad_norm": 0.19662728340224148,
      "learning_rate": 4.929337978071419e-07,
      "loss": 0.0083,
      "step": 9752
    },
    {
      "epoch": 2.6154465004022525,
      "grad_norm": 0.32034527896082726,
      "learning_rate": 4.9225851899667e-07,
      "loss": 0.0192,
      "step": 9753
    },
    {
      "epoch": 2.615714668812014,
      "grad_norm": 0.26566078713506464,
      "learning_rate": 4.91583679093019e-07,
      "loss": 0.0166,
      "step": 9754
    },
    {
      "epoch": 2.6159828372217753,
      "grad_norm": 0.2691305512811401,
      "learning_rate": 4.909092781618958e-07,
      "loss": 0.0204,
      "step": 9755
    },
    {
      "epoch": 2.6162510056315367,
      "grad_norm": 0.16020694821660136,
      "learning_rate": 4.902353162689632e-07,
      "loss": 0.0066,
      "step": 9756
    },
    {
      "epoch": 2.616519174041298,
      "grad_norm": 0.26211268997858794,
      "learning_rate": 4.895617934798463e-07,
      "loss": 0.0111,
      "step": 9757
    },
    {
      "epoch": 2.616787342451059,
      "grad_norm": 0.2144471290943061,
      "learning_rate": 4.888887098601214e-07,
      "loss": 0.0087,
      "step": 9758
    },
    {
      "epoch": 2.6170555108608204,
      "grad_norm": 0.2716440737664424,
      "learning_rate": 4.88216065475327e-07,
      "loss": 0.0155,
      "step": 9759
    },
    {
      "epoch": 2.617323679270582,
      "grad_norm": 0.27763575556640296,
      "learning_rate": 4.875438603909549e-07,
      "loss": 0.0138,
      "step": 9760
    },
    {
      "epoch": 2.6175918476803433,
      "grad_norm": 0.21515967624195942,
      "learning_rate": 4.86872094672457e-07,
      "loss": 0.0148,
      "step": 9761
    },
    {
      "epoch": 2.6178600160901047,
      "grad_norm": 0.2580595934868905,
      "learning_rate": 4.862007683852415e-07,
      "loss": 0.0115,
      "step": 9762
    },
    {
      "epoch": 2.618128184499866,
      "grad_norm": 0.24667810299490367,
      "learning_rate": 4.855298815946724e-07,
      "loss": 0.0138,
      "step": 9763
    },
    {
      "epoch": 2.6183963529096275,
      "grad_norm": 0.20610451413219072,
      "learning_rate": 4.848594343660745e-07,
      "loss": 0.009,
      "step": 9764
    },
    {
      "epoch": 2.6186645213193884,
      "grad_norm": 0.2741111709438867,
      "learning_rate": 4.841894267647246e-07,
      "loss": 0.0143,
      "step": 9765
    },
    {
      "epoch": 2.61893268972915,
      "grad_norm": 0.23834094004737522,
      "learning_rate": 4.83519858855862e-07,
      "loss": 0.0103,
      "step": 9766
    },
    {
      "epoch": 2.6192008581389112,
      "grad_norm": 0.26081984697330396,
      "learning_rate": 4.828507307046798e-07,
      "loss": 0.0132,
      "step": 9767
    },
    {
      "epoch": 2.6194690265486726,
      "grad_norm": 0.17148281182853317,
      "learning_rate": 4.821820423763285e-07,
      "loss": 0.0076,
      "step": 9768
    },
    {
      "epoch": 2.619737194958434,
      "grad_norm": 0.24628890803544362,
      "learning_rate": 4.815137939359176e-07,
      "loss": 0.0115,
      "step": 9769
    },
    {
      "epoch": 2.620005363368195,
      "grad_norm": 0.32299478922866015,
      "learning_rate": 4.808459854485114e-07,
      "loss": 0.0154,
      "step": 9770
    },
    {
      "epoch": 2.6202735317779564,
      "grad_norm": 0.2376250155801046,
      "learning_rate": 4.801786169791339e-07,
      "loss": 0.0139,
      "step": 9771
    },
    {
      "epoch": 2.620541700187718,
      "grad_norm": 0.23268253835744912,
      "learning_rate": 4.795116885927653e-07,
      "loss": 0.0141,
      "step": 9772
    },
    {
      "epoch": 2.620809868597479,
      "grad_norm": 0.41334527447785374,
      "learning_rate": 4.788452003543409e-07,
      "loss": 0.0239,
      "step": 9773
    },
    {
      "epoch": 2.6210780370072406,
      "grad_norm": 0.26891582381501156,
      "learning_rate": 4.781791523287566e-07,
      "loss": 0.0127,
      "step": 9774
    },
    {
      "epoch": 2.621346205417002,
      "grad_norm": 0.21680444918960082,
      "learning_rate": 4.77513544580862e-07,
      "loss": 0.0134,
      "step": 9775
    },
    {
      "epoch": 2.6216143738267634,
      "grad_norm": 0.27399730141778095,
      "learning_rate": 4.768483771754684e-07,
      "loss": 0.0153,
      "step": 9776
    },
    {
      "epoch": 2.6218825422365244,
      "grad_norm": 0.3548706812323244,
      "learning_rate": 4.761836501773376e-07,
      "loss": 0.014,
      "step": 9777
    },
    {
      "epoch": 2.622150710646286,
      "grad_norm": 0.21717215107681162,
      "learning_rate": 4.755193636511957e-07,
      "loss": 0.0069,
      "step": 9778
    },
    {
      "epoch": 2.622418879056047,
      "grad_norm": 0.2603111082292883,
      "learning_rate": 4.7485551766172146e-07,
      "loss": 0.0149,
      "step": 9779
    },
    {
      "epoch": 2.6226870474658086,
      "grad_norm": 0.2722953301480247,
      "learning_rate": 4.741921122735499e-07,
      "loss": 0.0166,
      "step": 9780
    },
    {
      "epoch": 2.62295521587557,
      "grad_norm": 0.21845657531053783,
      "learning_rate": 4.7352914755127764e-07,
      "loss": 0.0159,
      "step": 9781
    },
    {
      "epoch": 2.623223384285331,
      "grad_norm": 0.3281756138935632,
      "learning_rate": 4.7286662355945425e-07,
      "loss": 0.0139,
      "step": 9782
    },
    {
      "epoch": 2.6234915526950924,
      "grad_norm": 0.3985935920349267,
      "learning_rate": 4.7220454036258803e-07,
      "loss": 0.0131,
      "step": 9783
    },
    {
      "epoch": 2.623759721104854,
      "grad_norm": 0.2804698960480728,
      "learning_rate": 4.7154289802514574e-07,
      "loss": 0.0175,
      "step": 9784
    },
    {
      "epoch": 2.624027889514615,
      "grad_norm": 0.2511606691421615,
      "learning_rate": 4.708816966115476e-07,
      "loss": 0.013,
      "step": 9785
    },
    {
      "epoch": 2.6242960579243766,
      "grad_norm": 0.25376220842293135,
      "learning_rate": 4.702209361861748e-07,
      "loss": 0.011,
      "step": 9786
    },
    {
      "epoch": 2.624564226334138,
      "grad_norm": 0.20517549592946643,
      "learning_rate": 4.6956061681336263e-07,
      "loss": 0.009,
      "step": 9787
    },
    {
      "epoch": 2.6248323947438994,
      "grad_norm": 0.2343920918830164,
      "learning_rate": 4.689007385574046e-07,
      "loss": 0.0139,
      "step": 9788
    },
    {
      "epoch": 2.6251005631536604,
      "grad_norm": 0.33759617571009437,
      "learning_rate": 4.682413014825532e-07,
      "loss": 0.0167,
      "step": 9789
    },
    {
      "epoch": 2.6253687315634218,
      "grad_norm": 0.20875591195913856,
      "learning_rate": 4.6758230565301487e-07,
      "loss": 0.0133,
      "step": 9790
    },
    {
      "epoch": 2.625636899973183,
      "grad_norm": 0.20393149873070954,
      "learning_rate": 4.669237511329533e-07,
      "loss": 0.0119,
      "step": 9791
    },
    {
      "epoch": 2.6259050683829446,
      "grad_norm": 0.21029820252477016,
      "learning_rate": 4.6626563798649105e-07,
      "loss": 0.0118,
      "step": 9792
    },
    {
      "epoch": 2.626173236792706,
      "grad_norm": 0.30791521940225175,
      "learning_rate": 4.6560796627770577e-07,
      "loss": 0.0142,
      "step": 9793
    },
    {
      "epoch": 2.626441405202467,
      "grad_norm": 0.23836795309601663,
      "learning_rate": 4.649507360706362e-07,
      "loss": 0.019,
      "step": 9794
    },
    {
      "epoch": 2.6267095736122283,
      "grad_norm": 0.2723769139136709,
      "learning_rate": 4.642939474292713e-07,
      "loss": 0.0171,
      "step": 9795
    },
    {
      "epoch": 2.6269777420219897,
      "grad_norm": 0.2520573065321896,
      "learning_rate": 4.636376004175641e-07,
      "loss": 0.0139,
      "step": 9796
    },
    {
      "epoch": 2.627245910431751,
      "grad_norm": 0.25201012070466955,
      "learning_rate": 4.629816950994187e-07,
      "loss": 0.0138,
      "step": 9797
    },
    {
      "epoch": 2.6275140788415126,
      "grad_norm": 0.23223372355504587,
      "learning_rate": 4.623262315387017e-07,
      "loss": 0.0101,
      "step": 9798
    },
    {
      "epoch": 2.627782247251274,
      "grad_norm": 0.1977085873796665,
      "learning_rate": 4.616712097992304e-07,
      "loss": 0.0113,
      "step": 9799
    },
    {
      "epoch": 2.6280504156610354,
      "grad_norm": 0.2658375286942348,
      "learning_rate": 4.610166299447849e-07,
      "loss": 0.0131,
      "step": 9800
    },
    {
      "epoch": 2.6283185840707963,
      "grad_norm": 0.2848537874996271,
      "learning_rate": 4.603624920391003e-07,
      "loss": 0.0152,
      "step": 9801
    },
    {
      "epoch": 2.6285867524805577,
      "grad_norm": 0.2395367492887739,
      "learning_rate": 4.597087961458674e-07,
      "loss": 0.0136,
      "step": 9802
    },
    {
      "epoch": 2.628854920890319,
      "grad_norm": 0.24939657547570745,
      "learning_rate": 4.590555423287335e-07,
      "loss": 0.0085,
      "step": 9803
    },
    {
      "epoch": 2.6291230893000805,
      "grad_norm": 0.3005423809678072,
      "learning_rate": 4.584027306513067e-07,
      "loss": 0.0236,
      "step": 9804
    },
    {
      "epoch": 2.629391257709842,
      "grad_norm": 0.2680296278567531,
      "learning_rate": 4.577503611771472e-07,
      "loss": 0.0157,
      "step": 9805
    },
    {
      "epoch": 2.629659426119603,
      "grad_norm": 0.30993228623597807,
      "learning_rate": 4.5709843396977704e-07,
      "loss": 0.0188,
      "step": 9806
    },
    {
      "epoch": 2.6299275945293643,
      "grad_norm": 0.24181180880188047,
      "learning_rate": 4.5644694909266984e-07,
      "loss": 0.012,
      "step": 9807
    },
    {
      "epoch": 2.6301957629391257,
      "grad_norm": 0.33592756844666455,
      "learning_rate": 4.5579590660926097e-07,
      "loss": 0.0124,
      "step": 9808
    },
    {
      "epoch": 2.630463931348887,
      "grad_norm": 0.30988412959605455,
      "learning_rate": 4.551453065829392e-07,
      "loss": 0.0152,
      "step": 9809
    },
    {
      "epoch": 2.6307320997586485,
      "grad_norm": 0.2917006963761313,
      "learning_rate": 4.544951490770527e-07,
      "loss": 0.0184,
      "step": 9810
    },
    {
      "epoch": 2.6310002681684095,
      "grad_norm": 0.3244891526455222,
      "learning_rate": 4.5384543415490646e-07,
      "loss": 0.0206,
      "step": 9811
    },
    {
      "epoch": 2.6312684365781713,
      "grad_norm": 0.3140541456906654,
      "learning_rate": 4.5319616187975976e-07,
      "loss": 0.0217,
      "step": 9812
    },
    {
      "epoch": 2.6315366049879323,
      "grad_norm": 0.3144144139788163,
      "learning_rate": 4.5254733231483216e-07,
      "loss": 0.0177,
      "step": 9813
    },
    {
      "epoch": 2.6318047733976937,
      "grad_norm": 0.25031555801201266,
      "learning_rate": 4.5189894552329696e-07,
      "loss": 0.0111,
      "step": 9814
    },
    {
      "epoch": 2.632072941807455,
      "grad_norm": 0.297418032946646,
      "learning_rate": 4.5125100156828637e-07,
      "loss": 0.0162,
      "step": 9815
    },
    {
      "epoch": 2.6323411102172165,
      "grad_norm": 0.20373063294885077,
      "learning_rate": 4.506035005128895e-07,
      "loss": 0.0105,
      "step": 9816
    },
    {
      "epoch": 2.632609278626978,
      "grad_norm": 0.2437427539335146,
      "learning_rate": 4.4995644242015026e-07,
      "loss": 0.0119,
      "step": 9817
    },
    {
      "epoch": 2.632877447036739,
      "grad_norm": 0.22189272201829352,
      "learning_rate": 4.493098273530733e-07,
      "loss": 0.0111,
      "step": 9818
    },
    {
      "epoch": 2.6331456154465003,
      "grad_norm": 0.19138236566282554,
      "learning_rate": 4.4866365537461543e-07,
      "loss": 0.0124,
      "step": 9819
    },
    {
      "epoch": 2.6334137838562617,
      "grad_norm": 0.2233555139810297,
      "learning_rate": 4.4801792654769416e-07,
      "loss": 0.0121,
      "step": 9820
    },
    {
      "epoch": 2.633681952266023,
      "grad_norm": 0.22108736176406157,
      "learning_rate": 4.473726409351831e-07,
      "loss": 0.0098,
      "step": 9821
    },
    {
      "epoch": 2.6339501206757845,
      "grad_norm": 0.2953773818524073,
      "learning_rate": 4.467277985999097e-07,
      "loss": 0.0133,
      "step": 9822
    },
    {
      "epoch": 2.6342182890855455,
      "grad_norm": 0.4509975397616075,
      "learning_rate": 4.460833996046632e-07,
      "loss": 0.0127,
      "step": 9823
    },
    {
      "epoch": 2.6344864574953073,
      "grad_norm": 0.44251804461912303,
      "learning_rate": 4.454394440121845e-07,
      "loss": 0.0132,
      "step": 9824
    },
    {
      "epoch": 2.6347546259050683,
      "grad_norm": 0.2461787367495413,
      "learning_rate": 4.4479593188517576e-07,
      "loss": 0.0158,
      "step": 9825
    },
    {
      "epoch": 2.6350227943148297,
      "grad_norm": 0.3643113212431365,
      "learning_rate": 4.4415286328629346e-07,
      "loss": 0.0256,
      "step": 9826
    },
    {
      "epoch": 2.635290962724591,
      "grad_norm": 0.2777108198795846,
      "learning_rate": 4.435102382781503e-07,
      "loss": 0.015,
      "step": 9827
    },
    {
      "epoch": 2.6355591311343525,
      "grad_norm": 0.23314741268608813,
      "learning_rate": 4.4286805692331905e-07,
      "loss": 0.0151,
      "step": 9828
    },
    {
      "epoch": 2.635827299544114,
      "grad_norm": 0.26955267809173183,
      "learning_rate": 4.4222631928432566e-07,
      "loss": 0.0114,
      "step": 9829
    },
    {
      "epoch": 2.636095467953875,
      "grad_norm": 0.33377537973207594,
      "learning_rate": 4.415850254236542e-07,
      "loss": 0.0189,
      "step": 9830
    },
    {
      "epoch": 2.6363636363636362,
      "grad_norm": 0.24738662950534235,
      "learning_rate": 4.4094417540374745e-07,
      "loss": 0.0171,
      "step": 9831
    },
    {
      "epoch": 2.6366318047733976,
      "grad_norm": 0.2643976740493939,
      "learning_rate": 4.4030376928700156e-07,
      "loss": 0.0161,
      "step": 9832
    },
    {
      "epoch": 2.636899973183159,
      "grad_norm": 0.22302477916559263,
      "learning_rate": 4.396638071357728e-07,
      "loss": 0.0114,
      "step": 9833
    },
    {
      "epoch": 2.6371681415929205,
      "grad_norm": 0.23656257630278724,
      "learning_rate": 4.3902428901237083e-07,
      "loss": 0.0109,
      "step": 9834
    },
    {
      "epoch": 2.6374363100026814,
      "grad_norm": 0.25217439672653874,
      "learning_rate": 4.3838521497906574e-07,
      "loss": 0.0104,
      "step": 9835
    },
    {
      "epoch": 2.6377044784124433,
      "grad_norm": 0.2515161105036834,
      "learning_rate": 4.3774658509808055e-07,
      "loss": 0.009,
      "step": 9836
    },
    {
      "epoch": 2.6379726468222042,
      "grad_norm": 0.1979322847341452,
      "learning_rate": 4.371083994315972e-07,
      "loss": 0.012,
      "step": 9837
    },
    {
      "epoch": 2.6382408152319656,
      "grad_norm": 0.2135186068143742,
      "learning_rate": 4.3647065804175713e-07,
      "loss": 0.013,
      "step": 9838
    },
    {
      "epoch": 2.638508983641727,
      "grad_norm": 0.3826903203181729,
      "learning_rate": 4.358333609906512e-07,
      "loss": 0.0242,
      "step": 9839
    },
    {
      "epoch": 2.6387771520514884,
      "grad_norm": 0.22563769890435414,
      "learning_rate": 4.351965083403331e-07,
      "loss": 0.0134,
      "step": 9840
    },
    {
      "epoch": 2.63904532046125,
      "grad_norm": 0.2094758044403681,
      "learning_rate": 4.3456010015281267e-07,
      "loss": 0.0176,
      "step": 9841
    },
    {
      "epoch": 2.639313488871011,
      "grad_norm": 0.2957474311337995,
      "learning_rate": 4.339241364900537e-07,
      "loss": 0.0157,
      "step": 9842
    },
    {
      "epoch": 2.639581657280772,
      "grad_norm": 0.2944733433078101,
      "learning_rate": 4.332886174139794e-07,
      "loss": 0.0178,
      "step": 9843
    },
    {
      "epoch": 2.6398498256905336,
      "grad_norm": 0.22228963181782915,
      "learning_rate": 4.32653542986467e-07,
      "loss": 0.0113,
      "step": 9844
    },
    {
      "epoch": 2.640117994100295,
      "grad_norm": 0.24811066486274366,
      "learning_rate": 4.320189132693542e-07,
      "loss": 0.0119,
      "step": 9845
    },
    {
      "epoch": 2.6403861625100564,
      "grad_norm": 0.21756295545833926,
      "learning_rate": 4.31384728324431e-07,
      "loss": 0.0105,
      "step": 9846
    },
    {
      "epoch": 2.6406543309198174,
      "grad_norm": 0.29373081143815316,
      "learning_rate": 4.3075098821344754e-07,
      "loss": 0.0199,
      "step": 9847
    },
    {
      "epoch": 2.6409224993295792,
      "grad_norm": 0.2799336473878965,
      "learning_rate": 4.3011769299811047e-07,
      "loss": 0.0205,
      "step": 9848
    },
    {
      "epoch": 2.64119066773934,
      "grad_norm": 0.23688884869691956,
      "learning_rate": 4.294848427400794e-07,
      "loss": 0.013,
      "step": 9849
    },
    {
      "epoch": 2.6414588361491016,
      "grad_norm": 0.18740649651446645,
      "learning_rate": 4.288524375009756e-07,
      "loss": 0.0092,
      "step": 9850
    },
    {
      "epoch": 2.641727004558863,
      "grad_norm": 0.23925814990333194,
      "learning_rate": 4.282204773423737e-07,
      "loss": 0.0113,
      "step": 9851
    },
    {
      "epoch": 2.6419951729686244,
      "grad_norm": 0.2186495537695046,
      "learning_rate": 4.275889623258056e-07,
      "loss": 0.0112,
      "step": 9852
    },
    {
      "epoch": 2.642263341378386,
      "grad_norm": 0.22989413921777618,
      "learning_rate": 4.2695789251276154e-07,
      "loss": 0.0098,
      "step": 9853
    },
    {
      "epoch": 2.6425315097881468,
      "grad_norm": 0.2155083106354327,
      "learning_rate": 4.2632726796468515e-07,
      "loss": 0.0085,
      "step": 9854
    },
    {
      "epoch": 2.642799678197908,
      "grad_norm": 0.3364542638720638,
      "learning_rate": 4.2569708874298123e-07,
      "loss": 0.0197,
      "step": 9855
    },
    {
      "epoch": 2.6430678466076696,
      "grad_norm": 0.25283141312927687,
      "learning_rate": 4.2506735490900565e-07,
      "loss": 0.0185,
      "step": 9856
    },
    {
      "epoch": 2.643336015017431,
      "grad_norm": 0.23132672763976034,
      "learning_rate": 4.2443806652407605e-07,
      "loss": 0.0109,
      "step": 9857
    },
    {
      "epoch": 2.6436041834271924,
      "grad_norm": 0.25488881497009275,
      "learning_rate": 4.2380922364946455e-07,
      "loss": 0.0146,
      "step": 9858
    },
    {
      "epoch": 2.6438723518369533,
      "grad_norm": 0.2321640255945172,
      "learning_rate": 4.2318082634639877e-07,
      "loss": 0.0155,
      "step": 9859
    },
    {
      "epoch": 2.644140520246715,
      "grad_norm": 0.20403663081272436,
      "learning_rate": 4.225528746760654e-07,
      "loss": 0.0116,
      "step": 9860
    },
    {
      "epoch": 2.644408688656476,
      "grad_norm": 0.23332878834091897,
      "learning_rate": 4.21925368699605e-07,
      "loss": 0.0166,
      "step": 9861
    },
    {
      "epoch": 2.6446768570662376,
      "grad_norm": 0.24009026344660306,
      "learning_rate": 4.212983084781175e-07,
      "loss": 0.0126,
      "step": 9862
    },
    {
      "epoch": 2.644945025475999,
      "grad_norm": 0.24662279747164562,
      "learning_rate": 4.2067169407265793e-07,
      "loss": 0.012,
      "step": 9863
    },
    {
      "epoch": 2.6452131938857604,
      "grad_norm": 0.25749865654823295,
      "learning_rate": 4.200455255442365e-07,
      "loss": 0.0183,
      "step": 9864
    },
    {
      "epoch": 2.6454813622955218,
      "grad_norm": 0.2518921503901912,
      "learning_rate": 4.194198029538238e-07,
      "loss": 0.0166,
      "step": 9865
    },
    {
      "epoch": 2.6457495307052827,
      "grad_norm": 0.25365812183381214,
      "learning_rate": 4.187945263623422e-07,
      "loss": 0.0118,
      "step": 9866
    },
    {
      "epoch": 2.646017699115044,
      "grad_norm": 0.24524542855479292,
      "learning_rate": 4.1816969583067526e-07,
      "loss": 0.0171,
      "step": 9867
    },
    {
      "epoch": 2.6462858675248055,
      "grad_norm": 0.24311296221316883,
      "learning_rate": 4.175453114196615e-07,
      "loss": 0.0118,
      "step": 9868
    },
    {
      "epoch": 2.646554035934567,
      "grad_norm": 0.23222526506686028,
      "learning_rate": 4.169213731900934e-07,
      "loss": 0.014,
      "step": 9869
    },
    {
      "epoch": 2.6468222043443284,
      "grad_norm": 0.33713511906599675,
      "learning_rate": 4.162978812027241e-07,
      "loss": 0.0211,
      "step": 9870
    },
    {
      "epoch": 2.6470903727540893,
      "grad_norm": 0.31605268287240623,
      "learning_rate": 4.1567483551825995e-07,
      "loss": 0.0141,
      "step": 9871
    },
    {
      "epoch": 2.647358541163851,
      "grad_norm": 0.26787611519938204,
      "learning_rate": 4.150522361973669e-07,
      "loss": 0.013,
      "step": 9872
    },
    {
      "epoch": 2.647626709573612,
      "grad_norm": 0.2985614483839486,
      "learning_rate": 4.144300833006637e-07,
      "loss": 0.0143,
      "step": 9873
    },
    {
      "epoch": 2.6478948779833735,
      "grad_norm": 0.25781079857726485,
      "learning_rate": 4.138083768887291e-07,
      "loss": 0.0135,
      "step": 9874
    },
    {
      "epoch": 2.648163046393135,
      "grad_norm": 0.2764480029935245,
      "learning_rate": 4.13187117022098e-07,
      "loss": 0.0142,
      "step": 9875
    },
    {
      "epoch": 2.6484312148028963,
      "grad_norm": 0.24696743261577228,
      "learning_rate": 4.125663037612582e-07,
      "loss": 0.0171,
      "step": 9876
    },
    {
      "epoch": 2.6486993832126577,
      "grad_norm": 0.271617375629913,
      "learning_rate": 4.11945937166659e-07,
      "loss": 0.0225,
      "step": 9877
    },
    {
      "epoch": 2.6489675516224187,
      "grad_norm": 0.29396545343999836,
      "learning_rate": 4.113260172987016e-07,
      "loss": 0.0187,
      "step": 9878
    },
    {
      "epoch": 2.64923572003218,
      "grad_norm": 0.5704337946212032,
      "learning_rate": 4.1070654421774767e-07,
      "loss": 0.0102,
      "step": 9879
    },
    {
      "epoch": 2.6495038884419415,
      "grad_norm": 0.2939756820643174,
      "learning_rate": 4.100875179841135e-07,
      "loss": 0.0174,
      "step": 9880
    },
    {
      "epoch": 2.649772056851703,
      "grad_norm": 0.2531099141512188,
      "learning_rate": 4.0946893865807137e-07,
      "loss": 0.0137,
      "step": 9881
    },
    {
      "epoch": 2.6500402252614643,
      "grad_norm": 0.2415155202745602,
      "learning_rate": 4.0885080629985143e-07,
      "loss": 0.0122,
      "step": 9882
    },
    {
      "epoch": 2.6503083936712253,
      "grad_norm": 0.20514956260947687,
      "learning_rate": 4.0823312096963883e-07,
      "loss": 0.0118,
      "step": 9883
    },
    {
      "epoch": 2.650576562080987,
      "grad_norm": 0.23098217548736805,
      "learning_rate": 4.0761588272757614e-07,
      "loss": 0.0114,
      "step": 9884
    },
    {
      "epoch": 2.650844730490748,
      "grad_norm": 0.2800739071676577,
      "learning_rate": 4.0699909163376293e-07,
      "loss": 0.014,
      "step": 9885
    },
    {
      "epoch": 2.6511128989005095,
      "grad_norm": 0.27524636518215334,
      "learning_rate": 4.063827477482535e-07,
      "loss": 0.0153,
      "step": 9886
    },
    {
      "epoch": 2.651381067310271,
      "grad_norm": 0.19079089821662834,
      "learning_rate": 4.0576685113106196e-07,
      "loss": 0.0109,
      "step": 9887
    },
    {
      "epoch": 2.6516492357200323,
      "grad_norm": 0.2543506694931669,
      "learning_rate": 4.051514018421526e-07,
      "loss": 0.0153,
      "step": 9888
    },
    {
      "epoch": 2.6519174041297937,
      "grad_norm": 0.2010846590263328,
      "learning_rate": 4.0453639994145257e-07,
      "loss": 0.0126,
      "step": 9889
    },
    {
      "epoch": 2.6521855725395547,
      "grad_norm": 0.4028046531956334,
      "learning_rate": 4.0392184548884326e-07,
      "loss": 0.0228,
      "step": 9890
    },
    {
      "epoch": 2.652453740949316,
      "grad_norm": 0.2584006098997406,
      "learning_rate": 4.0330773854416025e-07,
      "loss": 0.0138,
      "step": 9891
    },
    {
      "epoch": 2.6527219093590775,
      "grad_norm": 0.3381562416319299,
      "learning_rate": 4.0269407916720016e-07,
      "loss": 0.0253,
      "step": 9892
    },
    {
      "epoch": 2.652990077768839,
      "grad_norm": 0.2820720644139439,
      "learning_rate": 4.020808674177107e-07,
      "loss": 0.0241,
      "step": 9893
    },
    {
      "epoch": 2.6532582461786003,
      "grad_norm": 0.29792382192122213,
      "learning_rate": 4.0146810335540074e-07,
      "loss": 0.015,
      "step": 9894
    },
    {
      "epoch": 2.6535264145883612,
      "grad_norm": 0.2398124551563542,
      "learning_rate": 4.008557870399321e-07,
      "loss": 0.0106,
      "step": 9895
    },
    {
      "epoch": 2.653794582998123,
      "grad_norm": 0.21783553808927528,
      "learning_rate": 4.002439185309248e-07,
      "loss": 0.0115,
      "step": 9896
    },
    {
      "epoch": 2.654062751407884,
      "grad_norm": 0.2614448084487735,
      "learning_rate": 3.996324978879562e-07,
      "loss": 0.0143,
      "step": 9897
    },
    {
      "epoch": 2.6543309198176455,
      "grad_norm": 0.3224886744583689,
      "learning_rate": 3.990215251705565e-07,
      "loss": 0.0193,
      "step": 9898
    },
    {
      "epoch": 2.654599088227407,
      "grad_norm": 0.24490896523396652,
      "learning_rate": 3.9841100043821637e-07,
      "loss": 0.0175,
      "step": 9899
    },
    {
      "epoch": 2.6548672566371683,
      "grad_norm": 0.21172278300241065,
      "learning_rate": 3.978009237503805e-07,
      "loss": 0.0115,
      "step": 9900
    },
    {
      "epoch": 2.6551354250469297,
      "grad_norm": 0.260501669870005,
      "learning_rate": 3.971912951664486e-07,
      "loss": 0.0148,
      "step": 9901
    },
    {
      "epoch": 2.6554035934566906,
      "grad_norm": 0.2383566029409828,
      "learning_rate": 3.9658211474578155e-07,
      "loss": 0.0131,
      "step": 9902
    },
    {
      "epoch": 2.655671761866452,
      "grad_norm": 0.2862660002085008,
      "learning_rate": 3.959733825476908e-07,
      "loss": 0.0136,
      "step": 9903
    },
    {
      "epoch": 2.6559399302762134,
      "grad_norm": 0.24669327237790925,
      "learning_rate": 3.9536509863144944e-07,
      "loss": 0.015,
      "step": 9904
    },
    {
      "epoch": 2.656208098685975,
      "grad_norm": 0.24262496711855822,
      "learning_rate": 3.9475726305628237e-07,
      "loss": 0.0139,
      "step": 9905
    },
    {
      "epoch": 2.6564762670957363,
      "grad_norm": 0.31579566406836074,
      "learning_rate": 3.941498758813739e-07,
      "loss": 0.0142,
      "step": 9906
    },
    {
      "epoch": 2.656744435505497,
      "grad_norm": 0.33095626950689366,
      "learning_rate": 3.9354293716586443e-07,
      "loss": 0.0182,
      "step": 9907
    },
    {
      "epoch": 2.6570126039152586,
      "grad_norm": 0.24178644664050422,
      "learning_rate": 3.9293644696884904e-07,
      "loss": 0.0159,
      "step": 9908
    },
    {
      "epoch": 2.65728077232502,
      "grad_norm": 0.282206204568637,
      "learning_rate": 3.923304053493804e-07,
      "loss": 0.012,
      "step": 9909
    },
    {
      "epoch": 2.6575489407347814,
      "grad_norm": 0.3021979266117166,
      "learning_rate": 3.9172481236646685e-07,
      "loss": 0.0195,
      "step": 9910
    },
    {
      "epoch": 2.657817109144543,
      "grad_norm": 0.208181351757412,
      "learning_rate": 3.911196680790741e-07,
      "loss": 0.0117,
      "step": 9911
    },
    {
      "epoch": 2.6580852775543042,
      "grad_norm": 0.2351755871462643,
      "learning_rate": 3.905149725461227e-07,
      "loss": 0.0168,
      "step": 9912
    },
    {
      "epoch": 2.6583534459640656,
      "grad_norm": 0.20800535440184445,
      "learning_rate": 3.8991072582649003e-07,
      "loss": 0.0108,
      "step": 9913
    },
    {
      "epoch": 2.6586216143738266,
      "grad_norm": 0.2209868783984825,
      "learning_rate": 3.893069279790113e-07,
      "loss": 0.0161,
      "step": 9914
    },
    {
      "epoch": 2.658889782783588,
      "grad_norm": 0.21700817440257628,
      "learning_rate": 3.8870357906247434e-07,
      "loss": 0.009,
      "step": 9915
    },
    {
      "epoch": 2.6591579511933494,
      "grad_norm": 0.25151635547065454,
      "learning_rate": 3.881006791356279e-07,
      "loss": 0.015,
      "step": 9916
    },
    {
      "epoch": 2.659426119603111,
      "grad_norm": 0.21806750130938918,
      "learning_rate": 3.874982282571743e-07,
      "loss": 0.0095,
      "step": 9917
    },
    {
      "epoch": 2.659694288012872,
      "grad_norm": 0.2104150451838226,
      "learning_rate": 3.868962264857717e-07,
      "loss": 0.0105,
      "step": 9918
    },
    {
      "epoch": 2.659962456422633,
      "grad_norm": 0.24542319231381937,
      "learning_rate": 3.8629467388003717e-07,
      "loss": 0.0163,
      "step": 9919
    },
    {
      "epoch": 2.6602306248323946,
      "grad_norm": 0.25043667118971946,
      "learning_rate": 3.8569357049853986e-07,
      "loss": 0.0136,
      "step": 9920
    },
    {
      "epoch": 2.660498793242156,
      "grad_norm": 0.20232480405205847,
      "learning_rate": 3.850929163998096e-07,
      "loss": 0.009,
      "step": 9921
    },
    {
      "epoch": 2.6607669616519174,
      "grad_norm": 0.2579181932483036,
      "learning_rate": 3.844927116423286e-07,
      "loss": 0.0125,
      "step": 9922
    },
    {
      "epoch": 2.661035130061679,
      "grad_norm": 0.27956837188709976,
      "learning_rate": 3.8389295628454003e-07,
      "loss": 0.0155,
      "step": 9923
    },
    {
      "epoch": 2.66130329847144,
      "grad_norm": 0.23746760177657622,
      "learning_rate": 3.8329365038483776e-07,
      "loss": 0.0126,
      "step": 9924
    },
    {
      "epoch": 2.6615714668812016,
      "grad_norm": 0.25289643184437777,
      "learning_rate": 3.826947940015757e-07,
      "loss": 0.0098,
      "step": 9925
    },
    {
      "epoch": 2.6618396352909626,
      "grad_norm": 0.24140113000033572,
      "learning_rate": 3.820963871930622e-07,
      "loss": 0.013,
      "step": 9926
    },
    {
      "epoch": 2.662107803700724,
      "grad_norm": 0.2744140528814186,
      "learning_rate": 3.814984300175645e-07,
      "loss": 0.0092,
      "step": 9927
    },
    {
      "epoch": 2.6623759721104854,
      "grad_norm": 0.28154436701574337,
      "learning_rate": 3.8090092253330156e-07,
      "loss": 0.0163,
      "step": 9928
    },
    {
      "epoch": 2.662644140520247,
      "grad_norm": 0.18346045057850063,
      "learning_rate": 3.803038647984536e-07,
      "loss": 0.0088,
      "step": 9929
    },
    {
      "epoch": 2.662912308930008,
      "grad_norm": 0.2980954695695256,
      "learning_rate": 3.7970725687115184e-07,
      "loss": 0.0127,
      "step": 9930
    },
    {
      "epoch": 2.663180477339769,
      "grad_norm": 0.2217523005011196,
      "learning_rate": 3.7911109880948924e-07,
      "loss": 0.0139,
      "step": 9931
    },
    {
      "epoch": 2.6634486457495306,
      "grad_norm": 0.25559064548276506,
      "learning_rate": 3.7851539067150944e-07,
      "loss": 0.0155,
      "step": 9932
    },
    {
      "epoch": 2.663716814159292,
      "grad_norm": 0.250660682316538,
      "learning_rate": 3.779201325152165e-07,
      "loss": 0.0128,
      "step": 9933
    },
    {
      "epoch": 2.6639849825690534,
      "grad_norm": 0.2774146533516982,
      "learning_rate": 3.7732532439856975e-07,
      "loss": 0.0145,
      "step": 9934
    },
    {
      "epoch": 2.6642531509788148,
      "grad_norm": 0.21089144504948068,
      "learning_rate": 3.767309663794833e-07,
      "loss": 0.0093,
      "step": 9935
    },
    {
      "epoch": 2.664521319388576,
      "grad_norm": 0.21204652663002455,
      "learning_rate": 3.761370585158269e-07,
      "loss": 0.0118,
      "step": 9936
    },
    {
      "epoch": 2.6647894877983376,
      "grad_norm": 0.6118608132707983,
      "learning_rate": 3.7554360086543005e-07,
      "loss": 0.0143,
      "step": 9937
    },
    {
      "epoch": 2.6650576562080985,
      "grad_norm": 0.2279560594254162,
      "learning_rate": 3.749505934860742e-07,
      "loss": 0.0131,
      "step": 9938
    },
    {
      "epoch": 2.66532582461786,
      "grad_norm": 0.20383163567640913,
      "learning_rate": 3.743580364355004e-07,
      "loss": 0.011,
      "step": 9939
    },
    {
      "epoch": 2.6655939930276213,
      "grad_norm": 0.2467789087161366,
      "learning_rate": 3.737659297714036e-07,
      "loss": 0.0151,
      "step": 9940
    },
    {
      "epoch": 2.6658621614373827,
      "grad_norm": 0.2474587250821761,
      "learning_rate": 3.7317427355143596e-07,
      "loss": 0.0149,
      "step": 9941
    },
    {
      "epoch": 2.666130329847144,
      "grad_norm": 0.31730773645879573,
      "learning_rate": 3.7258306783320484e-07,
      "loss": 0.0176,
      "step": 9942
    },
    {
      "epoch": 2.666398498256905,
      "grad_norm": 0.21477088936597902,
      "learning_rate": 3.719923126742747e-07,
      "loss": 0.0097,
      "step": 9943
    },
    {
      "epoch": 2.6666666666666665,
      "grad_norm": 0.21231930654647876,
      "learning_rate": 3.7140200813216677e-07,
      "loss": 0.011,
      "step": 9944
    },
    {
      "epoch": 2.666934835076428,
      "grad_norm": 0.3015072503116844,
      "learning_rate": 3.708121542643561e-07,
      "loss": 0.0218,
      "step": 9945
    },
    {
      "epoch": 2.6672030034861893,
      "grad_norm": 0.27169801782034875,
      "learning_rate": 3.7022275112827574e-07,
      "loss": 0.0144,
      "step": 9946
    },
    {
      "epoch": 2.6674711718959507,
      "grad_norm": 0.26881184706848005,
      "learning_rate": 3.6963379878131534e-07,
      "loss": 0.0169,
      "step": 9947
    },
    {
      "epoch": 2.667739340305712,
      "grad_norm": 0.3009968688747181,
      "learning_rate": 3.690452972808173e-07,
      "loss": 0.0191,
      "step": 9948
    },
    {
      "epoch": 2.6680075087154735,
      "grad_norm": 0.22071975309575997,
      "learning_rate": 3.684572466840841e-07,
      "loss": 0.012,
      "step": 9949
    },
    {
      "epoch": 2.6682756771252345,
      "grad_norm": 0.28707466484422095,
      "learning_rate": 3.678696470483717e-07,
      "loss": 0.0122,
      "step": 9950
    },
    {
      "epoch": 2.668543845534996,
      "grad_norm": 0.33301111582288656,
      "learning_rate": 3.672824984308948e-07,
      "loss": 0.0152,
      "step": 9951
    },
    {
      "epoch": 2.6688120139447573,
      "grad_norm": 0.21585744630120418,
      "learning_rate": 3.666958008888205e-07,
      "loss": 0.0111,
      "step": 9952
    },
    {
      "epoch": 2.6690801823545187,
      "grad_norm": 0.2069907174205666,
      "learning_rate": 3.6610955447927476e-07,
      "loss": 0.0112,
      "step": 9953
    },
    {
      "epoch": 2.66934835076428,
      "grad_norm": 0.2766982426267025,
      "learning_rate": 3.655237592593397e-07,
      "loss": 0.0103,
      "step": 9954
    },
    {
      "epoch": 2.669616519174041,
      "grad_norm": 0.22229575511824387,
      "learning_rate": 3.6493841528605134e-07,
      "loss": 0.0123,
      "step": 9955
    },
    {
      "epoch": 2.6698846875838025,
      "grad_norm": 0.207746300194602,
      "learning_rate": 3.643535226164041e-07,
      "loss": 0.0117,
      "step": 9956
    },
    {
      "epoch": 2.670152855993564,
      "grad_norm": 0.22165786019957717,
      "learning_rate": 3.637690813073469e-07,
      "loss": 0.011,
      "step": 9957
    },
    {
      "epoch": 2.6704210244033253,
      "grad_norm": 0.2308280488204009,
      "learning_rate": 3.631850914157853e-07,
      "loss": 0.0129,
      "step": 9958
    },
    {
      "epoch": 2.6706891928130867,
      "grad_norm": 0.2471966523237142,
      "learning_rate": 3.62601552998581e-07,
      "loss": 0.0129,
      "step": 9959
    },
    {
      "epoch": 2.670957361222848,
      "grad_norm": 0.2732889250894578,
      "learning_rate": 3.620184661125509e-07,
      "loss": 0.0139,
      "step": 9960
    },
    {
      "epoch": 2.6712255296326095,
      "grad_norm": 0.2131520220396545,
      "learning_rate": 3.6143583081447e-07,
      "loss": 0.0087,
      "step": 9961
    },
    {
      "epoch": 2.6714936980423705,
      "grad_norm": 0.2578249048950083,
      "learning_rate": 3.6085364716106575e-07,
      "loss": 0.0155,
      "step": 9962
    },
    {
      "epoch": 2.671761866452132,
      "grad_norm": 0.1970000339008575,
      "learning_rate": 3.602719152090256e-07,
      "loss": 0.0086,
      "step": 9963
    },
    {
      "epoch": 2.6720300348618933,
      "grad_norm": 0.2909319937868586,
      "learning_rate": 3.5969063501499146e-07,
      "loss": 0.0156,
      "step": 9964
    },
    {
      "epoch": 2.6722982032716547,
      "grad_norm": 0.34994095477867876,
      "learning_rate": 3.591098066355597e-07,
      "loss": 0.0227,
      "step": 9965
    },
    {
      "epoch": 2.672566371681416,
      "grad_norm": 0.23894680303076332,
      "learning_rate": 3.5852943012728567e-07,
      "loss": 0.0098,
      "step": 9966
    },
    {
      "epoch": 2.672834540091177,
      "grad_norm": 0.20276268045719037,
      "learning_rate": 3.579495055466775e-07,
      "loss": 0.0103,
      "step": 9967
    },
    {
      "epoch": 2.6731027085009385,
      "grad_norm": 0.3392770049157492,
      "learning_rate": 3.5737003295020165e-07,
      "loss": 0.014,
      "step": 9968
    },
    {
      "epoch": 2.6733708769107,
      "grad_norm": 0.32768824229427923,
      "learning_rate": 3.5679101239427973e-07,
      "loss": 0.0242,
      "step": 9969
    },
    {
      "epoch": 2.6736390453204613,
      "grad_norm": 0.27250459274638705,
      "learning_rate": 3.562124439352893e-07,
      "loss": 0.0131,
      "step": 9970
    },
    {
      "epoch": 2.6739072137302227,
      "grad_norm": 0.2701576415798605,
      "learning_rate": 3.556343276295654e-07,
      "loss": 0.0156,
      "step": 9971
    },
    {
      "epoch": 2.674175382139984,
      "grad_norm": 0.24944676155894252,
      "learning_rate": 3.550566635333952e-07,
      "loss": 0.0194,
      "step": 9972
    },
    {
      "epoch": 2.6744435505497455,
      "grad_norm": 0.2739204898929503,
      "learning_rate": 3.544794517030259e-07,
      "loss": 0.0177,
      "step": 9973
    },
    {
      "epoch": 2.6747117189595064,
      "grad_norm": 0.28043078030697444,
      "learning_rate": 3.539026921946587e-07,
      "loss": 0.016,
      "step": 9974
    },
    {
      "epoch": 2.674979887369268,
      "grad_norm": 0.30813056278004913,
      "learning_rate": 3.533263850644508e-07,
      "loss": 0.02,
      "step": 9975
    },
    {
      "epoch": 2.6752480557790292,
      "grad_norm": 0.20532239152065507,
      "learning_rate": 3.527505303685169e-07,
      "loss": 0.0102,
      "step": 9976
    },
    {
      "epoch": 2.6755162241887906,
      "grad_norm": 0.27545291431470365,
      "learning_rate": 3.5217512816292486e-07,
      "loss": 0.014,
      "step": 9977
    },
    {
      "epoch": 2.675784392598552,
      "grad_norm": 0.2306366449208292,
      "learning_rate": 3.5160017850370154e-07,
      "loss": 0.012,
      "step": 9978
    },
    {
      "epoch": 2.676052561008313,
      "grad_norm": 0.23143912620176887,
      "learning_rate": 3.5102568144682713e-07,
      "loss": 0.0094,
      "step": 9979
    },
    {
      "epoch": 2.6763207294180744,
      "grad_norm": 0.21756483303717467,
      "learning_rate": 3.5045163704823914e-07,
      "loss": 0.0123,
      "step": 9980
    },
    {
      "epoch": 2.676588897827836,
      "grad_norm": 0.2217713308136457,
      "learning_rate": 3.4987804536383174e-07,
      "loss": 0.0102,
      "step": 9981
    },
    {
      "epoch": 2.6768570662375972,
      "grad_norm": 0.19865529999219103,
      "learning_rate": 3.4930490644945246e-07,
      "loss": 0.0102,
      "step": 9982
    },
    {
      "epoch": 2.6771252346473586,
      "grad_norm": 0.2311762957651549,
      "learning_rate": 3.487322203609084e-07,
      "loss": 0.0128,
      "step": 9983
    },
    {
      "epoch": 2.67739340305712,
      "grad_norm": 0.20409204229367964,
      "learning_rate": 3.4815998715395816e-07,
      "loss": 0.0128,
      "step": 9984
    },
    {
      "epoch": 2.6776615714668814,
      "grad_norm": 0.23685875086382915,
      "learning_rate": 3.475882068843189e-07,
      "loss": 0.011,
      "step": 9985
    },
    {
      "epoch": 2.6779297398766424,
      "grad_norm": 0.20661139493652383,
      "learning_rate": 3.4701687960766504e-07,
      "loss": 0.0149,
      "step": 9986
    },
    {
      "epoch": 2.678197908286404,
      "grad_norm": 0.25825783266007146,
      "learning_rate": 3.464460053796237e-07,
      "loss": 0.0115,
      "step": 9987
    },
    {
      "epoch": 2.678466076696165,
      "grad_norm": 0.2449742580834682,
      "learning_rate": 3.458755842557804e-07,
      "loss": 0.0121,
      "step": 9988
    },
    {
      "epoch": 2.6787342451059266,
      "grad_norm": 0.22563721455622898,
      "learning_rate": 3.4530561629167457e-07,
      "loss": 0.0159,
      "step": 9989
    },
    {
      "epoch": 2.679002413515688,
      "grad_norm": 0.17115396014121265,
      "learning_rate": 3.44736101542803e-07,
      "loss": 0.0078,
      "step": 9990
    },
    {
      "epoch": 2.679270581925449,
      "grad_norm": 0.2981313032681861,
      "learning_rate": 3.4416704006461744e-07,
      "loss": 0.0169,
      "step": 9991
    },
    {
      "epoch": 2.6795387503352104,
      "grad_norm": 0.18473567185976986,
      "learning_rate": 3.435984319125263e-07,
      "loss": 0.0089,
      "step": 9992
    },
    {
      "epoch": 2.679806918744972,
      "grad_norm": 0.22467329248265597,
      "learning_rate": 3.4303027714189366e-07,
      "loss": 0.0092,
      "step": 9993
    },
    {
      "epoch": 2.680075087154733,
      "grad_norm": 0.20197785419558836,
      "learning_rate": 3.4246257580803855e-07,
      "loss": 0.0088,
      "step": 9994
    },
    {
      "epoch": 2.6803432555644946,
      "grad_norm": 0.2750637967870523,
      "learning_rate": 3.418953279662374e-07,
      "loss": 0.0113,
      "step": 9995
    },
    {
      "epoch": 2.6806114239742556,
      "grad_norm": 0.27544182920746385,
      "learning_rate": 3.4132853367172145e-07,
      "loss": 0.0133,
      "step": 9996
    },
    {
      "epoch": 2.6808795923840174,
      "grad_norm": 0.21877867924015149,
      "learning_rate": 3.407621929796767e-07,
      "loss": 0.0133,
      "step": 9997
    },
    {
      "epoch": 2.6811477607937784,
      "grad_norm": 0.3302875489444598,
      "learning_rate": 3.401963059452479e-07,
      "loss": 0.0139,
      "step": 9998
    },
    {
      "epoch": 2.6814159292035398,
      "grad_norm": 0.20531504165831338,
      "learning_rate": 3.396308726235326e-07,
      "loss": 0.0123,
      "step": 9999
    },
    {
      "epoch": 2.681684097613301,
      "grad_norm": 0.27873231096842677,
      "learning_rate": 3.3906589306958673e-07,
      "loss": 0.0153,
      "step": 10000
    },
    {
      "epoch": 2.6819522660230626,
      "grad_norm": 0.2673263275754631,
      "learning_rate": 3.3850136733841963e-07,
      "loss": 0.0133,
      "step": 10001
    },
    {
      "epoch": 2.682220434432824,
      "grad_norm": 0.22488874198961106,
      "learning_rate": 3.379372954849985e-07,
      "loss": 0.0117,
      "step": 10002
    },
    {
      "epoch": 2.682488602842585,
      "grad_norm": 0.21001963703954551,
      "learning_rate": 3.373736775642461e-07,
      "loss": 0.0103,
      "step": 10003
    },
    {
      "epoch": 2.6827567712523464,
      "grad_norm": 0.18567183311076185,
      "learning_rate": 3.3681051363103845e-07,
      "loss": 0.01,
      "step": 10004
    },
    {
      "epoch": 2.6830249396621078,
      "grad_norm": 0.2653104643175722,
      "learning_rate": 3.362478037402123e-07,
      "loss": 0.0131,
      "step": 10005
    },
    {
      "epoch": 2.683293108071869,
      "grad_norm": 0.28200697192598273,
      "learning_rate": 3.356855479465537e-07,
      "loss": 0.0144,
      "step": 10006
    },
    {
      "epoch": 2.6835612764816306,
      "grad_norm": 0.20752226266738377,
      "learning_rate": 3.3512374630481125e-07,
      "loss": 0.0105,
      "step": 10007
    },
    {
      "epoch": 2.6838294448913915,
      "grad_norm": 0.28352396377818323,
      "learning_rate": 3.345623988696844e-07,
      "loss": 0.0156,
      "step": 10008
    },
    {
      "epoch": 2.6840976133011534,
      "grad_norm": 0.2614648820818917,
      "learning_rate": 3.340015056958296e-07,
      "loss": 0.0143,
      "step": 10009
    },
    {
      "epoch": 2.6843657817109143,
      "grad_norm": 0.26206296025218345,
      "learning_rate": 3.334410668378607e-07,
      "loss": 0.0115,
      "step": 10010
    },
    {
      "epoch": 2.6846339501206757,
      "grad_norm": 0.23777404811317623,
      "learning_rate": 3.328810823503448e-07,
      "loss": 0.0191,
      "step": 10011
    },
    {
      "epoch": 2.684902118530437,
      "grad_norm": 0.23120094870085794,
      "learning_rate": 3.3232155228780704e-07,
      "loss": 0.0115,
      "step": 10012
    },
    {
      "epoch": 2.6851702869401985,
      "grad_norm": 0.27478399693499433,
      "learning_rate": 3.317624767047284e-07,
      "loss": 0.0175,
      "step": 10013
    },
    {
      "epoch": 2.68543845534996,
      "grad_norm": 0.2052603196988996,
      "learning_rate": 3.3120385565554247e-07,
      "loss": 0.0102,
      "step": 10014
    },
    {
      "epoch": 2.685706623759721,
      "grad_norm": 0.2042457132019438,
      "learning_rate": 3.3064568919464245e-07,
      "loss": 0.0122,
      "step": 10015
    },
    {
      "epoch": 2.6859747921694823,
      "grad_norm": 0.2208740827732014,
      "learning_rate": 3.300879773763743e-07,
      "loss": 0.0125,
      "step": 10016
    },
    {
      "epoch": 2.6862429605792437,
      "grad_norm": 0.260993393374754,
      "learning_rate": 3.2953072025504186e-07,
      "loss": 0.014,
      "step": 10017
    },
    {
      "epoch": 2.686511128989005,
      "grad_norm": 0.23594209576416675,
      "learning_rate": 3.289739178849027e-07,
      "loss": 0.0105,
      "step": 10018
    },
    {
      "epoch": 2.6867792973987665,
      "grad_norm": 0.2117900211064076,
      "learning_rate": 3.284175703201714e-07,
      "loss": 0.0125,
      "step": 10019
    },
    {
      "epoch": 2.6870474658085275,
      "grad_norm": 0.23049379191986713,
      "learning_rate": 3.2786167761502064e-07,
      "loss": 0.0172,
      "step": 10020
    },
    {
      "epoch": 2.6873156342182893,
      "grad_norm": 0.23672297894629166,
      "learning_rate": 3.2730623982357215e-07,
      "loss": 0.0133,
      "step": 10021
    },
    {
      "epoch": 2.6875838026280503,
      "grad_norm": 0.24690973843148198,
      "learning_rate": 3.267512569999093e-07,
      "loss": 0.0128,
      "step": 10022
    },
    {
      "epoch": 2.6878519710378117,
      "grad_norm": 0.3355428205491794,
      "learning_rate": 3.2619672919807054e-07,
      "loss": 0.0186,
      "step": 10023
    },
    {
      "epoch": 2.688120139447573,
      "grad_norm": 0.2436189415662277,
      "learning_rate": 3.2564265647204606e-07,
      "loss": 0.0151,
      "step": 10024
    },
    {
      "epoch": 2.6883883078573345,
      "grad_norm": 0.26350063222866427,
      "learning_rate": 3.2508903887578713e-07,
      "loss": 0.0171,
      "step": 10025
    },
    {
      "epoch": 2.688656476267096,
      "grad_norm": 0.2084271886703435,
      "learning_rate": 3.245358764631962e-07,
      "loss": 0.0164,
      "step": 10026
    },
    {
      "epoch": 2.688924644676857,
      "grad_norm": 0.28702299916448015,
      "learning_rate": 3.2398316928813454e-07,
      "loss": 0.0147,
      "step": 10027
    },
    {
      "epoch": 2.6891928130866183,
      "grad_norm": 0.29750609729810684,
      "learning_rate": 3.234309174044159e-07,
      "loss": 0.0116,
      "step": 10028
    },
    {
      "epoch": 2.6894609814963797,
      "grad_norm": 0.23142944801048917,
      "learning_rate": 3.228791208658133e-07,
      "loss": 0.0139,
      "step": 10029
    },
    {
      "epoch": 2.689729149906141,
      "grad_norm": 0.29508585384150066,
      "learning_rate": 3.2232777972605376e-07,
      "loss": 0.0204,
      "step": 10030
    },
    {
      "epoch": 2.6899973183159025,
      "grad_norm": 0.3598191051320857,
      "learning_rate": 3.2177689403881885e-07,
      "loss": 0.0115,
      "step": 10031
    },
    {
      "epoch": 2.6902654867256635,
      "grad_norm": 0.21732454575062218,
      "learning_rate": 3.212264638577478e-07,
      "loss": 0.0121,
      "step": 10032
    },
    {
      "epoch": 2.6905336551354253,
      "grad_norm": 0.21952435319814462,
      "learning_rate": 3.206764892364339e-07,
      "loss": 0.0109,
      "step": 10033
    },
    {
      "epoch": 2.6908018235451863,
      "grad_norm": 0.31091766598726395,
      "learning_rate": 3.2012697022842654e-07,
      "loss": 0.0155,
      "step": 10034
    },
    {
      "epoch": 2.6910699919549477,
      "grad_norm": 0.38660893547458436,
      "learning_rate": 3.195779068872318e-07,
      "loss": 0.018,
      "step": 10035
    },
    {
      "epoch": 2.691338160364709,
      "grad_norm": 0.25447225161005715,
      "learning_rate": 3.190292992663091e-07,
      "loss": 0.0148,
      "step": 10036
    },
    {
      "epoch": 2.6916063287744705,
      "grad_norm": 0.2376391543941223,
      "learning_rate": 3.1848114741907634e-07,
      "loss": 0.0114,
      "step": 10037
    },
    {
      "epoch": 2.691874497184232,
      "grad_norm": 0.21192542979570933,
      "learning_rate": 3.179334513989046e-07,
      "loss": 0.0118,
      "step": 10038
    },
    {
      "epoch": 2.692142665593993,
      "grad_norm": 0.23090341896737088,
      "learning_rate": 3.173862112591225e-07,
      "loss": 0.0116,
      "step": 10039
    },
    {
      "epoch": 2.6924108340037542,
      "grad_norm": 0.2543475828338398,
      "learning_rate": 3.1683942705301284e-07,
      "loss": 0.0114,
      "step": 10040
    },
    {
      "epoch": 2.6926790024135157,
      "grad_norm": 0.21631872674818173,
      "learning_rate": 3.162930988338148e-07,
      "loss": 0.0119,
      "step": 10041
    },
    {
      "epoch": 2.692947170823277,
      "grad_norm": 0.26669441135239735,
      "learning_rate": 3.157472266547229e-07,
      "loss": 0.0146,
      "step": 10042
    },
    {
      "epoch": 2.6932153392330385,
      "grad_norm": 0.1977208948752991,
      "learning_rate": 3.152018105688859e-07,
      "loss": 0.0121,
      "step": 10043
    },
    {
      "epoch": 2.6934835076427994,
      "grad_norm": 0.24658282883748822,
      "learning_rate": 3.1465685062941234e-07,
      "loss": 0.0141,
      "step": 10044
    },
    {
      "epoch": 2.6937516760525613,
      "grad_norm": 0.2806043403238001,
      "learning_rate": 3.1411234688936145e-07,
      "loss": 0.0143,
      "step": 10045
    },
    {
      "epoch": 2.6940198444623222,
      "grad_norm": 0.21415535357378834,
      "learning_rate": 3.1356829940174973e-07,
      "loss": 0.0117,
      "step": 10046
    },
    {
      "epoch": 2.6942880128720836,
      "grad_norm": 0.2992335510910722,
      "learning_rate": 3.1302470821955143e-07,
      "loss": 0.0148,
      "step": 10047
    },
    {
      "epoch": 2.694556181281845,
      "grad_norm": 0.21903298530033663,
      "learning_rate": 3.1248157339569265e-07,
      "loss": 0.0122,
      "step": 10048
    },
    {
      "epoch": 2.6948243496916064,
      "grad_norm": 0.2383975082075318,
      "learning_rate": 3.1193889498305816e-07,
      "loss": 0.0171,
      "step": 10049
    },
    {
      "epoch": 2.695092518101368,
      "grad_norm": 0.24396938238478916,
      "learning_rate": 3.113966730344875e-07,
      "loss": 0.0098,
      "step": 10050
    },
    {
      "epoch": 2.695360686511129,
      "grad_norm": 0.26109541547491394,
      "learning_rate": 3.108549076027739e-07,
      "loss": 0.0119,
      "step": 10051
    },
    {
      "epoch": 2.69562885492089,
      "grad_norm": 0.3039410893985119,
      "learning_rate": 3.103135987406697e-07,
      "loss": 0.0178,
      "step": 10052
    },
    {
      "epoch": 2.6958970233306516,
      "grad_norm": 0.19690200991629034,
      "learning_rate": 3.0977274650087817e-07,
      "loss": 0.0105,
      "step": 10053
    },
    {
      "epoch": 2.696165191740413,
      "grad_norm": 0.24034828942485903,
      "learning_rate": 3.092323509360634e-07,
      "loss": 0.0122,
      "step": 10054
    },
    {
      "epoch": 2.6964333601501744,
      "grad_norm": 0.44102258950536,
      "learning_rate": 3.086924120988399e-07,
      "loss": 0.0174,
      "step": 10055
    },
    {
      "epoch": 2.6967015285599354,
      "grad_norm": 0.2572631577938057,
      "learning_rate": 3.081529300417818e-07,
      "loss": 0.0125,
      "step": 10056
    },
    {
      "epoch": 2.6969696969696972,
      "grad_norm": 0.3642157881540682,
      "learning_rate": 3.076139048174159e-07,
      "loss": 0.0168,
      "step": 10057
    },
    {
      "epoch": 2.697237865379458,
      "grad_norm": 0.2522319538096927,
      "learning_rate": 3.0707533647822585e-07,
      "loss": 0.0106,
      "step": 10058
    },
    {
      "epoch": 2.6975060337892196,
      "grad_norm": 0.20718773525558293,
      "learning_rate": 3.0653722507665016e-07,
      "loss": 0.0082,
      "step": 10059
    },
    {
      "epoch": 2.697774202198981,
      "grad_norm": 0.3382134690946729,
      "learning_rate": 3.0599957066508534e-07,
      "loss": 0.0133,
      "step": 10060
    },
    {
      "epoch": 2.6980423706087424,
      "grad_norm": 0.24576192144405865,
      "learning_rate": 3.054623732958789e-07,
      "loss": 0.0119,
      "step": 10061
    },
    {
      "epoch": 2.698310539018504,
      "grad_norm": 0.24708451892766228,
      "learning_rate": 3.0492563302133846e-07,
      "loss": 0.0134,
      "step": 10062
    },
    {
      "epoch": 2.6985787074282648,
      "grad_norm": 0.2765739914259692,
      "learning_rate": 3.0438934989372284e-07,
      "loss": 0.0142,
      "step": 10063
    },
    {
      "epoch": 2.698846875838026,
      "grad_norm": 0.6960080105926635,
      "learning_rate": 3.038535239652507e-07,
      "loss": 0.0213,
      "step": 10064
    },
    {
      "epoch": 2.6991150442477876,
      "grad_norm": 0.19112631706057032,
      "learning_rate": 3.033181552880921e-07,
      "loss": 0.0124,
      "step": 10065
    },
    {
      "epoch": 2.699383212657549,
      "grad_norm": 0.349942462571877,
      "learning_rate": 3.027832439143752e-07,
      "loss": 0.0218,
      "step": 10066
    },
    {
      "epoch": 2.6996513810673104,
      "grad_norm": 0.3573100008294919,
      "learning_rate": 3.0224878989618346e-07,
      "loss": 0.0179,
      "step": 10067
    },
    {
      "epoch": 2.6999195494770714,
      "grad_norm": 0.22694198368133903,
      "learning_rate": 3.017147932855552e-07,
      "loss": 0.0146,
      "step": 10068
    },
    {
      "epoch": 2.700187717886833,
      "grad_norm": 0.19966955893551813,
      "learning_rate": 3.011812541344833e-07,
      "loss": 0.009,
      "step": 10069
    },
    {
      "epoch": 2.700455886296594,
      "grad_norm": 0.34611026006222856,
      "learning_rate": 3.006481724949173e-07,
      "loss": 0.0134,
      "step": 10070
    },
    {
      "epoch": 2.7007240547063556,
      "grad_norm": 0.2671818457342193,
      "learning_rate": 3.0011554841876236e-07,
      "loss": 0.0196,
      "step": 10071
    },
    {
      "epoch": 2.700992223116117,
      "grad_norm": 0.19766176087642223,
      "learning_rate": 2.995833819578786e-07,
      "loss": 0.0118,
      "step": 10072
    },
    {
      "epoch": 2.7012603915258784,
      "grad_norm": 0.2615912236081865,
      "learning_rate": 2.9905167316408135e-07,
      "loss": 0.0168,
      "step": 10073
    },
    {
      "epoch": 2.70152855993564,
      "grad_norm": 0.21245520264461423,
      "learning_rate": 2.9852042208914243e-07,
      "loss": 0.0096,
      "step": 10074
    },
    {
      "epoch": 2.7017967283454007,
      "grad_norm": 0.2195460463557986,
      "learning_rate": 2.979896287847872e-07,
      "loss": 0.0132,
      "step": 10075
    },
    {
      "epoch": 2.702064896755162,
      "grad_norm": 0.23276012930541176,
      "learning_rate": 2.974592933026982e-07,
      "loss": 0.0125,
      "step": 10076
    },
    {
      "epoch": 2.7023330651649236,
      "grad_norm": 0.22904691303340635,
      "learning_rate": 2.9692941569451296e-07,
      "loss": 0.0098,
      "step": 10077
    },
    {
      "epoch": 2.702601233574685,
      "grad_norm": 0.3002638794391005,
      "learning_rate": 2.963999960118241e-07,
      "loss": 0.0187,
      "step": 10078
    },
    {
      "epoch": 2.7028694019844464,
      "grad_norm": 0.9046273681117346,
      "learning_rate": 2.9587103430618045e-07,
      "loss": 0.0208,
      "step": 10079
    },
    {
      "epoch": 2.7031375703942073,
      "grad_norm": 0.2590412293383899,
      "learning_rate": 2.953425306290847e-07,
      "loss": 0.012,
      "step": 10080
    },
    {
      "epoch": 2.7034057388039687,
      "grad_norm": 0.33100045991560356,
      "learning_rate": 2.948144850319956e-07,
      "loss": 0.0148,
      "step": 10081
    },
    {
      "epoch": 2.70367390721373,
      "grad_norm": 0.23652284870884083,
      "learning_rate": 2.9428689756632876e-07,
      "loss": 0.013,
      "step": 10082
    },
    {
      "epoch": 2.7039420756234915,
      "grad_norm": 0.22557738497141674,
      "learning_rate": 2.9375976828345254e-07,
      "loss": 0.0105,
      "step": 10083
    },
    {
      "epoch": 2.704210244033253,
      "grad_norm": 0.21076126642046356,
      "learning_rate": 2.932330972346936e-07,
      "loss": 0.0145,
      "step": 10084
    },
    {
      "epoch": 2.7044784124430143,
      "grad_norm": 0.2591797880298394,
      "learning_rate": 2.92706884471331e-07,
      "loss": 0.0179,
      "step": 10085
    },
    {
      "epoch": 2.7047465808527758,
      "grad_norm": 0.2444132936226118,
      "learning_rate": 2.9218113004460204e-07,
      "loss": 0.0154,
      "step": 10086
    },
    {
      "epoch": 2.7050147492625367,
      "grad_norm": 0.3003104042494332,
      "learning_rate": 2.9165583400569686e-07,
      "loss": 0.0127,
      "step": 10087
    },
    {
      "epoch": 2.705282917672298,
      "grad_norm": 0.2469168704697331,
      "learning_rate": 2.9113099640576293e-07,
      "loss": 0.0114,
      "step": 10088
    },
    {
      "epoch": 2.7055510860820595,
      "grad_norm": 0.2807323904151426,
      "learning_rate": 2.906066172959021e-07,
      "loss": 0.013,
      "step": 10089
    },
    {
      "epoch": 2.705819254491821,
      "grad_norm": 0.2595370286899436,
      "learning_rate": 2.900826967271719e-07,
      "loss": 0.0119,
      "step": 10090
    },
    {
      "epoch": 2.7060874229015823,
      "grad_norm": 0.32656626934844674,
      "learning_rate": 2.895592347505849e-07,
      "loss": 0.0186,
      "step": 10091
    },
    {
      "epoch": 2.7063555913113433,
      "grad_norm": 0.20923410802601933,
      "learning_rate": 2.8903623141710966e-07,
      "loss": 0.0133,
      "step": 10092
    },
    {
      "epoch": 2.7066237597211047,
      "grad_norm": 0.24107661944610095,
      "learning_rate": 2.885136867776683e-07,
      "loss": 0.0096,
      "step": 10093
    },
    {
      "epoch": 2.706891928130866,
      "grad_norm": 0.21413002392570152,
      "learning_rate": 2.879916008831413e-07,
      "loss": 0.0146,
      "step": 10094
    },
    {
      "epoch": 2.7071600965406275,
      "grad_norm": 0.28638359566117866,
      "learning_rate": 2.8746997378436117e-07,
      "loss": 0.016,
      "step": 10095
    },
    {
      "epoch": 2.707428264950389,
      "grad_norm": 0.36549008060942845,
      "learning_rate": 2.8694880553211903e-07,
      "loss": 0.0106,
      "step": 10096
    },
    {
      "epoch": 2.7076964333601503,
      "grad_norm": 0.3131668139488774,
      "learning_rate": 2.864280961771582e-07,
      "loss": 0.0129,
      "step": 10097
    },
    {
      "epoch": 2.7079646017699117,
      "grad_norm": 0.2804486231139374,
      "learning_rate": 2.859078457701792e-07,
      "loss": 0.0128,
      "step": 10098
    },
    {
      "epoch": 2.7082327701796727,
      "grad_norm": 0.23309043282431077,
      "learning_rate": 2.8538805436183815e-07,
      "loss": 0.0112,
      "step": 10099
    },
    {
      "epoch": 2.708500938589434,
      "grad_norm": 0.2570046630685327,
      "learning_rate": 2.848687220027446e-07,
      "loss": 0.0132,
      "step": 10100
    },
    {
      "epoch": 2.7087691069991955,
      "grad_norm": 0.1885343389458234,
      "learning_rate": 2.843498487434659e-07,
      "loss": 0.0086,
      "step": 10101
    },
    {
      "epoch": 2.709037275408957,
      "grad_norm": 0.25947297903197725,
      "learning_rate": 2.838314346345217e-07,
      "loss": 0.0189,
      "step": 10102
    },
    {
      "epoch": 2.7093054438187183,
      "grad_norm": 0.2522663419739868,
      "learning_rate": 2.8331347972639034e-07,
      "loss": 0.0122,
      "step": 10103
    },
    {
      "epoch": 2.7095736122284793,
      "grad_norm": 0.3346132572757494,
      "learning_rate": 2.8279598406950324e-07,
      "loss": 0.0193,
      "step": 10104
    },
    {
      "epoch": 2.7098417806382407,
      "grad_norm": 0.18673409938462998,
      "learning_rate": 2.8227894771424624e-07,
      "loss": 0.0125,
      "step": 10105
    },
    {
      "epoch": 2.710109949048002,
      "grad_norm": 0.2208312631162379,
      "learning_rate": 2.817623707109635e-07,
      "loss": 0.0139,
      "step": 10106
    },
    {
      "epoch": 2.7103781174577635,
      "grad_norm": 0.20773393432408196,
      "learning_rate": 2.8124625310995136e-07,
      "loss": 0.0119,
      "step": 10107
    },
    {
      "epoch": 2.710646285867525,
      "grad_norm": 0.22906634615160432,
      "learning_rate": 2.807305949614636e-07,
      "loss": 0.0124,
      "step": 10108
    },
    {
      "epoch": 2.7109144542772863,
      "grad_norm": 0.28832244296912496,
      "learning_rate": 2.8021539631570896e-07,
      "loss": 0.0139,
      "step": 10109
    },
    {
      "epoch": 2.7111826226870477,
      "grad_norm": 0.20823692963071677,
      "learning_rate": 2.7970065722284944e-07,
      "loss": 0.0117,
      "step": 10110
    },
    {
      "epoch": 2.7114507910968086,
      "grad_norm": 0.409505423065092,
      "learning_rate": 2.7918637773300615e-07,
      "loss": 0.0173,
      "step": 10111
    },
    {
      "epoch": 2.71171895950657,
      "grad_norm": 0.26316303728179363,
      "learning_rate": 2.786725578962507e-07,
      "loss": 0.0142,
      "step": 10112
    },
    {
      "epoch": 2.7119871279163315,
      "grad_norm": 0.22194101728311771,
      "learning_rate": 2.7815919776261415e-07,
      "loss": 0.0148,
      "step": 10113
    },
    {
      "epoch": 2.712255296326093,
      "grad_norm": 0.28920723421608957,
      "learning_rate": 2.7764629738207927e-07,
      "loss": 0.0137,
      "step": 10114
    },
    {
      "epoch": 2.7125234647358543,
      "grad_norm": 0.2973945618538858,
      "learning_rate": 2.771338568045867e-07,
      "loss": 0.0157,
      "step": 10115
    },
    {
      "epoch": 2.7127916331456152,
      "grad_norm": 0.1694951523636434,
      "learning_rate": 2.766218760800327e-07,
      "loss": 0.0052,
      "step": 10116
    },
    {
      "epoch": 2.7130598015553766,
      "grad_norm": 0.22499071681850838,
      "learning_rate": 2.7611035525826614e-07,
      "loss": 0.0129,
      "step": 10117
    },
    {
      "epoch": 2.713327969965138,
      "grad_norm": 0.21686949477638687,
      "learning_rate": 2.755992943890923e-07,
      "loss": 0.0137,
      "step": 10118
    },
    {
      "epoch": 2.7135961383748994,
      "grad_norm": 0.19787165624272043,
      "learning_rate": 2.750886935222724e-07,
      "loss": 0.0109,
      "step": 10119
    },
    {
      "epoch": 2.713864306784661,
      "grad_norm": 0.27834288394583495,
      "learning_rate": 2.745785527075212e-07,
      "loss": 0.0135,
      "step": 10120
    },
    {
      "epoch": 2.7141324751944222,
      "grad_norm": 0.25931741692890964,
      "learning_rate": 2.740688719945112e-07,
      "loss": 0.0142,
      "step": 10121
    },
    {
      "epoch": 2.7144006436041836,
      "grad_norm": 0.22433002100203742,
      "learning_rate": 2.7355965143286753e-07,
      "loss": 0.0133,
      "step": 10122
    },
    {
      "epoch": 2.7146688120139446,
      "grad_norm": 0.2193055930782264,
      "learning_rate": 2.7305089107217295e-07,
      "loss": 0.015,
      "step": 10123
    },
    {
      "epoch": 2.714936980423706,
      "grad_norm": 0.22021854904232677,
      "learning_rate": 2.725425909619628e-07,
      "loss": 0.0123,
      "step": 10124
    },
    {
      "epoch": 2.7152051488334674,
      "grad_norm": 0.27222485997377344,
      "learning_rate": 2.720347511517291e-07,
      "loss": 0.014,
      "step": 10125
    },
    {
      "epoch": 2.715473317243229,
      "grad_norm": 0.2774959107938613,
      "learning_rate": 2.7152737169092005e-07,
      "loss": 0.0153,
      "step": 10126
    },
    {
      "epoch": 2.7157414856529902,
      "grad_norm": 0.23089316599624216,
      "learning_rate": 2.710204526289356e-07,
      "loss": 0.015,
      "step": 10127
    },
    {
      "epoch": 2.716009654062751,
      "grad_norm": 0.22850734753539792,
      "learning_rate": 2.705139940151358e-07,
      "loss": 0.0113,
      "step": 10128
    },
    {
      "epoch": 2.7162778224725126,
      "grad_norm": 0.21382124220516613,
      "learning_rate": 2.7000799589883166e-07,
      "loss": 0.0117,
      "step": 10129
    },
    {
      "epoch": 2.716545990882274,
      "grad_norm": 0.26662564640278064,
      "learning_rate": 2.6950245832929046e-07,
      "loss": 0.0163,
      "step": 10130
    },
    {
      "epoch": 2.7168141592920354,
      "grad_norm": 0.2660463732809721,
      "learning_rate": 2.689973813557367e-07,
      "loss": 0.0143,
      "step": 10131
    },
    {
      "epoch": 2.717082327701797,
      "grad_norm": 0.28877367165669693,
      "learning_rate": 2.684927650273461e-07,
      "loss": 0.017,
      "step": 10132
    },
    {
      "epoch": 2.717350496111558,
      "grad_norm": 0.23269313473078823,
      "learning_rate": 2.679886093932543e-07,
      "loss": 0.0131,
      "step": 10133
    },
    {
      "epoch": 2.7176186645213196,
      "grad_norm": 0.4221372247463844,
      "learning_rate": 2.674849145025471e-07,
      "loss": 0.0115,
      "step": 10134
    },
    {
      "epoch": 2.7178868329310806,
      "grad_norm": 0.272509445753347,
      "learning_rate": 2.6698168040426964e-07,
      "loss": 0.0143,
      "step": 10135
    },
    {
      "epoch": 2.718155001340842,
      "grad_norm": 0.25765841801038875,
      "learning_rate": 2.664789071474205e-07,
      "loss": 0.0129,
      "step": 10136
    },
    {
      "epoch": 2.7184231697506034,
      "grad_norm": 0.2262280714403418,
      "learning_rate": 2.659765947809523e-07,
      "loss": 0.0117,
      "step": 10137
    },
    {
      "epoch": 2.718691338160365,
      "grad_norm": 0.16487515671626357,
      "learning_rate": 2.6547474335377523e-07,
      "loss": 0.0065,
      "step": 10138
    },
    {
      "epoch": 2.718959506570126,
      "grad_norm": 0.25622958538467194,
      "learning_rate": 2.6497335291475133e-07,
      "loss": 0.0105,
      "step": 10139
    },
    {
      "epoch": 2.719227674979887,
      "grad_norm": 0.24826793676375233,
      "learning_rate": 2.644724235127016e-07,
      "loss": 0.0143,
      "step": 10140
    },
    {
      "epoch": 2.7194958433896486,
      "grad_norm": 0.2567674688866037,
      "learning_rate": 2.6397195519639983e-07,
      "loss": 0.0135,
      "step": 10141
    },
    {
      "epoch": 2.71976401179941,
      "grad_norm": 0.1691342962338418,
      "learning_rate": 2.634719480145742e-07,
      "loss": 0.0093,
      "step": 10142
    },
    {
      "epoch": 2.7200321802091714,
      "grad_norm": 0.28165669379629904,
      "learning_rate": 2.6297240201591025e-07,
      "loss": 0.0134,
      "step": 10143
    },
    {
      "epoch": 2.7203003486189328,
      "grad_norm": 0.27800160821264563,
      "learning_rate": 2.624733172490462e-07,
      "loss": 0.0142,
      "step": 10144
    },
    {
      "epoch": 2.720568517028694,
      "grad_norm": 0.23164947532696567,
      "learning_rate": 2.6197469376257724e-07,
      "loss": 0.0096,
      "step": 10145
    },
    {
      "epoch": 2.7208366854384556,
      "grad_norm": 0.2152848189530246,
      "learning_rate": 2.614765316050544e-07,
      "loss": 0.0173,
      "step": 10146
    },
    {
      "epoch": 2.7211048538482165,
      "grad_norm": 0.23289168459585774,
      "learning_rate": 2.6097883082498e-07,
      "loss": 0.0124,
      "step": 10147
    },
    {
      "epoch": 2.721373022257978,
      "grad_norm": 0.2218983801521475,
      "learning_rate": 2.604815914708153e-07,
      "loss": 0.0138,
      "step": 10148
    },
    {
      "epoch": 2.7216411906677394,
      "grad_norm": 0.2635082937707218,
      "learning_rate": 2.5998481359097496e-07,
      "loss": 0.0143,
      "step": 10149
    },
    {
      "epoch": 2.7219093590775008,
      "grad_norm": 0.28151299970223026,
      "learning_rate": 2.5948849723382906e-07,
      "loss": 0.0234,
      "step": 10150
    },
    {
      "epoch": 2.722177527487262,
      "grad_norm": 0.24786388950432012,
      "learning_rate": 2.589926424477018e-07,
      "loss": 0.016,
      "step": 10151
    },
    {
      "epoch": 2.722445695897023,
      "grad_norm": 0.25488643545589035,
      "learning_rate": 2.584972492808735e-07,
      "loss": 0.0165,
      "step": 10152
    },
    {
      "epoch": 2.7227138643067845,
      "grad_norm": 0.2717890470224508,
      "learning_rate": 2.580023177815816e-07,
      "loss": 0.011,
      "step": 10153
    },
    {
      "epoch": 2.722982032716546,
      "grad_norm": 0.27751522142135937,
      "learning_rate": 2.5750784799801255e-07,
      "loss": 0.0159,
      "step": 10154
    },
    {
      "epoch": 2.7232502011263073,
      "grad_norm": 0.2170607179433834,
      "learning_rate": 2.5701383997831284e-07,
      "loss": 0.0119,
      "step": 10155
    },
    {
      "epoch": 2.7235183695360687,
      "grad_norm": 0.2817797760124718,
      "learning_rate": 2.5652029377058465e-07,
      "loss": 0.0142,
      "step": 10156
    },
    {
      "epoch": 2.72378653794583,
      "grad_norm": 0.232999674125874,
      "learning_rate": 2.560272094228805e-07,
      "loss": 0.0119,
      "step": 10157
    },
    {
      "epoch": 2.7240547063555915,
      "grad_norm": 0.3220650106553259,
      "learning_rate": 2.5553458698321263e-07,
      "loss": 0.0161,
      "step": 10158
    },
    {
      "epoch": 2.7243228747653525,
      "grad_norm": 0.2558991183839685,
      "learning_rate": 2.5504242649954546e-07,
      "loss": 0.014,
      "step": 10159
    },
    {
      "epoch": 2.724591043175114,
      "grad_norm": 0.21497126511920372,
      "learning_rate": 2.545507280198001e-07,
      "loss": 0.0105,
      "step": 10160
    },
    {
      "epoch": 2.7248592115848753,
      "grad_norm": 0.3251135723498152,
      "learning_rate": 2.540594915918504e-07,
      "loss": 0.0274,
      "step": 10161
    },
    {
      "epoch": 2.7251273799946367,
      "grad_norm": 0.20574726414035674,
      "learning_rate": 2.5356871726352817e-07,
      "loss": 0.009,
      "step": 10162
    },
    {
      "epoch": 2.725395548404398,
      "grad_norm": 0.23393065244442354,
      "learning_rate": 2.5307840508261904e-07,
      "loss": 0.0116,
      "step": 10163
    },
    {
      "epoch": 2.725663716814159,
      "grad_norm": 0.2144140296155032,
      "learning_rate": 2.525885550968621e-07,
      "loss": 0.011,
      "step": 10164
    },
    {
      "epoch": 2.7259318852239205,
      "grad_norm": 0.22094296365370902,
      "learning_rate": 2.5209916735395467e-07,
      "loss": 0.012,
      "step": 10165
    },
    {
      "epoch": 2.726200053633682,
      "grad_norm": 0.1940292829081444,
      "learning_rate": 2.5161024190154427e-07,
      "loss": 0.009,
      "step": 10166
    },
    {
      "epoch": 2.7264682220434433,
      "grad_norm": 0.2154080547700454,
      "learning_rate": 2.5112177878723833e-07,
      "loss": 0.0127,
      "step": 10167
    },
    {
      "epoch": 2.7267363904532047,
      "grad_norm": 0.2538117030047245,
      "learning_rate": 2.506337780585971e-07,
      "loss": 0.0145,
      "step": 10168
    },
    {
      "epoch": 2.7270045588629657,
      "grad_norm": 0.2372653039010175,
      "learning_rate": 2.5014623976313546e-07,
      "loss": 0.0234,
      "step": 10169
    },
    {
      "epoch": 2.7272727272727275,
      "grad_norm": 0.3209603181066786,
      "learning_rate": 2.496591639483242e-07,
      "loss": 0.0149,
      "step": 10170
    },
    {
      "epoch": 2.7275408956824885,
      "grad_norm": 0.28903420557869525,
      "learning_rate": 2.491725506615872e-07,
      "loss": 0.0138,
      "step": 10171
    },
    {
      "epoch": 2.72780906409225,
      "grad_norm": 0.2319112836457248,
      "learning_rate": 2.4868639995030643e-07,
      "loss": 0.0112,
      "step": 10172
    },
    {
      "epoch": 2.7280772325020113,
      "grad_norm": 0.1765857314299491,
      "learning_rate": 2.4820071186181626e-07,
      "loss": 0.0084,
      "step": 10173
    },
    {
      "epoch": 2.7283454009117727,
      "grad_norm": 0.2847146295715286,
      "learning_rate": 2.477154864434067e-07,
      "loss": 0.0165,
      "step": 10174
    },
    {
      "epoch": 2.728613569321534,
      "grad_norm": 0.3082077564415258,
      "learning_rate": 2.472307237423244e-07,
      "loss": 0.015,
      "step": 10175
    },
    {
      "epoch": 2.728881737731295,
      "grad_norm": 0.32325184162292425,
      "learning_rate": 2.467464238057665e-07,
      "loss": 0.0095,
      "step": 10176
    },
    {
      "epoch": 2.7291499061410565,
      "grad_norm": 0.2070781417671562,
      "learning_rate": 2.46262586680891e-07,
      "loss": 0.0135,
      "step": 10177
    },
    {
      "epoch": 2.729418074550818,
      "grad_norm": 0.3202936888724756,
      "learning_rate": 2.457792124148062e-07,
      "loss": 0.0172,
      "step": 10178
    },
    {
      "epoch": 2.7296862429605793,
      "grad_norm": 0.26757917313394114,
      "learning_rate": 2.452963010545767e-07,
      "loss": 0.0131,
      "step": 10179
    },
    {
      "epoch": 2.7299544113703407,
      "grad_norm": 0.19102831234702866,
      "learning_rate": 2.448138526472238e-07,
      "loss": 0.0093,
      "step": 10180
    },
    {
      "epoch": 2.7302225797801016,
      "grad_norm": 0.22118426493052626,
      "learning_rate": 2.4433186723972045e-07,
      "loss": 0.0112,
      "step": 10181
    },
    {
      "epoch": 2.7304907481898635,
      "grad_norm": 0.18941046825797625,
      "learning_rate": 2.438503448789975e-07,
      "loss": 0.0111,
      "step": 10182
    },
    {
      "epoch": 2.7307589165996244,
      "grad_norm": 0.48831203652584326,
      "learning_rate": 2.4336928561193905e-07,
      "loss": 0.0137,
      "step": 10183
    },
    {
      "epoch": 2.731027085009386,
      "grad_norm": 0.23001800833565597,
      "learning_rate": 2.428886894853838e-07,
      "loss": 0.0137,
      "step": 10184
    },
    {
      "epoch": 2.7312952534191473,
      "grad_norm": 0.22151576935831416,
      "learning_rate": 2.4240855654612816e-07,
      "loss": 0.0121,
      "step": 10185
    },
    {
      "epoch": 2.7315634218289087,
      "grad_norm": 0.2636365617702101,
      "learning_rate": 2.4192888684091973e-07,
      "loss": 0.0158,
      "step": 10186
    },
    {
      "epoch": 2.73183159023867,
      "grad_norm": 0.25730794354940206,
      "learning_rate": 2.4144968041646343e-07,
      "loss": 0.0131,
      "step": 10187
    },
    {
      "epoch": 2.732099758648431,
      "grad_norm": 0.2266705224171433,
      "learning_rate": 2.40970937319418e-07,
      "loss": 0.0111,
      "step": 10188
    },
    {
      "epoch": 2.7323679270581924,
      "grad_norm": 0.1923055710247471,
      "learning_rate": 2.404926575963973e-07,
      "loss": 0.0082,
      "step": 10189
    },
    {
      "epoch": 2.732636095467954,
      "grad_norm": 0.2572208066024229,
      "learning_rate": 2.400148412939707e-07,
      "loss": 0.0137,
      "step": 10190
    },
    {
      "epoch": 2.7329042638777152,
      "grad_norm": 0.20868694979313432,
      "learning_rate": 2.3953748845866096e-07,
      "loss": 0.0114,
      "step": 10191
    },
    {
      "epoch": 2.7331724322874766,
      "grad_norm": 0.15256566167067673,
      "learning_rate": 2.390605991369477e-07,
      "loss": 0.0077,
      "step": 10192
    },
    {
      "epoch": 2.7334406006972376,
      "grad_norm": 0.26927236153892187,
      "learning_rate": 2.385841733752631e-07,
      "loss": 0.015,
      "step": 10193
    },
    {
      "epoch": 2.7337087691069994,
      "grad_norm": 0.2368625325813369,
      "learning_rate": 2.3810821121999616e-07,
      "loss": 0.0187,
      "step": 10194
    },
    {
      "epoch": 2.7339769375167604,
      "grad_norm": 0.2289195945449989,
      "learning_rate": 2.37632712717491e-07,
      "loss": 0.0128,
      "step": 10195
    },
    {
      "epoch": 2.734245105926522,
      "grad_norm": 0.22975656614025008,
      "learning_rate": 2.3715767791404387e-07,
      "loss": 0.016,
      "step": 10196
    },
    {
      "epoch": 2.734513274336283,
      "grad_norm": 0.2525111323513369,
      "learning_rate": 2.3668310685590945e-07,
      "loss": 0.0162,
      "step": 10197
    },
    {
      "epoch": 2.7347814427460446,
      "grad_norm": 0.3531149250183119,
      "learning_rate": 2.3620899958929354e-07,
      "loss": 0.0123,
      "step": 10198
    },
    {
      "epoch": 2.735049611155806,
      "grad_norm": 0.2363860754779539,
      "learning_rate": 2.3573535616036036e-07,
      "loss": 0.0124,
      "step": 10199
    },
    {
      "epoch": 2.735317779565567,
      "grad_norm": 0.33596924191149186,
      "learning_rate": 2.3526217661522632e-07,
      "loss": 0.0128,
      "step": 10200
    },
    {
      "epoch": 2.7355859479753284,
      "grad_norm": 0.23131245148846483,
      "learning_rate": 2.347894609999646e-07,
      "loss": 0.0113,
      "step": 10201
    },
    {
      "epoch": 2.73585411638509,
      "grad_norm": 0.19865156117447463,
      "learning_rate": 2.3431720936060166e-07,
      "loss": 0.009,
      "step": 10202
    },
    {
      "epoch": 2.736122284794851,
      "grad_norm": 0.27214901144528864,
      "learning_rate": 2.3384542174311908e-07,
      "loss": 0.012,
      "step": 10203
    },
    {
      "epoch": 2.7363904532046126,
      "grad_norm": 0.2697743120573721,
      "learning_rate": 2.3337409819345348e-07,
      "loss": 0.016,
      "step": 10204
    },
    {
      "epoch": 2.7366586216143736,
      "grad_norm": 0.2697348045195244,
      "learning_rate": 2.3290323875749754e-07,
      "loss": 0.0162,
      "step": 10205
    },
    {
      "epoch": 2.7369267900241354,
      "grad_norm": 0.22591343896685726,
      "learning_rate": 2.324328434810963e-07,
      "loss": 0.0125,
      "step": 10206
    },
    {
      "epoch": 2.7371949584338964,
      "grad_norm": 0.2908205132514924,
      "learning_rate": 2.3196291241005198e-07,
      "loss": 0.0195,
      "step": 10207
    },
    {
      "epoch": 2.737463126843658,
      "grad_norm": 0.2867602777133908,
      "learning_rate": 2.314934455901191e-07,
      "loss": 0.0151,
      "step": 10208
    },
    {
      "epoch": 2.737731295253419,
      "grad_norm": 0.1901515658051912,
      "learning_rate": 2.3102444306701e-07,
      "loss": 0.0124,
      "step": 10209
    },
    {
      "epoch": 2.7379994636631806,
      "grad_norm": 0.17653524232123627,
      "learning_rate": 2.3055590488638923e-07,
      "loss": 0.0102,
      "step": 10210
    },
    {
      "epoch": 2.738267632072942,
      "grad_norm": 0.28221149013141034,
      "learning_rate": 2.30087831093877e-07,
      "loss": 0.012,
      "step": 10211
    },
    {
      "epoch": 2.738535800482703,
      "grad_norm": 0.39051086649300387,
      "learning_rate": 2.296202217350496e-07,
      "loss": 0.0245,
      "step": 10212
    },
    {
      "epoch": 2.7388039688924644,
      "grad_norm": 0.295938389878031,
      "learning_rate": 2.291530768554362e-07,
      "loss": 0.014,
      "step": 10213
    },
    {
      "epoch": 2.7390721373022258,
      "grad_norm": 0.2392009215987486,
      "learning_rate": 2.286863965005204e-07,
      "loss": 0.0139,
      "step": 10214
    },
    {
      "epoch": 2.739340305711987,
      "grad_norm": 0.20829768416617755,
      "learning_rate": 2.282201807157436e-07,
      "loss": 0.0104,
      "step": 10215
    },
    {
      "epoch": 2.7396084741217486,
      "grad_norm": 0.20632633331738573,
      "learning_rate": 2.2775442954649786e-07,
      "loss": 0.0105,
      "step": 10216
    },
    {
      "epoch": 2.7398766425315095,
      "grad_norm": 0.2724529903820612,
      "learning_rate": 2.272891430381341e-07,
      "loss": 0.0142,
      "step": 10217
    },
    {
      "epoch": 2.7401448109412714,
      "grad_norm": 0.23027998567315175,
      "learning_rate": 2.26824321235955e-07,
      "loss": 0.0098,
      "step": 10218
    },
    {
      "epoch": 2.7404129793510323,
      "grad_norm": 0.3177990679203518,
      "learning_rate": 2.2635996418521932e-07,
      "loss": 0.0184,
      "step": 10219
    },
    {
      "epoch": 2.7406811477607937,
      "grad_norm": 0.18400364856294238,
      "learning_rate": 2.2589607193113982e-07,
      "loss": 0.014,
      "step": 10220
    },
    {
      "epoch": 2.740949316170555,
      "grad_norm": 0.1740607695303438,
      "learning_rate": 2.2543264451888424e-07,
      "loss": 0.0099,
      "step": 10221
    },
    {
      "epoch": 2.7412174845803166,
      "grad_norm": 0.36787119496236165,
      "learning_rate": 2.2496968199357705e-07,
      "loss": 0.0132,
      "step": 10222
    },
    {
      "epoch": 2.741485652990078,
      "grad_norm": 0.25222622983786247,
      "learning_rate": 2.2450718440029384e-07,
      "loss": 0.0114,
      "step": 10223
    },
    {
      "epoch": 2.741753821399839,
      "grad_norm": 0.23327250994105983,
      "learning_rate": 2.240451517840675e-07,
      "loss": 0.011,
      "step": 10224
    },
    {
      "epoch": 2.7420219898096003,
      "grad_norm": 0.2015999879553685,
      "learning_rate": 2.2358358418988536e-07,
      "loss": 0.0099,
      "step": 10225
    },
    {
      "epoch": 2.7422901582193617,
      "grad_norm": 0.25925054014848964,
      "learning_rate": 2.2312248166268758e-07,
      "loss": 0.0136,
      "step": 10226
    },
    {
      "epoch": 2.742558326629123,
      "grad_norm": 0.22678978116344337,
      "learning_rate": 2.2266184424737214e-07,
      "loss": 0.0125,
      "step": 10227
    },
    {
      "epoch": 2.7428264950388845,
      "grad_norm": 0.37391953056572996,
      "learning_rate": 2.2220167198878872e-07,
      "loss": 0.0158,
      "step": 10228
    },
    {
      "epoch": 2.7430946634486455,
      "grad_norm": 0.20504321962445998,
      "learning_rate": 2.2174196493174427e-07,
      "loss": 0.0104,
      "step": 10229
    },
    {
      "epoch": 2.7433628318584073,
      "grad_norm": 0.33702282905633757,
      "learning_rate": 2.2128272312099798e-07,
      "loss": 0.0128,
      "step": 10230
    },
    {
      "epoch": 2.7436310002681683,
      "grad_norm": 0.22776614844497967,
      "learning_rate": 2.208239466012657e-07,
      "loss": 0.0127,
      "step": 10231
    },
    {
      "epoch": 2.7438991686779297,
      "grad_norm": 0.3371623245606038,
      "learning_rate": 2.2036563541721789e-07,
      "loss": 0.024,
      "step": 10232
    },
    {
      "epoch": 2.744167337087691,
      "grad_norm": 0.22644136497706224,
      "learning_rate": 2.1990778961347768e-07,
      "loss": 0.0103,
      "step": 10233
    },
    {
      "epoch": 2.7444355054974525,
      "grad_norm": 0.28102350803875015,
      "learning_rate": 2.1945040923462614e-07,
      "loss": 0.0153,
      "step": 10234
    },
    {
      "epoch": 2.744703673907214,
      "grad_norm": 0.29960356666261806,
      "learning_rate": 2.1899349432519535e-07,
      "loss": 0.0222,
      "step": 10235
    },
    {
      "epoch": 2.744971842316975,
      "grad_norm": 0.296130382429142,
      "learning_rate": 2.1853704492967533e-07,
      "loss": 0.0128,
      "step": 10236
    },
    {
      "epoch": 2.7452400107267363,
      "grad_norm": 0.29711706491116263,
      "learning_rate": 2.180810610925077e-07,
      "loss": 0.0117,
      "step": 10237
    },
    {
      "epoch": 2.7455081791364977,
      "grad_norm": 0.2738283388550113,
      "learning_rate": 2.1762554285809258e-07,
      "loss": 0.0164,
      "step": 10238
    },
    {
      "epoch": 2.745776347546259,
      "grad_norm": 0.31280942182568827,
      "learning_rate": 2.1717049027078106e-07,
      "loss": 0.0181,
      "step": 10239
    },
    {
      "epoch": 2.7460445159560205,
      "grad_norm": 0.2682370193407817,
      "learning_rate": 2.1671590337488e-07,
      "loss": 0.0088,
      "step": 10240
    },
    {
      "epoch": 2.7463126843657815,
      "grad_norm": 0.200177021042343,
      "learning_rate": 2.1626178221465232e-07,
      "loss": 0.0113,
      "step": 10241
    },
    {
      "epoch": 2.7465808527755433,
      "grad_norm": 0.23299220982992275,
      "learning_rate": 2.1580812683431486e-07,
      "loss": 0.0121,
      "step": 10242
    },
    {
      "epoch": 2.7468490211853043,
      "grad_norm": 0.26113559828095206,
      "learning_rate": 2.153549372780378e-07,
      "loss": 0.0131,
      "step": 10243
    },
    {
      "epoch": 2.7471171895950657,
      "grad_norm": 0.30596950527762934,
      "learning_rate": 2.149022135899481e-07,
      "loss": 0.0112,
      "step": 10244
    },
    {
      "epoch": 2.747385358004827,
      "grad_norm": 0.2535548164256236,
      "learning_rate": 2.14449955814125e-07,
      "loss": 0.0158,
      "step": 10245
    },
    {
      "epoch": 2.7476535264145885,
      "grad_norm": 0.2241292329254599,
      "learning_rate": 2.1399816399460482e-07,
      "loss": 0.0125,
      "step": 10246
    },
    {
      "epoch": 2.74792169482435,
      "grad_norm": 0.29028886493608785,
      "learning_rate": 2.1354683817537637e-07,
      "loss": 0.0147,
      "step": 10247
    },
    {
      "epoch": 2.748189863234111,
      "grad_norm": 0.29754240709631247,
      "learning_rate": 2.1309597840038442e-07,
      "loss": 0.0134,
      "step": 10248
    },
    {
      "epoch": 2.7484580316438723,
      "grad_norm": 0.23200396880854102,
      "learning_rate": 2.1264558471352892e-07,
      "loss": 0.0134,
      "step": 10249
    },
    {
      "epoch": 2.7487262000536337,
      "grad_norm": 0.20940793849411454,
      "learning_rate": 2.1219565715866254e-07,
      "loss": 0.0093,
      "step": 10250
    },
    {
      "epoch": 2.748994368463395,
      "grad_norm": 0.24840800744348698,
      "learning_rate": 2.1174619577959355e-07,
      "loss": 0.0112,
      "step": 10251
    },
    {
      "epoch": 2.7492625368731565,
      "grad_norm": 0.23828654577646663,
      "learning_rate": 2.1129720062008418e-07,
      "loss": 0.0118,
      "step": 10252
    },
    {
      "epoch": 2.7495307052829174,
      "grad_norm": 0.22373162259808743,
      "learning_rate": 2.1084867172385282e-07,
      "loss": 0.0168,
      "step": 10253
    },
    {
      "epoch": 2.7497988736926793,
      "grad_norm": 0.23456597607422838,
      "learning_rate": 2.1040060913457171e-07,
      "loss": 0.0097,
      "step": 10254
    },
    {
      "epoch": 2.7500670421024402,
      "grad_norm": 0.2851256361029505,
      "learning_rate": 2.0995301289586656e-07,
      "loss": 0.0191,
      "step": 10255
    },
    {
      "epoch": 2.7503352105122016,
      "grad_norm": 0.19421952339888432,
      "learning_rate": 2.0950588305132024e-07,
      "loss": 0.0096,
      "step": 10256
    },
    {
      "epoch": 2.750603378921963,
      "grad_norm": 0.2081009268081321,
      "learning_rate": 2.0905921964446628e-07,
      "loss": 0.0071,
      "step": 10257
    },
    {
      "epoch": 2.7508715473317245,
      "grad_norm": 0.226446561290927,
      "learning_rate": 2.0861302271879713e-07,
      "loss": 0.0115,
      "step": 10258
    },
    {
      "epoch": 2.751139715741486,
      "grad_norm": 0.2510517821505098,
      "learning_rate": 2.0816729231775746e-07,
      "loss": 0.0134,
      "step": 10259
    },
    {
      "epoch": 2.751407884151247,
      "grad_norm": 0.3237875328770131,
      "learning_rate": 2.0772202848474587e-07,
      "loss": 0.0171,
      "step": 10260
    },
    {
      "epoch": 2.7516760525610082,
      "grad_norm": 0.23965919450221349,
      "learning_rate": 2.0727723126311715e-07,
      "loss": 0.0141,
      "step": 10261
    },
    {
      "epoch": 2.7519442209707696,
      "grad_norm": 0.2827376277757224,
      "learning_rate": 2.0683290069618055e-07,
      "loss": 0.0156,
      "step": 10262
    },
    {
      "epoch": 2.752212389380531,
      "grad_norm": 0.2506367706107894,
      "learning_rate": 2.0638903682719814e-07,
      "loss": 0.0131,
      "step": 10263
    },
    {
      "epoch": 2.7524805577902924,
      "grad_norm": 0.2766255687350105,
      "learning_rate": 2.059456396993892e-07,
      "loss": 0.0121,
      "step": 10264
    },
    {
      "epoch": 2.7527487262000534,
      "grad_norm": 0.2311344240626549,
      "learning_rate": 2.055027093559242e-07,
      "loss": 0.0122,
      "step": 10265
    },
    {
      "epoch": 2.753016894609815,
      "grad_norm": 0.2205222699875854,
      "learning_rate": 2.0506024583993256e-07,
      "loss": 0.0096,
      "step": 10266
    },
    {
      "epoch": 2.753285063019576,
      "grad_norm": 0.17159448343696415,
      "learning_rate": 2.0461824919449313e-07,
      "loss": 0.0078,
      "step": 10267
    },
    {
      "epoch": 2.7535532314293376,
      "grad_norm": 0.26279692574572383,
      "learning_rate": 2.0417671946264373e-07,
      "loss": 0.0125,
      "step": 10268
    },
    {
      "epoch": 2.753821399839099,
      "grad_norm": 0.6925207219733114,
      "learning_rate": 2.0373565668737493e-07,
      "loss": 0.0154,
      "step": 10269
    },
    {
      "epoch": 2.7540895682488604,
      "grad_norm": 0.22817350756203078,
      "learning_rate": 2.032950609116313e-07,
      "loss": 0.0127,
      "step": 10270
    },
    {
      "epoch": 2.754357736658622,
      "grad_norm": 0.48832810313717717,
      "learning_rate": 2.0285493217831242e-07,
      "loss": 0.0143,
      "step": 10271
    },
    {
      "epoch": 2.754625905068383,
      "grad_norm": 0.32622837516723335,
      "learning_rate": 2.0241527053027233e-07,
      "loss": 0.0257,
      "step": 10272
    },
    {
      "epoch": 2.754894073478144,
      "grad_norm": 0.24955000156224624,
      "learning_rate": 2.0197607601032067e-07,
      "loss": 0.0111,
      "step": 10273
    },
    {
      "epoch": 2.7551622418879056,
      "grad_norm": 0.3205979609858217,
      "learning_rate": 2.015373486612199e-07,
      "loss": 0.026,
      "step": 10274
    },
    {
      "epoch": 2.755430410297667,
      "grad_norm": 0.22140605952717737,
      "learning_rate": 2.010990885256875e-07,
      "loss": 0.0126,
      "step": 10275
    },
    {
      "epoch": 2.7556985787074284,
      "grad_norm": 0.2612654470878207,
      "learning_rate": 2.0066129564639658e-07,
      "loss": 0.0126,
      "step": 10276
    },
    {
      "epoch": 2.7559667471171894,
      "grad_norm": 0.379459263528785,
      "learning_rate": 2.002239700659725e-07,
      "loss": 0.0199,
      "step": 10277
    },
    {
      "epoch": 2.7562349155269508,
      "grad_norm": 0.36599950663530945,
      "learning_rate": 1.9978711182699783e-07,
      "loss": 0.0122,
      "step": 10278
    },
    {
      "epoch": 2.756503083936712,
      "grad_norm": 0.45018369586827184,
      "learning_rate": 1.9935072097200693e-07,
      "loss": 0.0165,
      "step": 10279
    },
    {
      "epoch": 2.7567712523464736,
      "grad_norm": 0.31162531621030426,
      "learning_rate": 1.989147975434913e-07,
      "loss": 0.0161,
      "step": 10280
    },
    {
      "epoch": 2.757039420756235,
      "grad_norm": 0.24638270124983994,
      "learning_rate": 1.9847934158389537e-07,
      "loss": 0.013,
      "step": 10281
    },
    {
      "epoch": 2.7573075891659964,
      "grad_norm": 0.2881351775983531,
      "learning_rate": 1.9804435313561798e-07,
      "loss": 0.0152,
      "step": 10282
    },
    {
      "epoch": 2.757575757575758,
      "grad_norm": 0.2815664557017304,
      "learning_rate": 1.976098322410136e-07,
      "loss": 0.0147,
      "step": 10283
    },
    {
      "epoch": 2.7578439259855188,
      "grad_norm": 0.4014799892547454,
      "learning_rate": 1.9717577894238893e-07,
      "loss": 0.0247,
      "step": 10284
    },
    {
      "epoch": 2.75811209439528,
      "grad_norm": 0.2727684225006125,
      "learning_rate": 1.9674219328200738e-07,
      "loss": 0.0113,
      "step": 10285
    },
    {
      "epoch": 2.7583802628050416,
      "grad_norm": 0.2464718439555285,
      "learning_rate": 1.9630907530208741e-07,
      "loss": 0.0125,
      "step": 10286
    },
    {
      "epoch": 2.758648431214803,
      "grad_norm": 0.25411832706798865,
      "learning_rate": 1.958764250447981e-07,
      "loss": 0.0173,
      "step": 10287
    },
    {
      "epoch": 2.7589165996245644,
      "grad_norm": 0.28624250104542454,
      "learning_rate": 1.9544424255226735e-07,
      "loss": 0.0166,
      "step": 10288
    },
    {
      "epoch": 2.7591847680343253,
      "grad_norm": 0.2241679073275532,
      "learning_rate": 1.9501252786657376e-07,
      "loss": 0.0114,
      "step": 10289
    },
    {
      "epoch": 2.7594529364440867,
      "grad_norm": 0.2624428034302116,
      "learning_rate": 1.945812810297537e-07,
      "loss": 0.0132,
      "step": 10290
    },
    {
      "epoch": 2.759721104853848,
      "grad_norm": 0.2370167165045738,
      "learning_rate": 1.9415050208379638e-07,
      "loss": 0.0138,
      "step": 10291
    },
    {
      "epoch": 2.7599892732636095,
      "grad_norm": 0.23993197510487468,
      "learning_rate": 1.9372019107064487e-07,
      "loss": 0.0141,
      "step": 10292
    },
    {
      "epoch": 2.760257441673371,
      "grad_norm": 0.19245829680014842,
      "learning_rate": 1.9329034803219903e-07,
      "loss": 0.0099,
      "step": 10293
    },
    {
      "epoch": 2.7605256100831324,
      "grad_norm": 0.23027509944270383,
      "learning_rate": 1.9286097301030927e-07,
      "loss": 0.0123,
      "step": 10294
    },
    {
      "epoch": 2.7607937784928938,
      "grad_norm": 0.2075698026182642,
      "learning_rate": 1.9243206604678487e-07,
      "loss": 0.0108,
      "step": 10295
    },
    {
      "epoch": 2.7610619469026547,
      "grad_norm": 0.17967105219804216,
      "learning_rate": 1.9200362718338528e-07,
      "loss": 0.0085,
      "step": 10296
    },
    {
      "epoch": 2.761330115312416,
      "grad_norm": 0.17285989310868857,
      "learning_rate": 1.915756564618282e-07,
      "loss": 0.0095,
      "step": 10297
    },
    {
      "epoch": 2.7615982837221775,
      "grad_norm": 0.2515371419058884,
      "learning_rate": 1.9114815392378362e-07,
      "loss": 0.0141,
      "step": 10298
    },
    {
      "epoch": 2.761866452131939,
      "grad_norm": 0.3648060177137761,
      "learning_rate": 1.9072111961087546e-07,
      "loss": 0.0265,
      "step": 10299
    },
    {
      "epoch": 2.7621346205417003,
      "grad_norm": 0.2574701574198003,
      "learning_rate": 1.902945535646833e-07,
      "loss": 0.0153,
      "step": 10300
    },
    {
      "epoch": 2.7624027889514613,
      "grad_norm": 0.2412810247703087,
      "learning_rate": 1.8986845582674164e-07,
      "loss": 0.0132,
      "step": 10301
    },
    {
      "epoch": 2.7626709573612227,
      "grad_norm": 0.20926607655787768,
      "learning_rate": 1.894428264385373e-07,
      "loss": 0.0099,
      "step": 10302
    },
    {
      "epoch": 2.762939125770984,
      "grad_norm": 0.1819578201015022,
      "learning_rate": 1.890176654415138e-07,
      "loss": 0.0094,
      "step": 10303
    },
    {
      "epoch": 2.7632072941807455,
      "grad_norm": 0.2701035941216311,
      "learning_rate": 1.8859297287706636e-07,
      "loss": 0.0136,
      "step": 10304
    },
    {
      "epoch": 2.763475462590507,
      "grad_norm": 0.20569477944370845,
      "learning_rate": 1.88168748786548e-07,
      "loss": 0.0102,
      "step": 10305
    },
    {
      "epoch": 2.7637436310002683,
      "grad_norm": 0.23322955056802552,
      "learning_rate": 1.8774499321126293e-07,
      "loss": 0.01,
      "step": 10306
    },
    {
      "epoch": 2.7640117994100297,
      "grad_norm": 0.23871855309907736,
      "learning_rate": 1.8732170619247093e-07,
      "loss": 0.0191,
      "step": 10307
    },
    {
      "epoch": 2.7642799678197907,
      "grad_norm": 0.2149647640690599,
      "learning_rate": 1.8689888777138843e-07,
      "loss": 0.015,
      "step": 10308
    },
    {
      "epoch": 2.764548136229552,
      "grad_norm": 0.3155170795165717,
      "learning_rate": 1.864765379891814e-07,
      "loss": 0.0124,
      "step": 10309
    },
    {
      "epoch": 2.7648163046393135,
      "grad_norm": 0.21049758398944282,
      "learning_rate": 1.8605465688697532e-07,
      "loss": 0.0105,
      "step": 10310
    },
    {
      "epoch": 2.765084473049075,
      "grad_norm": 0.2863906521356286,
      "learning_rate": 1.856332445058462e-07,
      "loss": 0.0164,
      "step": 10311
    },
    {
      "epoch": 2.7653526414588363,
      "grad_norm": 0.2270089325208188,
      "learning_rate": 1.8521230088682564e-07,
      "loss": 0.0127,
      "step": 10312
    },
    {
      "epoch": 2.7656208098685973,
      "grad_norm": 0.27355287480396784,
      "learning_rate": 1.8479182607090085e-07,
      "loss": 0.0147,
      "step": 10313
    },
    {
      "epoch": 2.7658889782783587,
      "grad_norm": 0.17183495368980595,
      "learning_rate": 1.8437182009901134e-07,
      "loss": 0.0077,
      "step": 10314
    },
    {
      "epoch": 2.76615714668812,
      "grad_norm": 0.5598178295342714,
      "learning_rate": 1.8395228301205325e-07,
      "loss": 0.0142,
      "step": 10315
    },
    {
      "epoch": 2.7664253150978815,
      "grad_norm": 0.43847833037313527,
      "learning_rate": 1.8353321485087394e-07,
      "loss": 0.0163,
      "step": 10316
    },
    {
      "epoch": 2.766693483507643,
      "grad_norm": 0.21230966128304207,
      "learning_rate": 1.8311461565627852e-07,
      "loss": 0.0119,
      "step": 10317
    },
    {
      "epoch": 2.7669616519174043,
      "grad_norm": 0.28864930293786856,
      "learning_rate": 1.826964854690244e-07,
      "loss": 0.0129,
      "step": 10318
    },
    {
      "epoch": 2.7672298203271657,
      "grad_norm": 0.2991216039914053,
      "learning_rate": 1.8227882432982347e-07,
      "loss": 0.0146,
      "step": 10319
    },
    {
      "epoch": 2.7674979887369267,
      "grad_norm": 0.2704800470352621,
      "learning_rate": 1.8186163227934316e-07,
      "loss": 0.0143,
      "step": 10320
    },
    {
      "epoch": 2.767766157146688,
      "grad_norm": 0.204038761943951,
      "learning_rate": 1.814449093582038e-07,
      "loss": 0.0098,
      "step": 10321
    },
    {
      "epoch": 2.7680343255564495,
      "grad_norm": 0.3319624950140885,
      "learning_rate": 1.8102865560698068e-07,
      "loss": 0.0122,
      "step": 10322
    },
    {
      "epoch": 2.768302493966211,
      "grad_norm": 0.3257821344791158,
      "learning_rate": 1.8061287106620308e-07,
      "loss": 0.0181,
      "step": 10323
    },
    {
      "epoch": 2.7685706623759723,
      "grad_norm": 0.20001357546567808,
      "learning_rate": 1.8019755577635468e-07,
      "loss": 0.0109,
      "step": 10324
    },
    {
      "epoch": 2.7688388307857332,
      "grad_norm": 0.23585049574639466,
      "learning_rate": 1.7978270977787426e-07,
      "loss": 0.01,
      "step": 10325
    },
    {
      "epoch": 2.7691069991954946,
      "grad_norm": 0.20227424915313286,
      "learning_rate": 1.7936833311115398e-07,
      "loss": 0.0094,
      "step": 10326
    },
    {
      "epoch": 2.769375167605256,
      "grad_norm": 0.2320709775885682,
      "learning_rate": 1.7895442581653988e-07,
      "loss": 0.0118,
      "step": 10327
    },
    {
      "epoch": 2.7696433360150174,
      "grad_norm": 0.19016772083065708,
      "learning_rate": 1.7854098793433472e-07,
      "loss": 0.0078,
      "step": 10328
    },
    {
      "epoch": 2.769911504424779,
      "grad_norm": 0.25879980862355717,
      "learning_rate": 1.7812801950479186e-07,
      "loss": 0.0111,
      "step": 10329
    },
    {
      "epoch": 2.7701796728345403,
      "grad_norm": 0.22364016472566298,
      "learning_rate": 1.7771552056812303e-07,
      "loss": 0.0112,
      "step": 10330
    },
    {
      "epoch": 2.7704478412443017,
      "grad_norm": 0.24812615777637864,
      "learning_rate": 1.7730349116448998e-07,
      "loss": 0.0133,
      "step": 10331
    },
    {
      "epoch": 2.7707160096540626,
      "grad_norm": 0.32595556281030735,
      "learning_rate": 1.7689193133401284e-07,
      "loss": 0.0201,
      "step": 10332
    },
    {
      "epoch": 2.770984178063824,
      "grad_norm": 0.335697145560089,
      "learning_rate": 1.764808411167629e-07,
      "loss": 0.0201,
      "step": 10333
    },
    {
      "epoch": 2.7712523464735854,
      "grad_norm": 0.23592271550726876,
      "learning_rate": 1.7607022055276758e-07,
      "loss": 0.0101,
      "step": 10334
    },
    {
      "epoch": 2.771520514883347,
      "grad_norm": 0.24657334384179988,
      "learning_rate": 1.7566006968200712e-07,
      "loss": 0.0118,
      "step": 10335
    },
    {
      "epoch": 2.7717886832931082,
      "grad_norm": 0.24704361849458073,
      "learning_rate": 1.7525038854441734e-07,
      "loss": 0.0177,
      "step": 10336
    },
    {
      "epoch": 2.772056851702869,
      "grad_norm": 0.265406988641684,
      "learning_rate": 1.7484117717988746e-07,
      "loss": 0.0135,
      "step": 10337
    },
    {
      "epoch": 2.7723250201126306,
      "grad_norm": 0.181827170597123,
      "learning_rate": 1.7443243562826228e-07,
      "loss": 0.0082,
      "step": 10338
    },
    {
      "epoch": 2.772593188522392,
      "grad_norm": 0.2827419545725674,
      "learning_rate": 1.7402416392933885e-07,
      "loss": 0.0112,
      "step": 10339
    },
    {
      "epoch": 2.7728613569321534,
      "grad_norm": 0.29644413744040143,
      "learning_rate": 1.7361636212287036e-07,
      "loss": 0.0124,
      "step": 10340
    },
    {
      "epoch": 2.773129525341915,
      "grad_norm": 0.2218596425328855,
      "learning_rate": 1.7320903024856227e-07,
      "loss": 0.013,
      "step": 10341
    },
    {
      "epoch": 2.773397693751676,
      "grad_norm": 0.21982000024457082,
      "learning_rate": 1.728021683460762e-07,
      "loss": 0.0117,
      "step": 10342
    },
    {
      "epoch": 2.7736658621614376,
      "grad_norm": 0.2526653705505683,
      "learning_rate": 1.7239577645502714e-07,
      "loss": 0.0112,
      "step": 10343
    },
    {
      "epoch": 2.7739340305711986,
      "grad_norm": 0.20579035471585921,
      "learning_rate": 1.7198985461498398e-07,
      "loss": 0.0105,
      "step": 10344
    },
    {
      "epoch": 2.77420219898096,
      "grad_norm": 0.18570489694528705,
      "learning_rate": 1.7158440286547117e-07,
      "loss": 0.0097,
      "step": 10345
    },
    {
      "epoch": 2.7744703673907214,
      "grad_norm": 0.24743635872723702,
      "learning_rate": 1.7117942124596555e-07,
      "loss": 0.0102,
      "step": 10346
    },
    {
      "epoch": 2.774738535800483,
      "grad_norm": 0.27089635845612203,
      "learning_rate": 1.7077490979589996e-07,
      "loss": 0.016,
      "step": 10347
    },
    {
      "epoch": 2.775006704210244,
      "grad_norm": 0.2658242080535775,
      "learning_rate": 1.7037086855465902e-07,
      "loss": 0.0147,
      "step": 10348
    },
    {
      "epoch": 2.775274872620005,
      "grad_norm": 0.2420758360240541,
      "learning_rate": 1.6996729756158403e-07,
      "loss": 0.0112,
      "step": 10349
    },
    {
      "epoch": 2.7755430410297666,
      "grad_norm": 0.1666718013045961,
      "learning_rate": 1.695641968559708e-07,
      "loss": 0.0122,
      "step": 10350
    },
    {
      "epoch": 2.775811209439528,
      "grad_norm": 0.24409090648180884,
      "learning_rate": 1.6916156647706683e-07,
      "loss": 0.0173,
      "step": 10351
    },
    {
      "epoch": 2.7760793778492894,
      "grad_norm": 0.23448958354029764,
      "learning_rate": 1.687594064640752e-07,
      "loss": 0.0103,
      "step": 10352
    },
    {
      "epoch": 2.776347546259051,
      "grad_norm": 0.351794842035431,
      "learning_rate": 1.6835771685615343e-07,
      "loss": 0.0256,
      "step": 10353
    },
    {
      "epoch": 2.7766157146688117,
      "grad_norm": 0.2847824787892679,
      "learning_rate": 1.6795649769241307e-07,
      "loss": 0.0152,
      "step": 10354
    },
    {
      "epoch": 2.7768838830785736,
      "grad_norm": 0.21984114751109476,
      "learning_rate": 1.6755574901192007e-07,
      "loss": 0.0121,
      "step": 10355
    },
    {
      "epoch": 2.7771520514883345,
      "grad_norm": 0.2768227547924538,
      "learning_rate": 1.6715547085369378e-07,
      "loss": 0.0149,
      "step": 10356
    },
    {
      "epoch": 2.777420219898096,
      "grad_norm": 0.27439159070393127,
      "learning_rate": 1.6675566325670856e-07,
      "loss": 0.0172,
      "step": 10357
    },
    {
      "epoch": 2.7776883883078574,
      "grad_norm": 0.31181356151709716,
      "learning_rate": 1.6635632625989163e-07,
      "loss": 0.0136,
      "step": 10358
    },
    {
      "epoch": 2.7779565567176188,
      "grad_norm": 0.24751533358959624,
      "learning_rate": 1.6595745990212686e-07,
      "loss": 0.0095,
      "step": 10359
    },
    {
      "epoch": 2.77822472512738,
      "grad_norm": 0.18923389868414806,
      "learning_rate": 1.6555906422225043e-07,
      "loss": 0.0102,
      "step": 10360
    },
    {
      "epoch": 2.778492893537141,
      "grad_norm": 0.24941852697524652,
      "learning_rate": 1.6516113925905185e-07,
      "loss": 0.0125,
      "step": 10361
    },
    {
      "epoch": 2.7787610619469025,
      "grad_norm": 0.35410922350237184,
      "learning_rate": 1.647636850512774e-07,
      "loss": 0.0288,
      "step": 10362
    },
    {
      "epoch": 2.779029230356664,
      "grad_norm": 0.2262450056972405,
      "learning_rate": 1.6436670163762492e-07,
      "loss": 0.0118,
      "step": 10363
    },
    {
      "epoch": 2.7792973987664253,
      "grad_norm": 0.29306291273751933,
      "learning_rate": 1.6397018905674912e-07,
      "loss": 0.0147,
      "step": 10364
    },
    {
      "epoch": 2.7795655671761867,
      "grad_norm": 0.32147155883807826,
      "learning_rate": 1.635741473472563e-07,
      "loss": 0.0145,
      "step": 10365
    },
    {
      "epoch": 2.7798337355859477,
      "grad_norm": 0.2551642323049274,
      "learning_rate": 1.6317857654770787e-07,
      "loss": 0.011,
      "step": 10366
    },
    {
      "epoch": 2.7801019039957096,
      "grad_norm": 0.3308911839901085,
      "learning_rate": 1.6278347669662077e-07,
      "loss": 0.0187,
      "step": 10367
    },
    {
      "epoch": 2.7803700724054705,
      "grad_norm": 0.3146301942924829,
      "learning_rate": 1.6238884783246312e-07,
      "loss": 0.0187,
      "step": 10368
    },
    {
      "epoch": 2.780638240815232,
      "grad_norm": 0.3258296054709916,
      "learning_rate": 1.6199468999366086e-07,
      "loss": 0.0263,
      "step": 10369
    },
    {
      "epoch": 2.7809064092249933,
      "grad_norm": 0.203573694491409,
      "learning_rate": 1.6160100321858996e-07,
      "loss": 0.0131,
      "step": 10370
    },
    {
      "epoch": 2.7811745776347547,
      "grad_norm": 0.30626492319516063,
      "learning_rate": 1.6120778754558418e-07,
      "loss": 0.0107,
      "step": 10371
    },
    {
      "epoch": 2.781442746044516,
      "grad_norm": 0.17102970758164596,
      "learning_rate": 1.608150430129296e-07,
      "loss": 0.0088,
      "step": 10372
    },
    {
      "epoch": 2.781710914454277,
      "grad_norm": 0.260349658145614,
      "learning_rate": 1.604227696588656e-07,
      "loss": 0.0154,
      "step": 10373
    },
    {
      "epoch": 2.7819790828640385,
      "grad_norm": 0.27792940635653424,
      "learning_rate": 1.6003096752158886e-07,
      "loss": 0.0154,
      "step": 10374
    },
    {
      "epoch": 2.7822472512738,
      "grad_norm": 0.20612069696434848,
      "learning_rate": 1.596396366392461e-07,
      "loss": 0.0143,
      "step": 10375
    },
    {
      "epoch": 2.7825154196835613,
      "grad_norm": 0.22436612925722413,
      "learning_rate": 1.5924877704994068e-07,
      "loss": 0.0125,
      "step": 10376
    },
    {
      "epoch": 2.7827835880933227,
      "grad_norm": 0.21744654798322047,
      "learning_rate": 1.588583887917311e-07,
      "loss": 0.0122,
      "step": 10377
    },
    {
      "epoch": 2.7830517565030837,
      "grad_norm": 0.3061791582866806,
      "learning_rate": 1.584684719026258e-07,
      "loss": 0.0132,
      "step": 10378
    },
    {
      "epoch": 2.7833199249128455,
      "grad_norm": 0.243636010690925,
      "learning_rate": 1.5807902642059282e-07,
      "loss": 0.0096,
      "step": 10379
    },
    {
      "epoch": 2.7835880933226065,
      "grad_norm": 0.2261238031706784,
      "learning_rate": 1.5769005238354894e-07,
      "loss": 0.0147,
      "step": 10380
    },
    {
      "epoch": 2.783856261732368,
      "grad_norm": 0.2553476864221084,
      "learning_rate": 1.573015498293684e-07,
      "loss": 0.0222,
      "step": 10381
    },
    {
      "epoch": 2.7841244301421293,
      "grad_norm": 0.6847567360528842,
      "learning_rate": 1.569135187958798e-07,
      "loss": 0.0154,
      "step": 10382
    },
    {
      "epoch": 2.7843925985518907,
      "grad_norm": 0.2406006914957366,
      "learning_rate": 1.5652595932086346e-07,
      "loss": 0.0115,
      "step": 10383
    },
    {
      "epoch": 2.784660766961652,
      "grad_norm": 0.1809979877173751,
      "learning_rate": 1.561388714420553e-07,
      "loss": 0.0098,
      "step": 10384
    },
    {
      "epoch": 2.784928935371413,
      "grad_norm": 0.267644877138911,
      "learning_rate": 1.5575225519714464e-07,
      "loss": 0.0101,
      "step": 10385
    },
    {
      "epoch": 2.7851971037811745,
      "grad_norm": 0.2444677341742721,
      "learning_rate": 1.5536611062377583e-07,
      "loss": 0.0094,
      "step": 10386
    },
    {
      "epoch": 2.785465272190936,
      "grad_norm": 0.2893272278301197,
      "learning_rate": 1.5498043775954707e-07,
      "loss": 0.0149,
      "step": 10387
    },
    {
      "epoch": 2.7857334406006973,
      "grad_norm": 0.2326711247304831,
      "learning_rate": 1.5459523664200894e-07,
      "loss": 0.0158,
      "step": 10388
    },
    {
      "epoch": 2.7860016090104587,
      "grad_norm": 0.5013544426240463,
      "learning_rate": 1.5421050730866914e-07,
      "loss": 0.0156,
      "step": 10389
    },
    {
      "epoch": 2.7862697774202196,
      "grad_norm": 0.19445489988059045,
      "learning_rate": 1.5382624979698613e-07,
      "loss": 0.0107,
      "step": 10390
    },
    {
      "epoch": 2.7865379458299815,
      "grad_norm": 0.23395153437828692,
      "learning_rate": 1.5344246414437546e-07,
      "loss": 0.0125,
      "step": 10391
    },
    {
      "epoch": 2.7868061142397424,
      "grad_norm": 0.22559644842053048,
      "learning_rate": 1.5305915038820397e-07,
      "loss": 0.0138,
      "step": 10392
    },
    {
      "epoch": 2.787074282649504,
      "grad_norm": 0.22746131041139123,
      "learning_rate": 1.5267630856579508e-07,
      "loss": 0.0085,
      "step": 10393
    },
    {
      "epoch": 2.7873424510592653,
      "grad_norm": 0.27648221746913276,
      "learning_rate": 1.522939387144251e-07,
      "loss": 0.0124,
      "step": 10394
    },
    {
      "epoch": 2.7876106194690267,
      "grad_norm": 0.19119714348072736,
      "learning_rate": 1.519120408713237e-07,
      "loss": 0.0094,
      "step": 10395
    },
    {
      "epoch": 2.787878787878788,
      "grad_norm": 0.22126481607399828,
      "learning_rate": 1.515306150736745e-07,
      "loss": 0.0124,
      "step": 10396
    },
    {
      "epoch": 2.788146956288549,
      "grad_norm": 0.2074502966363226,
      "learning_rate": 1.511496613586183e-07,
      "loss": 0.0092,
      "step": 10397
    },
    {
      "epoch": 2.7884151246983104,
      "grad_norm": 0.2194114895785131,
      "learning_rate": 1.507691797632449e-07,
      "loss": 0.0134,
      "step": 10398
    },
    {
      "epoch": 2.788683293108072,
      "grad_norm": 0.18321003190076166,
      "learning_rate": 1.5038917032460298e-07,
      "loss": 0.0078,
      "step": 10399
    },
    {
      "epoch": 2.7889514615178332,
      "grad_norm": 0.2879240989180353,
      "learning_rate": 1.500096330796913e-07,
      "loss": 0.0199,
      "step": 10400
    },
    {
      "epoch": 2.7892196299275946,
      "grad_norm": 0.2804627396054786,
      "learning_rate": 1.4963056806546582e-07,
      "loss": 0.0169,
      "step": 10401
    },
    {
      "epoch": 2.7894877983373556,
      "grad_norm": 0.2515257040649049,
      "learning_rate": 1.4925197531883374e-07,
      "loss": 0.0133,
      "step": 10402
    },
    {
      "epoch": 2.7897559667471175,
      "grad_norm": 0.28994729813499437,
      "learning_rate": 1.4887385487665828e-07,
      "loss": 0.0126,
      "step": 10403
    },
    {
      "epoch": 2.7900241351568784,
      "grad_norm": 0.20802634294868103,
      "learning_rate": 1.484962067757567e-07,
      "loss": 0.0121,
      "step": 10404
    },
    {
      "epoch": 2.79029230356664,
      "grad_norm": 0.24242197527195733,
      "learning_rate": 1.4811903105289904e-07,
      "loss": 0.0112,
      "step": 10405
    },
    {
      "epoch": 2.7905604719764012,
      "grad_norm": 0.2480661374254817,
      "learning_rate": 1.4774232774480978e-07,
      "loss": 0.0117,
      "step": 10406
    },
    {
      "epoch": 2.7908286403861626,
      "grad_norm": 0.24038377504023176,
      "learning_rate": 1.4736609688816738e-07,
      "loss": 0.0168,
      "step": 10407
    },
    {
      "epoch": 2.791096808795924,
      "grad_norm": 0.2560227372163319,
      "learning_rate": 1.469903385196042e-07,
      "loss": 0.0141,
      "step": 10408
    },
    {
      "epoch": 2.791364977205685,
      "grad_norm": 0.22674291457777546,
      "learning_rate": 1.466150526757082e-07,
      "loss": 0.0108,
      "step": 10409
    },
    {
      "epoch": 2.7916331456154464,
      "grad_norm": 0.2887108267352762,
      "learning_rate": 1.4624023939301845e-07,
      "loss": 0.0135,
      "step": 10410
    },
    {
      "epoch": 2.791901314025208,
      "grad_norm": 0.2622124591864205,
      "learning_rate": 1.4586589870803025e-07,
      "loss": 0.0129,
      "step": 10411
    },
    {
      "epoch": 2.792169482434969,
      "grad_norm": 0.2562398963460012,
      "learning_rate": 1.4549203065719163e-07,
      "loss": 0.0098,
      "step": 10412
    },
    {
      "epoch": 2.7924376508447306,
      "grad_norm": 0.30412303316149014,
      "learning_rate": 1.4511863527690573e-07,
      "loss": 0.0142,
      "step": 10413
    },
    {
      "epoch": 2.7927058192544916,
      "grad_norm": 0.31076244578127227,
      "learning_rate": 1.4474571260352898e-07,
      "loss": 0.0212,
      "step": 10414
    },
    {
      "epoch": 2.7929739876642534,
      "grad_norm": 0.3657686841788158,
      "learning_rate": 1.443732626733718e-07,
      "loss": 0.0211,
      "step": 10415
    },
    {
      "epoch": 2.7932421560740144,
      "grad_norm": 0.24214448364605679,
      "learning_rate": 1.4400128552269855e-07,
      "loss": 0.0107,
      "step": 10416
    },
    {
      "epoch": 2.793510324483776,
      "grad_norm": 0.23584827647361203,
      "learning_rate": 1.436297811877274e-07,
      "loss": 0.0115,
      "step": 10417
    },
    {
      "epoch": 2.793778492893537,
      "grad_norm": 0.28820192373450965,
      "learning_rate": 1.4325874970463172e-07,
      "loss": 0.015,
      "step": 10418
    },
    {
      "epoch": 2.7940466613032986,
      "grad_norm": 0.2237656459835972,
      "learning_rate": 1.42888191109537e-07,
      "loss": 0.0109,
      "step": 10419
    },
    {
      "epoch": 2.79431482971306,
      "grad_norm": 0.24680954567322236,
      "learning_rate": 1.425181054385233e-07,
      "loss": 0.012,
      "step": 10420
    },
    {
      "epoch": 2.794582998122821,
      "grad_norm": 0.232064208063594,
      "learning_rate": 1.4214849272762565e-07,
      "loss": 0.0105,
      "step": 10421
    },
    {
      "epoch": 2.7948511665325824,
      "grad_norm": 0.2785547264126087,
      "learning_rate": 1.417793530128314e-07,
      "loss": 0.0184,
      "step": 10422
    },
    {
      "epoch": 2.7951193349423438,
      "grad_norm": 0.26641413334934144,
      "learning_rate": 1.414106863300835e-07,
      "loss": 0.0162,
      "step": 10423
    },
    {
      "epoch": 2.795387503352105,
      "grad_norm": 0.21820182422666062,
      "learning_rate": 1.4104249271527815e-07,
      "loss": 0.0167,
      "step": 10424
    },
    {
      "epoch": 2.7956556717618666,
      "grad_norm": 0.2559928207004348,
      "learning_rate": 1.406747722042645e-07,
      "loss": 0.0155,
      "step": 10425
    },
    {
      "epoch": 2.7959238401716275,
      "grad_norm": 0.2977755717115644,
      "learning_rate": 1.4030752483284727e-07,
      "loss": 0.0178,
      "step": 10426
    },
    {
      "epoch": 2.7961920085813894,
      "grad_norm": 0.2220591626514631,
      "learning_rate": 1.3994075063678393e-07,
      "loss": 0.0097,
      "step": 10427
    },
    {
      "epoch": 2.7964601769911503,
      "grad_norm": 0.19230928808945955,
      "learning_rate": 1.3957444965178703e-07,
      "loss": 0.0086,
      "step": 10428
    },
    {
      "epoch": 2.7967283454009118,
      "grad_norm": 0.9443152836682657,
      "learning_rate": 1.392086219135208e-07,
      "loss": 0.0083,
      "step": 10429
    },
    {
      "epoch": 2.796996513810673,
      "grad_norm": 0.21101384917978036,
      "learning_rate": 1.3884326745760622e-07,
      "loss": 0.0123,
      "step": 10430
    },
    {
      "epoch": 2.7972646822204346,
      "grad_norm": 0.22971679317096116,
      "learning_rate": 1.3847838631961764e-07,
      "loss": 0.013,
      "step": 10431
    },
    {
      "epoch": 2.797532850630196,
      "grad_norm": 0.28036280132388014,
      "learning_rate": 1.3811397853508047e-07,
      "loss": 0.0106,
      "step": 10432
    },
    {
      "epoch": 2.797801019039957,
      "grad_norm": 0.2576678558789783,
      "learning_rate": 1.377500441394769e-07,
      "loss": 0.0133,
      "step": 10433
    },
    {
      "epoch": 2.7980691874497183,
      "grad_norm": 0.4126172021834964,
      "learning_rate": 1.373865831682436e-07,
      "loss": 0.0333,
      "step": 10434
    },
    {
      "epoch": 2.7983373558594797,
      "grad_norm": 0.32729245952100927,
      "learning_rate": 1.370235956567678e-07,
      "loss": 0.0374,
      "step": 10435
    },
    {
      "epoch": 2.798605524269241,
      "grad_norm": 0.3309749441799219,
      "learning_rate": 1.3666108164039406e-07,
      "loss": 0.0167,
      "step": 10436
    },
    {
      "epoch": 2.7988736926790025,
      "grad_norm": 0.2787860629229127,
      "learning_rate": 1.362990411544185e-07,
      "loss": 0.0148,
      "step": 10437
    },
    {
      "epoch": 2.7991418610887635,
      "grad_norm": 0.2478465003549631,
      "learning_rate": 1.35937474234093e-07,
      "loss": 0.0133,
      "step": 10438
    },
    {
      "epoch": 2.799410029498525,
      "grad_norm": 0.2194953338305714,
      "learning_rate": 1.3557638091462099e-07,
      "loss": 0.0096,
      "step": 10439
    },
    {
      "epoch": 2.7996781979082863,
      "grad_norm": 0.2449033353376612,
      "learning_rate": 1.352157612311622e-07,
      "loss": 0.0139,
      "step": 10440
    },
    {
      "epoch": 2.7999463663180477,
      "grad_norm": 0.28415259940058196,
      "learning_rate": 1.3485561521882906e-07,
      "loss": 0.0145,
      "step": 10441
    },
    {
      "epoch": 2.800214534727809,
      "grad_norm": 0.2637487643923356,
      "learning_rate": 1.3449594291268797e-07,
      "loss": 0.0171,
      "step": 10442
    },
    {
      "epoch": 2.8004827031375705,
      "grad_norm": 0.33749852024804883,
      "learning_rate": 1.341367443477598e-07,
      "loss": 0.0139,
      "step": 10443
    },
    {
      "epoch": 2.800750871547332,
      "grad_norm": 0.20998618750802264,
      "learning_rate": 1.337780195590177e-07,
      "loss": 0.0105,
      "step": 10444
    },
    {
      "epoch": 2.801019039957093,
      "grad_norm": 0.23861127514112854,
      "learning_rate": 1.3341976858139038e-07,
      "loss": 0.0173,
      "step": 10445
    },
    {
      "epoch": 2.8012872083668543,
      "grad_norm": 0.2455684868640165,
      "learning_rate": 1.330619914497594e-07,
      "loss": 0.0111,
      "step": 10446
    },
    {
      "epoch": 2.8015553767766157,
      "grad_norm": 0.25615967860042893,
      "learning_rate": 1.3270468819896076e-07,
      "loss": 0.0112,
      "step": 10447
    },
    {
      "epoch": 2.801823545186377,
      "grad_norm": 0.2703378774023744,
      "learning_rate": 1.3234785886378499e-07,
      "loss": 0.0155,
      "step": 10448
    },
    {
      "epoch": 2.8020917135961385,
      "grad_norm": 0.26591268382646427,
      "learning_rate": 1.3199150347897427e-07,
      "loss": 0.0164,
      "step": 10449
    },
    {
      "epoch": 2.8023598820058995,
      "grad_norm": 0.2619501554580531,
      "learning_rate": 1.316356220792264e-07,
      "loss": 0.0152,
      "step": 10450
    },
    {
      "epoch": 2.802628050415661,
      "grad_norm": 0.22985707092005225,
      "learning_rate": 1.3128021469919317e-07,
      "loss": 0.0117,
      "step": 10451
    },
    {
      "epoch": 2.8028962188254223,
      "grad_norm": 0.2450159160562203,
      "learning_rate": 1.3092528137347848e-07,
      "loss": 0.011,
      "step": 10452
    },
    {
      "epoch": 2.8031643872351837,
      "grad_norm": 0.32912648627599583,
      "learning_rate": 1.3057082213664307e-07,
      "loss": 0.0137,
      "step": 10453
    },
    {
      "epoch": 2.803432555644945,
      "grad_norm": 0.2278205689066198,
      "learning_rate": 1.3021683702319766e-07,
      "loss": 0.0138,
      "step": 10454
    },
    {
      "epoch": 2.8037007240547065,
      "grad_norm": 0.253014305577512,
      "learning_rate": 1.2986332606761077e-07,
      "loss": 0.017,
      "step": 10455
    },
    {
      "epoch": 2.803968892464468,
      "grad_norm": 0.42745883173923505,
      "learning_rate": 1.2951028930430155e-07,
      "loss": 0.0166,
      "step": 10456
    },
    {
      "epoch": 2.804237060874229,
      "grad_norm": 0.325634582241657,
      "learning_rate": 1.2915772676764416e-07,
      "loss": 0.0173,
      "step": 10457
    },
    {
      "epoch": 2.8045052292839903,
      "grad_norm": 0.3077726016166347,
      "learning_rate": 1.2880563849196725e-07,
      "loss": 0.0178,
      "step": 10458
    },
    {
      "epoch": 2.8047733976937517,
      "grad_norm": 0.2282917340907862,
      "learning_rate": 1.2845402451155231e-07,
      "loss": 0.0117,
      "step": 10459
    },
    {
      "epoch": 2.805041566103513,
      "grad_norm": 0.26700993642872733,
      "learning_rate": 1.2810288486063582e-07,
      "loss": 0.0115,
      "step": 10460
    },
    {
      "epoch": 2.8053097345132745,
      "grad_norm": 0.20885392443741777,
      "learning_rate": 1.2775221957340654e-07,
      "loss": 0.0109,
      "step": 10461
    },
    {
      "epoch": 2.8055779029230354,
      "grad_norm": 0.18612941625924867,
      "learning_rate": 1.2740202868400776e-07,
      "loss": 0.0116,
      "step": 10462
    },
    {
      "epoch": 2.805846071332797,
      "grad_norm": 0.3363810787023966,
      "learning_rate": 1.2705231222653768e-07,
      "loss": 0.0263,
      "step": 10463
    },
    {
      "epoch": 2.8061142397425582,
      "grad_norm": 0.2539166688734785,
      "learning_rate": 1.267030702350458e-07,
      "loss": 0.0147,
      "step": 10464
    },
    {
      "epoch": 2.8063824081523197,
      "grad_norm": 0.22304391928935519,
      "learning_rate": 1.263543027435382e-07,
      "loss": 0.0092,
      "step": 10465
    },
    {
      "epoch": 2.806650576562081,
      "grad_norm": 0.2653321377312349,
      "learning_rate": 1.2600600978597222e-07,
      "loss": 0.0191,
      "step": 10466
    },
    {
      "epoch": 2.8069187449718425,
      "grad_norm": 0.2287778530073877,
      "learning_rate": 1.2565819139626123e-07,
      "loss": 0.0139,
      "step": 10467
    },
    {
      "epoch": 2.807186913381604,
      "grad_norm": 0.2986165970589626,
      "learning_rate": 1.253108476082715e-07,
      "loss": 0.0177,
      "step": 10468
    },
    {
      "epoch": 2.807455081791365,
      "grad_norm": 0.2389811203872515,
      "learning_rate": 1.2496397845582152e-07,
      "loss": 0.0172,
      "step": 10469
    },
    {
      "epoch": 2.8077232502011262,
      "grad_norm": 0.2313254011790213,
      "learning_rate": 1.246175839726871e-07,
      "loss": 0.0093,
      "step": 10470
    },
    {
      "epoch": 2.8079914186108876,
      "grad_norm": 0.2540078090779224,
      "learning_rate": 1.242716641925934e-07,
      "loss": 0.0134,
      "step": 10471
    },
    {
      "epoch": 2.808259587020649,
      "grad_norm": 0.23870018947824648,
      "learning_rate": 1.2392621914922354e-07,
      "loss": 0.0121,
      "step": 10472
    },
    {
      "epoch": 2.8085277554304104,
      "grad_norm": 0.20133152690268585,
      "learning_rate": 1.2358124887621225e-07,
      "loss": 0.0081,
      "step": 10473
    },
    {
      "epoch": 2.8087959238401714,
      "grad_norm": 0.2951443022709948,
      "learning_rate": 1.2323675340714768e-07,
      "loss": 0.0155,
      "step": 10474
    },
    {
      "epoch": 2.809064092249933,
      "grad_norm": 0.20949239363459768,
      "learning_rate": 1.2289273277557356e-07,
      "loss": 0.0105,
      "step": 10475
    },
    {
      "epoch": 2.809332260659694,
      "grad_norm": 0.25121939776565294,
      "learning_rate": 1.2254918701498474e-07,
      "loss": 0.0095,
      "step": 10476
    },
    {
      "epoch": 2.8096004290694556,
      "grad_norm": 0.25489599586964196,
      "learning_rate": 1.222061161588328e-07,
      "loss": 0.0142,
      "step": 10477
    },
    {
      "epoch": 2.809868597479217,
      "grad_norm": 0.3713140867020353,
      "learning_rate": 1.218635202405205e-07,
      "loss": 0.0146,
      "step": 10478
    },
    {
      "epoch": 2.8101367658889784,
      "grad_norm": 0.26517108704676373,
      "learning_rate": 1.215213992934061e-07,
      "loss": 0.0121,
      "step": 10479
    },
    {
      "epoch": 2.81040493429874,
      "grad_norm": 0.23486400485530132,
      "learning_rate": 1.211797533508019e-07,
      "loss": 0.0104,
      "step": 10480
    },
    {
      "epoch": 2.810673102708501,
      "grad_norm": 0.22152003535505607,
      "learning_rate": 1.2083858244597124e-07,
      "loss": 0.0121,
      "step": 10481
    },
    {
      "epoch": 2.810941271118262,
      "grad_norm": 0.21521823373193297,
      "learning_rate": 1.2049788661213367e-07,
      "loss": 0.0139,
      "step": 10482
    },
    {
      "epoch": 2.8112094395280236,
      "grad_norm": 0.24934471274133665,
      "learning_rate": 1.2015766588246268e-07,
      "loss": 0.0168,
      "step": 10483
    },
    {
      "epoch": 2.811477607937785,
      "grad_norm": 0.18632973811217737,
      "learning_rate": 1.1981792029008342e-07,
      "loss": 0.011,
      "step": 10484
    },
    {
      "epoch": 2.8117457763475464,
      "grad_norm": 0.24445712819532772,
      "learning_rate": 1.1947864986807722e-07,
      "loss": 0.0133,
      "step": 10485
    },
    {
      "epoch": 2.8120139447573074,
      "grad_norm": 0.3464566346990064,
      "learning_rate": 1.1913985464947652e-07,
      "loss": 0.0162,
      "step": 10486
    },
    {
      "epoch": 2.8122821131670688,
      "grad_norm": 0.2157737623394709,
      "learning_rate": 1.1880153466727107e-07,
      "loss": 0.0086,
      "step": 10487
    },
    {
      "epoch": 2.81255028157683,
      "grad_norm": 0.221391901442086,
      "learning_rate": 1.184636899543995e-07,
      "loss": 0.0111,
      "step": 10488
    },
    {
      "epoch": 2.8128184499865916,
      "grad_norm": 0.3390963824654852,
      "learning_rate": 1.1812632054375883e-07,
      "loss": 0.0166,
      "step": 10489
    },
    {
      "epoch": 2.813086618396353,
      "grad_norm": 0.24287132858084956,
      "learning_rate": 1.1778942646819724e-07,
      "loss": 0.0114,
      "step": 10490
    },
    {
      "epoch": 2.8133547868061144,
      "grad_norm": 0.34294226074080464,
      "learning_rate": 1.1745300776051683e-07,
      "loss": 0.0164,
      "step": 10491
    },
    {
      "epoch": 2.813622955215876,
      "grad_norm": 0.23270887858535036,
      "learning_rate": 1.171170644534747e-07,
      "loss": 0.0089,
      "step": 10492
    },
    {
      "epoch": 2.8138911236256368,
      "grad_norm": 0.23224269257798463,
      "learning_rate": 1.1678159657978027e-07,
      "loss": 0.0144,
      "step": 10493
    },
    {
      "epoch": 2.814159292035398,
      "grad_norm": 0.18104815265243018,
      "learning_rate": 1.164466041720963e-07,
      "loss": 0.0073,
      "step": 10494
    },
    {
      "epoch": 2.8144274604451596,
      "grad_norm": 0.21350395185096221,
      "learning_rate": 1.1611208726304169e-07,
      "loss": 0.0094,
      "step": 10495
    },
    {
      "epoch": 2.814695628854921,
      "grad_norm": 0.2486309338339223,
      "learning_rate": 1.1577804588518593e-07,
      "loss": 0.0174,
      "step": 10496
    },
    {
      "epoch": 2.8149637972646824,
      "grad_norm": 0.28649282558673866,
      "learning_rate": 1.1544448007105524e-07,
      "loss": 0.0117,
      "step": 10497
    },
    {
      "epoch": 2.8152319656744433,
      "grad_norm": 0.2431961914125149,
      "learning_rate": 1.1511138985312698e-07,
      "loss": 0.0089,
      "step": 10498
    },
    {
      "epoch": 2.8155001340842047,
      "grad_norm": 0.3257855778533044,
      "learning_rate": 1.1477877526383296e-07,
      "loss": 0.0126,
      "step": 10499
    },
    {
      "epoch": 2.815768302493966,
      "grad_norm": 0.18654397407433732,
      "learning_rate": 1.1444663633556063e-07,
      "loss": 0.0073,
      "step": 10500
    },
    {
      "epoch": 2.8160364709037276,
      "grad_norm": 0.21329784236664553,
      "learning_rate": 1.1411497310064745e-07,
      "loss": 0.0112,
      "step": 10501
    },
    {
      "epoch": 2.816304639313489,
      "grad_norm": 0.2330606826695789,
      "learning_rate": 1.1378378559138814e-07,
      "loss": 0.0086,
      "step": 10502
    },
    {
      "epoch": 2.8165728077232504,
      "grad_norm": 0.3100593980493565,
      "learning_rate": 1.1345307384002857e-07,
      "loss": 0.024,
      "step": 10503
    },
    {
      "epoch": 2.8168409761330118,
      "grad_norm": 0.2350491979618477,
      "learning_rate": 1.131228378787702e-07,
      "loss": 0.0147,
      "step": 10504
    },
    {
      "epoch": 2.8171091445427727,
      "grad_norm": 0.2775991165155188,
      "learning_rate": 1.1279307773976677e-07,
      "loss": 0.0136,
      "step": 10505
    },
    {
      "epoch": 2.817377312952534,
      "grad_norm": 0.2707360254237531,
      "learning_rate": 1.1246379345512593e-07,
      "loss": 0.0122,
      "step": 10506
    },
    {
      "epoch": 2.8176454813622955,
      "grad_norm": 0.227220094865465,
      "learning_rate": 1.1213498505690924e-07,
      "loss": 0.0086,
      "step": 10507
    },
    {
      "epoch": 2.817913649772057,
      "grad_norm": 0.22817919608034942,
      "learning_rate": 1.1180665257713164e-07,
      "loss": 0.01,
      "step": 10508
    },
    {
      "epoch": 2.8181818181818183,
      "grad_norm": 0.28890061337198,
      "learning_rate": 1.1147879604776258e-07,
      "loss": 0.018,
      "step": 10509
    },
    {
      "epoch": 2.8184499865915793,
      "grad_norm": 0.3070013579283648,
      "learning_rate": 1.1115141550072483e-07,
      "loss": 0.0104,
      "step": 10510
    },
    {
      "epoch": 2.8187181550013407,
      "grad_norm": 0.2435290939246613,
      "learning_rate": 1.1082451096789348e-07,
      "loss": 0.0134,
      "step": 10511
    },
    {
      "epoch": 2.818986323411102,
      "grad_norm": 0.27000591558114306,
      "learning_rate": 1.1049808248109973e-07,
      "loss": 0.0139,
      "step": 10512
    },
    {
      "epoch": 2.8192544918208635,
      "grad_norm": 0.1944386742758072,
      "learning_rate": 1.1017213007212536e-07,
      "loss": 0.0102,
      "step": 10513
    },
    {
      "epoch": 2.819522660230625,
      "grad_norm": 0.2396153728456049,
      "learning_rate": 1.0984665377270942e-07,
      "loss": 0.0097,
      "step": 10514
    },
    {
      "epoch": 2.8197908286403863,
      "grad_norm": 0.18574527659173157,
      "learning_rate": 1.0952165361454103e-07,
      "loss": 0.0096,
      "step": 10515
    },
    {
      "epoch": 2.8200589970501477,
      "grad_norm": 0.27264315456353905,
      "learning_rate": 1.0919712962926543e-07,
      "loss": 0.0178,
      "step": 10516
    },
    {
      "epoch": 2.8203271654599087,
      "grad_norm": 0.23467576324785558,
      "learning_rate": 1.0887308184848122e-07,
      "loss": 0.0166,
      "step": 10517
    },
    {
      "epoch": 2.82059533386967,
      "grad_norm": 0.23237151439835077,
      "learning_rate": 1.0854951030373817e-07,
      "loss": 0.0126,
      "step": 10518
    },
    {
      "epoch": 2.8208635022794315,
      "grad_norm": 0.2265132858856544,
      "learning_rate": 1.082264150265433e-07,
      "loss": 0.012,
      "step": 10519
    },
    {
      "epoch": 2.821131670689193,
      "grad_norm": 0.2392742797338822,
      "learning_rate": 1.0790379604835477e-07,
      "loss": 0.0132,
      "step": 10520
    },
    {
      "epoch": 2.8213998390989543,
      "grad_norm": 0.2795524375002084,
      "learning_rate": 1.0758165340058524e-07,
      "loss": 0.0178,
      "step": 10521
    },
    {
      "epoch": 2.8216680075087153,
      "grad_norm": 0.2661763907993042,
      "learning_rate": 1.0725998711460184e-07,
      "loss": 0.0144,
      "step": 10522
    },
    {
      "epoch": 2.8219361759184767,
      "grad_norm": 0.22888801914804655,
      "learning_rate": 1.0693879722172285e-07,
      "loss": 0.0136,
      "step": 10523
    },
    {
      "epoch": 2.822204344328238,
      "grad_norm": 0.32246337040229756,
      "learning_rate": 1.0661808375322269e-07,
      "loss": 0.0181,
      "step": 10524
    },
    {
      "epoch": 2.8224725127379995,
      "grad_norm": 0.2958465744163367,
      "learning_rate": 1.0629784674032751e-07,
      "loss": 0.0149,
      "step": 10525
    },
    {
      "epoch": 2.822740681147761,
      "grad_norm": 0.24649871652539673,
      "learning_rate": 1.0597808621421846e-07,
      "loss": 0.0169,
      "step": 10526
    },
    {
      "epoch": 2.823008849557522,
      "grad_norm": 0.3134110662921053,
      "learning_rate": 1.0565880220603009e-07,
      "loss": 0.0105,
      "step": 10527
    },
    {
      "epoch": 2.8232770179672837,
      "grad_norm": 0.17443324557004608,
      "learning_rate": 1.053399947468503e-07,
      "loss": 0.008,
      "step": 10528
    },
    {
      "epoch": 2.8235451863770447,
      "grad_norm": 0.2187714883265498,
      "learning_rate": 1.0502166386771929e-07,
      "loss": 0.0199,
      "step": 10529
    },
    {
      "epoch": 2.823813354786806,
      "grad_norm": 0.22886551935926505,
      "learning_rate": 1.0470380959963388e-07,
      "loss": 0.0122,
      "step": 10530
    },
    {
      "epoch": 2.8240815231965675,
      "grad_norm": 0.2484388046265019,
      "learning_rate": 1.0438643197354103e-07,
      "loss": 0.0153,
      "step": 10531
    },
    {
      "epoch": 2.824349691606329,
      "grad_norm": 0.29250340854280743,
      "learning_rate": 1.0406953102034434e-07,
      "loss": 0.0129,
      "step": 10532
    },
    {
      "epoch": 2.8246178600160903,
      "grad_norm": 0.2011810687082208,
      "learning_rate": 1.03753106770898e-07,
      "loss": 0.0092,
      "step": 10533
    },
    {
      "epoch": 2.8248860284258512,
      "grad_norm": 0.25207777330616665,
      "learning_rate": 1.034371592560135e-07,
      "loss": 0.0131,
      "step": 10534
    },
    {
      "epoch": 2.8251541968356126,
      "grad_norm": 0.23081352811138797,
      "learning_rate": 1.0312168850645177e-07,
      "loss": 0.0116,
      "step": 10535
    },
    {
      "epoch": 2.825422365245374,
      "grad_norm": 0.2778166475281058,
      "learning_rate": 1.0280669455293046e-07,
      "loss": 0.0152,
      "step": 10536
    },
    {
      "epoch": 2.8256905336551355,
      "grad_norm": 0.19949269306913872,
      "learning_rate": 1.0249217742612005e-07,
      "loss": 0.0092,
      "step": 10537
    },
    {
      "epoch": 2.825958702064897,
      "grad_norm": 0.3158668509886965,
      "learning_rate": 1.0217813715664327e-07,
      "loss": 0.0226,
      "step": 10538
    },
    {
      "epoch": 2.826226870474658,
      "grad_norm": 0.20996423286752994,
      "learning_rate": 1.0186457377507786e-07,
      "loss": 0.009,
      "step": 10539
    },
    {
      "epoch": 2.8264950388844197,
      "grad_norm": 0.208105971675455,
      "learning_rate": 1.0155148731195497e-07,
      "loss": 0.0084,
      "step": 10540
    },
    {
      "epoch": 2.8267632072941806,
      "grad_norm": 0.2774483154697275,
      "learning_rate": 1.0123887779775798e-07,
      "loss": 0.0125,
      "step": 10541
    },
    {
      "epoch": 2.827031375703942,
      "grad_norm": 0.24656591178592982,
      "learning_rate": 1.009267452629259e-07,
      "loss": 0.0135,
      "step": 10542
    },
    {
      "epoch": 2.8272995441137034,
      "grad_norm": 0.19404473740136532,
      "learning_rate": 1.0061508973784995e-07,
      "loss": 0.0114,
      "step": 10543
    },
    {
      "epoch": 2.827567712523465,
      "grad_norm": 0.23654619646755948,
      "learning_rate": 1.0030391125287476e-07,
      "loss": 0.015,
      "step": 10544
    },
    {
      "epoch": 2.8278358809332262,
      "grad_norm": 0.23251931017808042,
      "learning_rate": 9.999320983829941e-08,
      "loss": 0.0118,
      "step": 10545
    },
    {
      "epoch": 2.828104049342987,
      "grad_norm": 0.22423685520950068,
      "learning_rate": 9.968298552437583e-08,
      "loss": 0.0111,
      "step": 10546
    },
    {
      "epoch": 2.8283722177527486,
      "grad_norm": 0.2283074961901126,
      "learning_rate": 9.937323834130985e-08,
      "loss": 0.0112,
      "step": 10547
    },
    {
      "epoch": 2.82864038616251,
      "grad_norm": 0.1998370102471561,
      "learning_rate": 9.906396831926069e-08,
      "loss": 0.011,
      "step": 10548
    },
    {
      "epoch": 2.8289085545722714,
      "grad_norm": 0.2712724812963372,
      "learning_rate": 9.875517548834201e-08,
      "loss": 0.0151,
      "step": 10549
    },
    {
      "epoch": 2.829176722982033,
      "grad_norm": 0.2497759348634584,
      "learning_rate": 9.844685987861813e-08,
      "loss": 0.0126,
      "step": 10550
    },
    {
      "epoch": 2.829444891391794,
      "grad_norm": 0.2854105222814435,
      "learning_rate": 9.813902152011112e-08,
      "loss": 0.017,
      "step": 10551
    },
    {
      "epoch": 2.8297130598015556,
      "grad_norm": 0.22328515175330568,
      "learning_rate": 9.783166044279313e-08,
      "loss": 0.0151,
      "step": 10552
    },
    {
      "epoch": 2.8299812282113166,
      "grad_norm": 0.27221734064435466,
      "learning_rate": 9.752477667659133e-08,
      "loss": 0.0126,
      "step": 10553
    },
    {
      "epoch": 2.830249396621078,
      "grad_norm": 0.26142095298663853,
      "learning_rate": 9.721837025138625e-08,
      "loss": 0.0131,
      "step": 10554
    },
    {
      "epoch": 2.8305175650308394,
      "grad_norm": 0.29516283162324486,
      "learning_rate": 9.691244119701126e-08,
      "loss": 0.0106,
      "step": 10555
    },
    {
      "epoch": 2.830785733440601,
      "grad_norm": 0.29419935012549886,
      "learning_rate": 9.660698954325531e-08,
      "loss": 0.0151,
      "step": 10556
    },
    {
      "epoch": 2.831053901850362,
      "grad_norm": 0.2644848867063579,
      "learning_rate": 9.630201531985794e-08,
      "loss": 0.0156,
      "step": 10557
    },
    {
      "epoch": 2.831322070260123,
      "grad_norm": 0.3452619118433279,
      "learning_rate": 9.599751855651429e-08,
      "loss": 0.0198,
      "step": 10558
    },
    {
      "epoch": 2.8315902386698846,
      "grad_norm": 0.2911449097193264,
      "learning_rate": 9.569349928287231e-08,
      "loss": 0.0132,
      "step": 10559
    },
    {
      "epoch": 2.831858407079646,
      "grad_norm": 0.2733649681675306,
      "learning_rate": 9.538995752853387e-08,
      "loss": 0.0138,
      "step": 10560
    },
    {
      "epoch": 2.8321265754894074,
      "grad_norm": 0.2712613627377445,
      "learning_rate": 9.508689332305365e-08,
      "loss": 0.0092,
      "step": 10561
    },
    {
      "epoch": 2.832394743899169,
      "grad_norm": 0.3491678286581166,
      "learning_rate": 9.47843066959403e-08,
      "loss": 0.0234,
      "step": 10562
    },
    {
      "epoch": 2.8326629123089297,
      "grad_norm": 0.1939961486760394,
      "learning_rate": 9.448219767665579e-08,
      "loss": 0.0094,
      "step": 10563
    },
    {
      "epoch": 2.8329310807186916,
      "grad_norm": 0.2276638070120413,
      "learning_rate": 9.418056629461658e-08,
      "loss": 0.0103,
      "step": 10564
    },
    {
      "epoch": 2.8331992491284526,
      "grad_norm": 0.24073772279406377,
      "learning_rate": 9.38794125791903e-08,
      "loss": 0.0181,
      "step": 10565
    },
    {
      "epoch": 2.833467417538214,
      "grad_norm": 0.2871623719892741,
      "learning_rate": 9.357873655970018e-08,
      "loss": 0.0121,
      "step": 10566
    },
    {
      "epoch": 2.8337355859479754,
      "grad_norm": 0.2167964775352199,
      "learning_rate": 9.327853826542221e-08,
      "loss": 0.0097,
      "step": 10567
    },
    {
      "epoch": 2.8340037543577368,
      "grad_norm": 0.2450520580095833,
      "learning_rate": 9.297881772558581e-08,
      "loss": 0.0148,
      "step": 10568
    },
    {
      "epoch": 2.834271922767498,
      "grad_norm": 0.246440215240128,
      "learning_rate": 9.267957496937429e-08,
      "loss": 0.0129,
      "step": 10569
    },
    {
      "epoch": 2.834540091177259,
      "grad_norm": 0.28089009870122456,
      "learning_rate": 9.238081002592326e-08,
      "loss": 0.011,
      "step": 10570
    },
    {
      "epoch": 2.8348082595870205,
      "grad_norm": 0.20602307036459055,
      "learning_rate": 9.208252292432385e-08,
      "loss": 0.0096,
      "step": 10571
    },
    {
      "epoch": 2.835076427996782,
      "grad_norm": 0.2528165779942726,
      "learning_rate": 9.178471369361841e-08,
      "loss": 0.0133,
      "step": 10572
    },
    {
      "epoch": 2.8353445964065433,
      "grad_norm": 0.31999369787628684,
      "learning_rate": 9.148738236280541e-08,
      "loss": 0.0235,
      "step": 10573
    },
    {
      "epoch": 2.8356127648163048,
      "grad_norm": 0.20547858624632034,
      "learning_rate": 9.11905289608328e-08,
      "loss": 0.0096,
      "step": 10574
    },
    {
      "epoch": 2.8358809332260657,
      "grad_norm": 0.22721597528761336,
      "learning_rate": 9.089415351660635e-08,
      "loss": 0.0099,
      "step": 10575
    },
    {
      "epoch": 2.8361491016358276,
      "grad_norm": 0.1888936459211925,
      "learning_rate": 9.059825605898354e-08,
      "loss": 0.0086,
      "step": 10576
    },
    {
      "epoch": 2.8364172700455885,
      "grad_norm": 0.20703990127078115,
      "learning_rate": 9.030283661677297e-08,
      "loss": 0.0115,
      "step": 10577
    },
    {
      "epoch": 2.83668543845535,
      "grad_norm": 0.29628996574905,
      "learning_rate": 9.000789521874054e-08,
      "loss": 0.0148,
      "step": 10578
    },
    {
      "epoch": 2.8369536068651113,
      "grad_norm": 0.25111679464742376,
      "learning_rate": 8.971343189360437e-08,
      "loss": 0.0122,
      "step": 10579
    },
    {
      "epoch": 2.8372217752748727,
      "grad_norm": 0.3115563093804111,
      "learning_rate": 8.941944667003377e-08,
      "loss": 0.014,
      "step": 10580
    },
    {
      "epoch": 2.837489943684634,
      "grad_norm": 0.2266011230599328,
      "learning_rate": 8.912593957665527e-08,
      "loss": 0.0096,
      "step": 10581
    },
    {
      "epoch": 2.837758112094395,
      "grad_norm": 0.2559349868780593,
      "learning_rate": 8.883291064204492e-08,
      "loss": 0.0184,
      "step": 10582
    },
    {
      "epoch": 2.8380262805041565,
      "grad_norm": 0.2190602977356901,
      "learning_rate": 8.854035989473597e-08,
      "loss": 0.0121,
      "step": 10583
    },
    {
      "epoch": 2.838294448913918,
      "grad_norm": 0.38782373003621023,
      "learning_rate": 8.824828736321178e-08,
      "loss": 0.021,
      "step": 10584
    },
    {
      "epoch": 2.8385626173236793,
      "grad_norm": 0.2365477169937749,
      "learning_rate": 8.795669307591182e-08,
      "loss": 0.0156,
      "step": 10585
    },
    {
      "epoch": 2.8388307857334407,
      "grad_norm": 0.23297278833207105,
      "learning_rate": 8.76655770612278e-08,
      "loss": 0.0115,
      "step": 10586
    },
    {
      "epoch": 2.8390989541432017,
      "grad_norm": 0.2454591733971592,
      "learning_rate": 8.737493934750374e-08,
      "loss": 0.0088,
      "step": 10587
    },
    {
      "epoch": 2.8393671225529635,
      "grad_norm": 0.3079366078069074,
      "learning_rate": 8.708477996303977e-08,
      "loss": 0.0187,
      "step": 10588
    },
    {
      "epoch": 2.8396352909627245,
      "grad_norm": 0.259339823241484,
      "learning_rate": 8.679509893608717e-08,
      "loss": 0.0122,
      "step": 10589
    },
    {
      "epoch": 2.839903459372486,
      "grad_norm": 0.2650153092394716,
      "learning_rate": 8.650589629485118e-08,
      "loss": 0.0171,
      "step": 10590
    },
    {
      "epoch": 2.8401716277822473,
      "grad_norm": 0.2797122545855559,
      "learning_rate": 8.621717206749092e-08,
      "loss": 0.0147,
      "step": 10591
    },
    {
      "epoch": 2.8404397961920087,
      "grad_norm": 0.24417525891297182,
      "learning_rate": 8.59289262821189e-08,
      "loss": 0.014,
      "step": 10592
    },
    {
      "epoch": 2.84070796460177,
      "grad_norm": 0.18874377319686708,
      "learning_rate": 8.564115896680103e-08,
      "loss": 0.0088,
      "step": 10593
    },
    {
      "epoch": 2.840976133011531,
      "grad_norm": 0.2573554419635773,
      "learning_rate": 8.535387014955544e-08,
      "loss": 0.011,
      "step": 10594
    },
    {
      "epoch": 2.8412443014212925,
      "grad_norm": 0.23311093464832702,
      "learning_rate": 8.506705985835584e-08,
      "loss": 0.0133,
      "step": 10595
    },
    {
      "epoch": 2.841512469831054,
      "grad_norm": 0.2679061021423933,
      "learning_rate": 8.47807281211277e-08,
      "loss": 0.0163,
      "step": 10596
    },
    {
      "epoch": 2.8417806382408153,
      "grad_norm": 0.19947671978766254,
      "learning_rate": 8.449487496575037e-08,
      "loss": 0.0149,
      "step": 10597
    },
    {
      "epoch": 2.8420488066505767,
      "grad_norm": 0.2926554532419335,
      "learning_rate": 8.420950042005716e-08,
      "loss": 0.0197,
      "step": 10598
    },
    {
      "epoch": 2.8423169750603376,
      "grad_norm": 0.2313980388441465,
      "learning_rate": 8.392460451183304e-08,
      "loss": 0.0101,
      "step": 10599
    },
    {
      "epoch": 2.8425851434700995,
      "grad_norm": 0.22247884604244084,
      "learning_rate": 8.364018726881861e-08,
      "loss": 0.0146,
      "step": 10600
    },
    {
      "epoch": 2.8428533118798605,
      "grad_norm": 0.22514501709865536,
      "learning_rate": 8.335624871870618e-08,
      "loss": 0.0126,
      "step": 10601
    },
    {
      "epoch": 2.843121480289622,
      "grad_norm": 0.2792580500414105,
      "learning_rate": 8.307278888914249e-08,
      "loss": 0.0141,
      "step": 10602
    },
    {
      "epoch": 2.8433896486993833,
      "grad_norm": 0.2445241118676122,
      "learning_rate": 8.278980780772717e-08,
      "loss": 0.0091,
      "step": 10603
    },
    {
      "epoch": 2.8436578171091447,
      "grad_norm": 0.3082495604497813,
      "learning_rate": 8.250730550201314e-08,
      "loss": 0.0242,
      "step": 10604
    },
    {
      "epoch": 2.843925985518906,
      "grad_norm": 0.1942839327026283,
      "learning_rate": 8.22252819995073e-08,
      "loss": 0.0081,
      "step": 10605
    },
    {
      "epoch": 2.844194153928667,
      "grad_norm": 0.3192349591541725,
      "learning_rate": 8.194373732766936e-08,
      "loss": 0.0146,
      "step": 10606
    },
    {
      "epoch": 2.8444623223384284,
      "grad_norm": 0.2933682104098624,
      "learning_rate": 8.166267151391238e-08,
      "loss": 0.0191,
      "step": 10607
    },
    {
      "epoch": 2.84473049074819,
      "grad_norm": 0.2531452625630813,
      "learning_rate": 8.138208458560337e-08,
      "loss": 0.0151,
      "step": 10608
    },
    {
      "epoch": 2.8449986591579512,
      "grad_norm": 0.3110276138444187,
      "learning_rate": 8.110197657006158e-08,
      "loss": 0.0179,
      "step": 10609
    },
    {
      "epoch": 2.8452668275677127,
      "grad_norm": 0.19220130459381576,
      "learning_rate": 8.082234749456186e-08,
      "loss": 0.0096,
      "step": 10610
    },
    {
      "epoch": 2.8455349959774736,
      "grad_norm": 0.2628642329696975,
      "learning_rate": 8.05431973863291e-08,
      "loss": 0.011,
      "step": 10611
    },
    {
      "epoch": 2.845803164387235,
      "grad_norm": 0.22975694864832294,
      "learning_rate": 8.026452627254432e-08,
      "loss": 0.0111,
      "step": 10612
    },
    {
      "epoch": 2.8460713327969964,
      "grad_norm": 0.30306437937548053,
      "learning_rate": 7.998633418034196e-08,
      "loss": 0.0197,
      "step": 10613
    },
    {
      "epoch": 2.846339501206758,
      "grad_norm": 0.244860342718741,
      "learning_rate": 7.9708621136807e-08,
      "loss": 0.0126,
      "step": 10614
    },
    {
      "epoch": 2.8466076696165192,
      "grad_norm": 0.19685454132794,
      "learning_rate": 7.943138716898058e-08,
      "loss": 0.0107,
      "step": 10615
    },
    {
      "epoch": 2.8468758380262806,
      "grad_norm": 0.3121269895594631,
      "learning_rate": 7.915463230385611e-08,
      "loss": 0.02,
      "step": 10616
    },
    {
      "epoch": 2.847144006436042,
      "grad_norm": 0.2642243776815256,
      "learning_rate": 7.887835656838038e-08,
      "loss": 0.0127,
      "step": 10617
    },
    {
      "epoch": 2.847412174845803,
      "grad_norm": 0.25861735624457133,
      "learning_rate": 7.86025599894541e-08,
      "loss": 0.0129,
      "step": 10618
    },
    {
      "epoch": 2.8476803432555644,
      "grad_norm": 0.25801462181122226,
      "learning_rate": 7.832724259393021e-08,
      "loss": 0.0129,
      "step": 10619
    },
    {
      "epoch": 2.847948511665326,
      "grad_norm": 0.23044693288051027,
      "learning_rate": 7.805240440861673e-08,
      "loss": 0.0111,
      "step": 10620
    },
    {
      "epoch": 2.848216680075087,
      "grad_norm": 0.26756974928200905,
      "learning_rate": 7.777804546027224e-08,
      "loss": 0.0133,
      "step": 10621
    },
    {
      "epoch": 2.8484848484848486,
      "grad_norm": 0.27012373730553907,
      "learning_rate": 7.750416577561092e-08,
      "loss": 0.0174,
      "step": 10622
    },
    {
      "epoch": 2.8487530168946096,
      "grad_norm": 0.268666903106677,
      "learning_rate": 7.723076538130093e-08,
      "loss": 0.0121,
      "step": 10623
    },
    {
      "epoch": 2.849021185304371,
      "grad_norm": 0.25512591981013816,
      "learning_rate": 7.695784430396092e-08,
      "loss": 0.0115,
      "step": 10624
    },
    {
      "epoch": 2.8492893537141324,
      "grad_norm": 0.19976715728678632,
      "learning_rate": 7.668540257016577e-08,
      "loss": 0.0082,
      "step": 10625
    },
    {
      "epoch": 2.849557522123894,
      "grad_norm": 0.3158987901622547,
      "learning_rate": 7.641344020644148e-08,
      "loss": 0.0186,
      "step": 10626
    },
    {
      "epoch": 2.849825690533655,
      "grad_norm": 0.24445990152292052,
      "learning_rate": 7.614195723926854e-08,
      "loss": 0.0125,
      "step": 10627
    },
    {
      "epoch": 2.8500938589434166,
      "grad_norm": 0.23900855271479524,
      "learning_rate": 7.58709536950808e-08,
      "loss": 0.0089,
      "step": 10628
    },
    {
      "epoch": 2.850362027353178,
      "grad_norm": 0.2994783391640885,
      "learning_rate": 7.560042960026493e-08,
      "loss": 0.0228,
      "step": 10629
    },
    {
      "epoch": 2.850630195762939,
      "grad_norm": 0.2339018765302426,
      "learning_rate": 7.533038498116152e-08,
      "loss": 0.0156,
      "step": 10630
    },
    {
      "epoch": 2.8508983641727004,
      "grad_norm": 0.18157292859375423,
      "learning_rate": 7.506081986406344e-08,
      "loss": 0.0089,
      "step": 10631
    },
    {
      "epoch": 2.8511665325824618,
      "grad_norm": 0.23329198027324297,
      "learning_rate": 7.479173427521747e-08,
      "loss": 0.013,
      "step": 10632
    },
    {
      "epoch": 2.851434700992223,
      "grad_norm": 0.1953503453061723,
      "learning_rate": 7.452312824082486e-08,
      "loss": 0.0082,
      "step": 10633
    },
    {
      "epoch": 2.8517028694019846,
      "grad_norm": 0.17172112633992723,
      "learning_rate": 7.425500178703804e-08,
      "loss": 0.0074,
      "step": 10634
    },
    {
      "epoch": 2.8519710378117455,
      "grad_norm": 0.22745110833554352,
      "learning_rate": 7.398735493996445e-08,
      "loss": 0.0093,
      "step": 10635
    },
    {
      "epoch": 2.852239206221507,
      "grad_norm": 0.22617404795611154,
      "learning_rate": 7.372018772566325e-08,
      "loss": 0.0087,
      "step": 10636
    },
    {
      "epoch": 2.8525073746312684,
      "grad_norm": 0.21104348577197363,
      "learning_rate": 7.345350017014863e-08,
      "loss": 0.0113,
      "step": 10637
    },
    {
      "epoch": 2.8527755430410298,
      "grad_norm": 0.20690144836158955,
      "learning_rate": 7.318729229938759e-08,
      "loss": 0.0094,
      "step": 10638
    },
    {
      "epoch": 2.853043711450791,
      "grad_norm": 0.31499414877338716,
      "learning_rate": 7.292156413929829e-08,
      "loss": 0.012,
      "step": 10639
    },
    {
      "epoch": 2.8533118798605526,
      "grad_norm": 0.25110972579920054,
      "learning_rate": 7.265631571575616e-08,
      "loss": 0.0098,
      "step": 10640
    },
    {
      "epoch": 2.853580048270314,
      "grad_norm": 0.29669863099853355,
      "learning_rate": 7.239154705458606e-08,
      "loss": 0.0148,
      "step": 10641
    },
    {
      "epoch": 2.853848216680075,
      "grad_norm": 0.20904327978688195,
      "learning_rate": 7.212725818156852e-08,
      "loss": 0.0128,
      "step": 10642
    },
    {
      "epoch": 2.8541163850898363,
      "grad_norm": 0.24825748832240835,
      "learning_rate": 7.186344912243681e-08,
      "loss": 0.0131,
      "step": 10643
    },
    {
      "epoch": 2.8543845534995977,
      "grad_norm": 0.30219151793847576,
      "learning_rate": 7.160011990287708e-08,
      "loss": 0.0154,
      "step": 10644
    },
    {
      "epoch": 2.854652721909359,
      "grad_norm": 0.29647750132174816,
      "learning_rate": 7.133727054852935e-08,
      "loss": 0.0224,
      "step": 10645
    },
    {
      "epoch": 2.8549208903191206,
      "grad_norm": 0.29879632130258627,
      "learning_rate": 7.107490108498539e-08,
      "loss": 0.014,
      "step": 10646
    },
    {
      "epoch": 2.8551890587288815,
      "grad_norm": 0.34309410047854594,
      "learning_rate": 7.081301153779308e-08,
      "loss": 0.0119,
      "step": 10647
    },
    {
      "epoch": 2.855457227138643,
      "grad_norm": 0.23692651617330582,
      "learning_rate": 7.055160193245036e-08,
      "loss": 0.0113,
      "step": 10648
    },
    {
      "epoch": 2.8557253955484043,
      "grad_norm": 0.25867194544708405,
      "learning_rate": 7.029067229441133e-08,
      "loss": 0.0095,
      "step": 10649
    },
    {
      "epoch": 2.8559935639581657,
      "grad_norm": 0.23751708220075085,
      "learning_rate": 7.003022264908177e-08,
      "loss": 0.0143,
      "step": 10650
    },
    {
      "epoch": 2.856261732367927,
      "grad_norm": 0.24329228810271727,
      "learning_rate": 6.977025302181973e-08,
      "loss": 0.0118,
      "step": 10651
    },
    {
      "epoch": 2.8565299007776885,
      "grad_norm": 0.25372366632987153,
      "learning_rate": 6.951076343793939e-08,
      "loss": 0.0129,
      "step": 10652
    },
    {
      "epoch": 2.85679806918745,
      "grad_norm": 0.2449690571642807,
      "learning_rate": 6.9251753922705e-08,
      "loss": 0.0135,
      "step": 10653
    },
    {
      "epoch": 2.857066237597211,
      "grad_norm": 0.2269502125287009,
      "learning_rate": 6.899322450133694e-08,
      "loss": 0.0161,
      "step": 10654
    },
    {
      "epoch": 2.8573344060069723,
      "grad_norm": 0.23648397133003965,
      "learning_rate": 6.873517519900729e-08,
      "loss": 0.0114,
      "step": 10655
    },
    {
      "epoch": 2.8576025744167337,
      "grad_norm": 0.18302301346135239,
      "learning_rate": 6.847760604084098e-08,
      "loss": 0.0102,
      "step": 10656
    },
    {
      "epoch": 2.857870742826495,
      "grad_norm": 0.21852575731600501,
      "learning_rate": 6.822051705191734e-08,
      "loss": 0.01,
      "step": 10657
    },
    {
      "epoch": 2.8581389112362565,
      "grad_norm": 0.27364786077113185,
      "learning_rate": 6.796390825726807e-08,
      "loss": 0.0136,
      "step": 10658
    },
    {
      "epoch": 2.8584070796460175,
      "grad_norm": 0.2099567349482354,
      "learning_rate": 6.77077796818787e-08,
      "loss": 0.0102,
      "step": 10659
    },
    {
      "epoch": 2.858675248055779,
      "grad_norm": 0.18837511541531632,
      "learning_rate": 6.745213135068818e-08,
      "loss": 0.0091,
      "step": 10660
    },
    {
      "epoch": 2.8589434164655403,
      "grad_norm": 0.23162943414378795,
      "learning_rate": 6.719696328858827e-08,
      "loss": 0.0097,
      "step": 10661
    },
    {
      "epoch": 2.8592115848753017,
      "grad_norm": 0.2685984372048917,
      "learning_rate": 6.694227552042354e-08,
      "loss": 0.0147,
      "step": 10662
    },
    {
      "epoch": 2.859479753285063,
      "grad_norm": 0.23250500136628197,
      "learning_rate": 6.668806807099193e-08,
      "loss": 0.013,
      "step": 10663
    },
    {
      "epoch": 2.8597479216948245,
      "grad_norm": 0.23767922398000396,
      "learning_rate": 6.643434096504587e-08,
      "loss": 0.0138,
      "step": 10664
    },
    {
      "epoch": 2.860016090104586,
      "grad_norm": 0.27494063792007367,
      "learning_rate": 6.618109422728947e-08,
      "loss": 0.0113,
      "step": 10665
    },
    {
      "epoch": 2.860284258514347,
      "grad_norm": 0.22921451603165832,
      "learning_rate": 6.59283278823808e-08,
      "loss": 0.014,
      "step": 10666
    },
    {
      "epoch": 2.8605524269241083,
      "grad_norm": 0.290705815372379,
      "learning_rate": 6.567604195493182e-08,
      "loss": 0.012,
      "step": 10667
    },
    {
      "epoch": 2.8608205953338697,
      "grad_norm": 0.26276898336995486,
      "learning_rate": 6.542423646950569e-08,
      "loss": 0.0181,
      "step": 10668
    },
    {
      "epoch": 2.861088763743631,
      "grad_norm": 0.23743219836329005,
      "learning_rate": 6.517291145062054e-08,
      "loss": 0.0092,
      "step": 10669
    },
    {
      "epoch": 2.8613569321533925,
      "grad_norm": 0.24654596737261228,
      "learning_rate": 6.492206692274739e-08,
      "loss": 0.0135,
      "step": 10670
    },
    {
      "epoch": 2.8616251005631534,
      "grad_norm": 0.3248934213139609,
      "learning_rate": 6.467170291030999e-08,
      "loss": 0.0222,
      "step": 10671
    },
    {
      "epoch": 2.861893268972915,
      "grad_norm": 0.22896643140062836,
      "learning_rate": 6.44218194376861e-08,
      "loss": 0.0113,
      "step": 10672
    },
    {
      "epoch": 2.8621614373826763,
      "grad_norm": 0.256494872984811,
      "learning_rate": 6.417241652920625e-08,
      "loss": 0.016,
      "step": 10673
    },
    {
      "epoch": 2.8624296057924377,
      "grad_norm": 0.2714888626904641,
      "learning_rate": 6.392349420915323e-08,
      "loss": 0.0136,
      "step": 10674
    },
    {
      "epoch": 2.862697774202199,
      "grad_norm": 0.28316952574319487,
      "learning_rate": 6.367505250176543e-08,
      "loss": 0.0117,
      "step": 10675
    },
    {
      "epoch": 2.8629659426119605,
      "grad_norm": 0.24094608967141434,
      "learning_rate": 6.342709143123126e-08,
      "loss": 0.0148,
      "step": 10676
    },
    {
      "epoch": 2.863234111021722,
      "grad_norm": 0.21256041140210563,
      "learning_rate": 6.317961102169534e-08,
      "loss": 0.0128,
      "step": 10677
    },
    {
      "epoch": 2.863502279431483,
      "grad_norm": 0.2500259222631522,
      "learning_rate": 6.293261129725281e-08,
      "loss": 0.0112,
      "step": 10678
    },
    {
      "epoch": 2.8637704478412442,
      "grad_norm": 0.23845455651228067,
      "learning_rate": 6.268609228195499e-08,
      "loss": 0.0117,
      "step": 10679
    },
    {
      "epoch": 2.8640386162510056,
      "grad_norm": 0.18522798161035695,
      "learning_rate": 6.24400539998038e-08,
      "loss": 0.0122,
      "step": 10680
    },
    {
      "epoch": 2.864306784660767,
      "grad_norm": 0.2958123940959807,
      "learning_rate": 6.219449647475562e-08,
      "loss": 0.0179,
      "step": 10681
    },
    {
      "epoch": 2.8645749530705285,
      "grad_norm": 0.2652717578234367,
      "learning_rate": 6.194941973072021e-08,
      "loss": 0.0124,
      "step": 10682
    },
    {
      "epoch": 2.8648431214802894,
      "grad_norm": 0.2524821898966749,
      "learning_rate": 6.170482379155907e-08,
      "loss": 0.018,
      "step": 10683
    },
    {
      "epoch": 2.865111289890051,
      "grad_norm": 0.22064700791201303,
      "learning_rate": 6.146070868108811e-08,
      "loss": 0.0121,
      "step": 10684
    },
    {
      "epoch": 2.865379458299812,
      "grad_norm": 0.2808869203904827,
      "learning_rate": 6.121707442307722e-08,
      "loss": 0.0129,
      "step": 10685
    },
    {
      "epoch": 2.8656476267095736,
      "grad_norm": 0.27018142674042606,
      "learning_rate": 6.097392104124633e-08,
      "loss": 0.0134,
      "step": 10686
    },
    {
      "epoch": 2.865915795119335,
      "grad_norm": 0.26159361132421055,
      "learning_rate": 6.073124855927315e-08,
      "loss": 0.0136,
      "step": 10687
    },
    {
      "epoch": 2.8661839635290964,
      "grad_norm": 0.22421743481175047,
      "learning_rate": 6.048905700078378e-08,
      "loss": 0.0095,
      "step": 10688
    },
    {
      "epoch": 2.866452131938858,
      "grad_norm": 0.30935380844032195,
      "learning_rate": 6.024734638936159e-08,
      "loss": 0.0152,
      "step": 10689
    },
    {
      "epoch": 2.866720300348619,
      "grad_norm": 0.4071782032792132,
      "learning_rate": 6.000611674853995e-08,
      "loss": 0.0167,
      "step": 10690
    },
    {
      "epoch": 2.86698846875838,
      "grad_norm": 0.20571066499942672,
      "learning_rate": 5.976536810180677e-08,
      "loss": 0.0103,
      "step": 10691
    },
    {
      "epoch": 2.8672566371681416,
      "grad_norm": 0.3646466863011554,
      "learning_rate": 5.9525100472604955e-08,
      "loss": 0.0169,
      "step": 10692
    },
    {
      "epoch": 2.867524805577903,
      "grad_norm": 0.2568548624802712,
      "learning_rate": 5.928531388432635e-08,
      "loss": 0.0101,
      "step": 10693
    },
    {
      "epoch": 2.8677929739876644,
      "grad_norm": 0.19578701561309472,
      "learning_rate": 5.904600836032004e-08,
      "loss": 0.0086,
      "step": 10694
    },
    {
      "epoch": 2.8680611423974254,
      "grad_norm": 0.2379866992449065,
      "learning_rate": 5.880718392388518e-08,
      "loss": 0.0113,
      "step": 10695
    },
    {
      "epoch": 2.868329310807187,
      "grad_norm": 0.22682817567827396,
      "learning_rate": 5.8568840598277054e-08,
      "loss": 0.009,
      "step": 10696
    },
    {
      "epoch": 2.868597479216948,
      "grad_norm": 0.24001908118358514,
      "learning_rate": 5.833097840670155e-08,
      "loss": 0.013,
      "step": 10697
    },
    {
      "epoch": 2.8688656476267096,
      "grad_norm": 0.19874948778386625,
      "learning_rate": 5.809359737231846e-08,
      "loss": 0.0091,
      "step": 10698
    },
    {
      "epoch": 2.869133816036471,
      "grad_norm": 0.23576132917668016,
      "learning_rate": 5.7856697518241524e-08,
      "loss": 0.0084,
      "step": 10699
    },
    {
      "epoch": 2.869401984446232,
      "grad_norm": 0.18163783281297288,
      "learning_rate": 5.7620278867536185e-08,
      "loss": 0.0099,
      "step": 10700
    },
    {
      "epoch": 2.869670152855994,
      "grad_norm": 0.2369892362471996,
      "learning_rate": 5.738434144322291e-08,
      "loss": 0.0138,
      "step": 10701
    },
    {
      "epoch": 2.8699383212657548,
      "grad_norm": 0.23892673665658076,
      "learning_rate": 5.714888526827444e-08,
      "loss": 0.0137,
      "step": 10702
    },
    {
      "epoch": 2.870206489675516,
      "grad_norm": 0.2926573374691794,
      "learning_rate": 5.691391036561522e-08,
      "loss": 0.017,
      "step": 10703
    },
    {
      "epoch": 2.8704746580852776,
      "grad_norm": 0.21972722960786198,
      "learning_rate": 5.667941675812527e-08,
      "loss": 0.0132,
      "step": 10704
    },
    {
      "epoch": 2.870742826495039,
      "grad_norm": 0.24287825711764252,
      "learning_rate": 5.6445404468636334e-08,
      "loss": 0.0135,
      "step": 10705
    },
    {
      "epoch": 2.8710109949048004,
      "grad_norm": 0.28202639186980843,
      "learning_rate": 5.6211873519933516e-08,
      "loss": 0.0132,
      "step": 10706
    },
    {
      "epoch": 2.8712791633145613,
      "grad_norm": 0.19571743792785304,
      "learning_rate": 5.597882393475473e-08,
      "loss": 0.0121,
      "step": 10707
    },
    {
      "epoch": 2.8715473317243227,
      "grad_norm": 0.2081721081017145,
      "learning_rate": 5.5746255735791846e-08,
      "loss": 0.0088,
      "step": 10708
    },
    {
      "epoch": 2.871815500134084,
      "grad_norm": 0.2672145414492506,
      "learning_rate": 5.5514168945689506e-08,
      "loss": 0.0112,
      "step": 10709
    },
    {
      "epoch": 2.8720836685438456,
      "grad_norm": 0.23324105151108748,
      "learning_rate": 5.52825635870452e-08,
      "loss": 0.0091,
      "step": 10710
    },
    {
      "epoch": 2.872351836953607,
      "grad_norm": 0.22366036685290314,
      "learning_rate": 5.505143968240923e-08,
      "loss": 0.0172,
      "step": 10711
    },
    {
      "epoch": 2.872620005363368,
      "grad_norm": 0.2159042028953645,
      "learning_rate": 5.4820797254286904e-08,
      "loss": 0.0088,
      "step": 10712
    },
    {
      "epoch": 2.8728881737731298,
      "grad_norm": 0.23077847823432823,
      "learning_rate": 5.4590636325133615e-08,
      "loss": 0.0134,
      "step": 10713
    },
    {
      "epoch": 2.8731563421828907,
      "grad_norm": 0.2030900613957855,
      "learning_rate": 5.436095691736032e-08,
      "loss": 0.0109,
      "step": 10714
    },
    {
      "epoch": 2.873424510592652,
      "grad_norm": 0.22615386980389088,
      "learning_rate": 5.413175905333024e-08,
      "loss": 0.0132,
      "step": 10715
    },
    {
      "epoch": 2.8736926790024135,
      "grad_norm": 0.26292300237158145,
      "learning_rate": 5.390304275535996e-08,
      "loss": 0.0177,
      "step": 10716
    },
    {
      "epoch": 2.873960847412175,
      "grad_norm": 0.3110479559754243,
      "learning_rate": 5.367480804571834e-08,
      "loss": 0.0212,
      "step": 10717
    },
    {
      "epoch": 2.8742290158219364,
      "grad_norm": 0.2147478759117049,
      "learning_rate": 5.3447054946628715e-08,
      "loss": 0.0092,
      "step": 10718
    },
    {
      "epoch": 2.8744971842316973,
      "grad_norm": 0.2323908150578418,
      "learning_rate": 5.3219783480266685e-08,
      "loss": 0.0124,
      "step": 10719
    },
    {
      "epoch": 2.8747653526414587,
      "grad_norm": 0.2282675714810508,
      "learning_rate": 5.2992993668760654e-08,
      "loss": 0.0118,
      "step": 10720
    },
    {
      "epoch": 2.87503352105122,
      "grad_norm": 0.18849140745619442,
      "learning_rate": 5.276668553419295e-08,
      "loss": 0.0107,
      "step": 10721
    },
    {
      "epoch": 2.8753016894609815,
      "grad_norm": 0.3035223877169693,
      "learning_rate": 5.254085909859874e-08,
      "loss": 0.0141,
      "step": 10722
    },
    {
      "epoch": 2.875569857870743,
      "grad_norm": 0.25746366177247537,
      "learning_rate": 5.231551438396543e-08,
      "loss": 0.0163,
      "step": 10723
    },
    {
      "epoch": 2.875838026280504,
      "grad_norm": 0.22775690719760497,
      "learning_rate": 5.2090651412234906e-08,
      "loss": 0.0121,
      "step": 10724
    },
    {
      "epoch": 2.8761061946902657,
      "grad_norm": 0.2014942840297933,
      "learning_rate": 5.186627020530077e-08,
      "loss": 0.0103,
      "step": 10725
    },
    {
      "epoch": 2.8763743631000267,
      "grad_norm": 0.29869823714863425,
      "learning_rate": 5.164237078501111e-08,
      "loss": 0.0217,
      "step": 10726
    },
    {
      "epoch": 2.876642531509788,
      "grad_norm": 0.2490907457747659,
      "learning_rate": 5.1418953173165695e-08,
      "loss": 0.0164,
      "step": 10727
    },
    {
      "epoch": 2.8769106999195495,
      "grad_norm": 0.19742446996431975,
      "learning_rate": 5.1196017391518804e-08,
      "loss": 0.0078,
      "step": 10728
    },
    {
      "epoch": 2.877178868329311,
      "grad_norm": 0.19184811122076945,
      "learning_rate": 5.097356346177751e-08,
      "loss": 0.0103,
      "step": 10729
    },
    {
      "epoch": 2.8774470367390723,
      "grad_norm": 0.20822349845271776,
      "learning_rate": 5.075159140560004e-08,
      "loss": 0.0092,
      "step": 10730
    },
    {
      "epoch": 2.8777152051488333,
      "grad_norm": 0.23490334002865088,
      "learning_rate": 5.053010124460078e-08,
      "loss": 0.0162,
      "step": 10731
    },
    {
      "epoch": 2.8779833735585947,
      "grad_norm": 0.32736689171025674,
      "learning_rate": 5.03090930003447e-08,
      "loss": 0.0134,
      "step": 10732
    },
    {
      "epoch": 2.878251541968356,
      "grad_norm": 0.2676807397876617,
      "learning_rate": 5.008856669435125e-08,
      "loss": 0.0152,
      "step": 10733
    },
    {
      "epoch": 2.8785197103781175,
      "grad_norm": 0.3198441360541822,
      "learning_rate": 4.986852234809214e-08,
      "loss": 0.0155,
      "step": 10734
    },
    {
      "epoch": 2.878787878787879,
      "grad_norm": 0.2427503049227703,
      "learning_rate": 4.964895998299246e-08,
      "loss": 0.0104,
      "step": 10735
    },
    {
      "epoch": 2.87905604719764,
      "grad_norm": 0.25758445885932013,
      "learning_rate": 4.9429879620431223e-08,
      "loss": 0.0132,
      "step": 10736
    },
    {
      "epoch": 2.8793242156074017,
      "grad_norm": 0.38261435675620703,
      "learning_rate": 4.9211281281738574e-08,
      "loss": 0.0093,
      "step": 10737
    },
    {
      "epoch": 2.8795923840171627,
      "grad_norm": 0.21337977691120827,
      "learning_rate": 4.8993164988199174e-08,
      "loss": 0.0122,
      "step": 10738
    },
    {
      "epoch": 2.879860552426924,
      "grad_norm": 0.27265397310305844,
      "learning_rate": 4.877553076105157e-08,
      "loss": 0.0151,
      "step": 10739
    },
    {
      "epoch": 2.8801287208366855,
      "grad_norm": 0.24466895999626964,
      "learning_rate": 4.8558378621484384e-08,
      "loss": 0.0136,
      "step": 10740
    },
    {
      "epoch": 2.880396889246447,
      "grad_norm": 0.1634364335481162,
      "learning_rate": 4.834170859064236e-08,
      "loss": 0.0064,
      "step": 10741
    },
    {
      "epoch": 2.8806650576562083,
      "grad_norm": 0.22797094254589756,
      "learning_rate": 4.812552068962195e-08,
      "loss": 0.0109,
      "step": 10742
    },
    {
      "epoch": 2.8809332260659692,
      "grad_norm": 0.20989060110905947,
      "learning_rate": 4.790981493947244e-08,
      "loss": 0.0132,
      "step": 10743
    },
    {
      "epoch": 2.8812013944757306,
      "grad_norm": 0.232073402958727,
      "learning_rate": 4.769459136119647e-08,
      "loss": 0.0113,
      "step": 10744
    },
    {
      "epoch": 2.881469562885492,
      "grad_norm": 0.39347502436718024,
      "learning_rate": 4.747984997575006e-08,
      "loss": 0.0216,
      "step": 10745
    },
    {
      "epoch": 2.8817377312952535,
      "grad_norm": 0.2418846290147638,
      "learning_rate": 4.726559080404258e-08,
      "loss": 0.0132,
      "step": 10746
    },
    {
      "epoch": 2.882005899705015,
      "grad_norm": 0.5233123498295074,
      "learning_rate": 4.705181386693458e-08,
      "loss": 0.0157,
      "step": 10747
    },
    {
      "epoch": 2.882274068114776,
      "grad_norm": 0.2476118016700826,
      "learning_rate": 4.6838519185242184e-08,
      "loss": 0.0159,
      "step": 10748
    },
    {
      "epoch": 2.8825422365245377,
      "grad_norm": 0.23716619977480552,
      "learning_rate": 4.662570677973266e-08,
      "loss": 0.0164,
      "step": 10749
    },
    {
      "epoch": 2.8828104049342986,
      "grad_norm": 0.3396500767859851,
      "learning_rate": 4.6413376671126665e-08,
      "loss": 0.0134,
      "step": 10750
    },
    {
      "epoch": 2.88307857334406,
      "grad_norm": 0.18867047756361685,
      "learning_rate": 4.620152888009932e-08,
      "loss": 0.01,
      "step": 10751
    },
    {
      "epoch": 2.8833467417538214,
      "grad_norm": 0.28418149568434675,
      "learning_rate": 4.599016342727636e-08,
      "loss": 0.02,
      "step": 10752
    },
    {
      "epoch": 2.883614910163583,
      "grad_norm": 0.19196365125978784,
      "learning_rate": 4.5779280333239086e-08,
      "loss": 0.0105,
      "step": 10753
    },
    {
      "epoch": 2.8838830785733442,
      "grad_norm": 0.33045287147534574,
      "learning_rate": 4.556887961851941e-08,
      "loss": 0.028,
      "step": 10754
    },
    {
      "epoch": 2.884151246983105,
      "grad_norm": 0.21781141614604496,
      "learning_rate": 4.5358961303604845e-08,
      "loss": 0.0101,
      "step": 10755
    },
    {
      "epoch": 2.8844194153928666,
      "grad_norm": 0.25561495282876745,
      "learning_rate": 4.514952540893347e-08,
      "loss": 0.012,
      "step": 10756
    },
    {
      "epoch": 2.884687583802628,
      "grad_norm": 0.24450844934152924,
      "learning_rate": 4.494057195489843e-08,
      "loss": 0.0142,
      "step": 10757
    },
    {
      "epoch": 2.8849557522123894,
      "grad_norm": 0.22941195540687426,
      "learning_rate": 4.473210096184455e-08,
      "loss": 0.0113,
      "step": 10758
    },
    {
      "epoch": 2.885223920622151,
      "grad_norm": 0.22881737929831394,
      "learning_rate": 4.452411245006949e-08,
      "loss": 0.0149,
      "step": 10759
    },
    {
      "epoch": 2.885492089031912,
      "grad_norm": 0.30357110617921057,
      "learning_rate": 4.431660643982483e-08,
      "loss": 0.013,
      "step": 10760
    },
    {
      "epoch": 2.8857602574416736,
      "grad_norm": 0.20301544718138229,
      "learning_rate": 4.410958295131551e-08,
      "loss": 0.0089,
      "step": 10761
    },
    {
      "epoch": 2.8860284258514346,
      "grad_norm": 0.34250606222542357,
      "learning_rate": 4.390304200469875e-08,
      "loss": 0.0168,
      "step": 10762
    },
    {
      "epoch": 2.886296594261196,
      "grad_norm": 0.2811058101704099,
      "learning_rate": 4.3696983620084566e-08,
      "loss": 0.0134,
      "step": 10763
    },
    {
      "epoch": 2.8865647626709574,
      "grad_norm": 0.23686941892929175,
      "learning_rate": 4.34914078175358e-08,
      "loss": 0.0113,
      "step": 10764
    },
    {
      "epoch": 2.886832931080719,
      "grad_norm": 0.2530079734053365,
      "learning_rate": 4.328631461706978e-08,
      "loss": 0.0144,
      "step": 10765
    },
    {
      "epoch": 2.88710109949048,
      "grad_norm": 0.284550594947064,
      "learning_rate": 4.308170403865497e-08,
      "loss": 0.0152,
      "step": 10766
    },
    {
      "epoch": 2.887369267900241,
      "grad_norm": 0.2586890597539836,
      "learning_rate": 4.287757610221488e-08,
      "loss": 0.0185,
      "step": 10767
    },
    {
      "epoch": 2.8876374363100026,
      "grad_norm": 0.24639457734596898,
      "learning_rate": 4.2673930827624166e-08,
      "loss": 0.0085,
      "step": 10768
    },
    {
      "epoch": 2.887905604719764,
      "grad_norm": 0.2833016823495599,
      "learning_rate": 4.247076823471141e-08,
      "loss": 0.014,
      "step": 10769
    },
    {
      "epoch": 2.8881737731295254,
      "grad_norm": 0.21993450036100445,
      "learning_rate": 4.226808834325802e-08,
      "loss": 0.0133,
      "step": 10770
    },
    {
      "epoch": 2.888441941539287,
      "grad_norm": 0.18793939372181975,
      "learning_rate": 4.20658911729982e-08,
      "loss": 0.0079,
      "step": 10771
    },
    {
      "epoch": 2.8887101099490478,
      "grad_norm": 0.24432389409290467,
      "learning_rate": 4.1864176743619535e-08,
      "loss": 0.0116,
      "step": 10772
    },
    {
      "epoch": 2.8889782783588096,
      "grad_norm": 0.23823909659121906,
      "learning_rate": 4.166294507476243e-08,
      "loss": 0.0111,
      "step": 10773
    },
    {
      "epoch": 2.8892464467685706,
      "grad_norm": 0.26261729110912074,
      "learning_rate": 4.1462196186019546e-08,
      "loss": 0.0133,
      "step": 10774
    },
    {
      "epoch": 2.889514615178332,
      "grad_norm": 0.29790486687422574,
      "learning_rate": 4.126193009693913e-08,
      "loss": 0.0191,
      "step": 10775
    },
    {
      "epoch": 2.8897827835880934,
      "grad_norm": 0.3491887357239462,
      "learning_rate": 4.106214682701837e-08,
      "loss": 0.0116,
      "step": 10776
    },
    {
      "epoch": 2.8900509519978548,
      "grad_norm": 0.24965399030364568,
      "learning_rate": 4.0862846395711145e-08,
      "loss": 0.0155,
      "step": 10777
    },
    {
      "epoch": 2.890319120407616,
      "grad_norm": 0.20786935512612767,
      "learning_rate": 4.0664028822422486e-08,
      "loss": 0.0102,
      "step": 10778
    },
    {
      "epoch": 2.890587288817377,
      "grad_norm": 0.24718692303597062,
      "learning_rate": 4.046569412651025e-08,
      "loss": 0.0177,
      "step": 10779
    },
    {
      "epoch": 2.8908554572271385,
      "grad_norm": 0.34224180609098015,
      "learning_rate": 4.02678423272862e-08,
      "loss": 0.0152,
      "step": 10780
    },
    {
      "epoch": 2.8911236256369,
      "grad_norm": 0.16358025461594786,
      "learning_rate": 4.007047344401438e-08,
      "loss": 0.0078,
      "step": 10781
    },
    {
      "epoch": 2.8913917940466614,
      "grad_norm": 0.2062651264482024,
      "learning_rate": 3.987358749591219e-08,
      "loss": 0.009,
      "step": 10782
    },
    {
      "epoch": 2.8916599624564228,
      "grad_norm": 0.2775558503949398,
      "learning_rate": 3.9677184502149855e-08,
      "loss": 0.0225,
      "step": 10783
    },
    {
      "epoch": 2.8919281308661837,
      "grad_norm": 0.284556850669693,
      "learning_rate": 3.948126448185097e-08,
      "loss": 0.0119,
      "step": 10784
    },
    {
      "epoch": 2.8921962992759456,
      "grad_norm": 0.9538459684149021,
      "learning_rate": 3.928582745409137e-08,
      "loss": 0.0128,
      "step": 10785
    },
    {
      "epoch": 2.8924644676857065,
      "grad_norm": 0.25046198133831027,
      "learning_rate": 3.90908734379003e-08,
      "loss": 0.0122,
      "step": 10786
    },
    {
      "epoch": 2.892732636095468,
      "grad_norm": 0.2113598638935938,
      "learning_rate": 3.889640245225923e-08,
      "loss": 0.0115,
      "step": 10787
    },
    {
      "epoch": 2.8930008045052293,
      "grad_norm": 0.2302661518063801,
      "learning_rate": 3.870241451610468e-08,
      "loss": 0.0129,
      "step": 10788
    },
    {
      "epoch": 2.8932689729149907,
      "grad_norm": 0.2448927180305606,
      "learning_rate": 3.850890964832377e-08,
      "loss": 0.0109,
      "step": 10789
    },
    {
      "epoch": 2.893537141324752,
      "grad_norm": 0.3047814027105091,
      "learning_rate": 3.8315887867758637e-08,
      "loss": 0.0187,
      "step": 10790
    },
    {
      "epoch": 2.893805309734513,
      "grad_norm": 0.21818109995242277,
      "learning_rate": 3.8123349193201484e-08,
      "loss": 0.0119,
      "step": 10791
    },
    {
      "epoch": 2.8940734781442745,
      "grad_norm": 0.19917567444670137,
      "learning_rate": 3.793129364340064e-08,
      "loss": 0.0066,
      "step": 10792
    },
    {
      "epoch": 2.894341646554036,
      "grad_norm": 0.17103409023366323,
      "learning_rate": 3.77397212370556e-08,
      "loss": 0.0084,
      "step": 10793
    },
    {
      "epoch": 2.8946098149637973,
      "grad_norm": 0.2780854285939121,
      "learning_rate": 3.754863199281977e-08,
      "loss": 0.0133,
      "step": 10794
    },
    {
      "epoch": 2.8948779833735587,
      "grad_norm": 0.2683967631065976,
      "learning_rate": 3.735802592929827e-08,
      "loss": 0.0183,
      "step": 10795
    },
    {
      "epoch": 2.8951461517833197,
      "grad_norm": 0.2780625250176772,
      "learning_rate": 3.7167903065050136e-08,
      "loss": 0.0106,
      "step": 10796
    },
    {
      "epoch": 2.895414320193081,
      "grad_norm": 0.3111365881263348,
      "learning_rate": 3.6978263418587235e-08,
      "loss": 0.0194,
      "step": 10797
    },
    {
      "epoch": 2.8956824886028425,
      "grad_norm": 0.33059402853027936,
      "learning_rate": 3.6789107008374235e-08,
      "loss": 0.0138,
      "step": 10798
    },
    {
      "epoch": 2.895950657012604,
      "grad_norm": 0.22064836189162967,
      "learning_rate": 3.6600433852829186e-08,
      "loss": 0.0132,
      "step": 10799
    },
    {
      "epoch": 2.8962188254223653,
      "grad_norm": 0.24614463949527357,
      "learning_rate": 3.641224397032184e-08,
      "loss": 0.0131,
      "step": 10800
    },
    {
      "epoch": 2.8964869938321267,
      "grad_norm": 0.227065095793287,
      "learning_rate": 3.622453737917642e-08,
      "loss": 0.0101,
      "step": 10801
    },
    {
      "epoch": 2.896755162241888,
      "grad_norm": 0.32632311494254657,
      "learning_rate": 3.603731409766942e-08,
      "loss": 0.0256,
      "step": 10802
    },
    {
      "epoch": 2.897023330651649,
      "grad_norm": 0.21838724873055596,
      "learning_rate": 3.585057414402959e-08,
      "loss": 0.0113,
      "step": 10803
    },
    {
      "epoch": 2.8972914990614105,
      "grad_norm": 0.27074453620563943,
      "learning_rate": 3.566431753644017e-08,
      "loss": 0.0134,
      "step": 10804
    },
    {
      "epoch": 2.897559667471172,
      "grad_norm": 0.30515959920856817,
      "learning_rate": 3.54785442930361e-08,
      "loss": 0.0163,
      "step": 10805
    },
    {
      "epoch": 2.8978278358809333,
      "grad_norm": 0.26337186650335676,
      "learning_rate": 3.529325443190568e-08,
      "loss": 0.0142,
      "step": 10806
    },
    {
      "epoch": 2.8980960042906947,
      "grad_norm": 0.19879095868834398,
      "learning_rate": 3.5108447971090034e-08,
      "loss": 0.0101,
      "step": 10807
    },
    {
      "epoch": 2.8983641727004557,
      "grad_norm": 0.17335105053057231,
      "learning_rate": 3.4924124928583104e-08,
      "loss": 0.0071,
      "step": 10808
    },
    {
      "epoch": 2.898632341110217,
      "grad_norm": 0.23356749881765212,
      "learning_rate": 3.47402853223322e-08,
      "loss": 0.0128,
      "step": 10809
    },
    {
      "epoch": 2.8989005095199785,
      "grad_norm": 0.3764161792507979,
      "learning_rate": 3.45569291702369e-08,
      "loss": 0.018,
      "step": 10810
    },
    {
      "epoch": 2.89916867792974,
      "grad_norm": 0.3370187586851141,
      "learning_rate": 3.437405649015069e-08,
      "loss": 0.0169,
      "step": 10811
    },
    {
      "epoch": 2.8994368463395013,
      "grad_norm": 0.3002358114283504,
      "learning_rate": 3.4191667299879325e-08,
      "loss": 0.0199,
      "step": 10812
    },
    {
      "epoch": 2.8997050147492627,
      "grad_norm": 0.29005365081811896,
      "learning_rate": 3.4009761617180834e-08,
      "loss": 0.0163,
      "step": 10813
    },
    {
      "epoch": 2.899973183159024,
      "grad_norm": 0.2410781304595417,
      "learning_rate": 3.382833945976771e-08,
      "loss": 0.0142,
      "step": 10814
    },
    {
      "epoch": 2.900241351568785,
      "grad_norm": 0.2224664254228026,
      "learning_rate": 3.364740084530416e-08,
      "loss": 0.0115,
      "step": 10815
    },
    {
      "epoch": 2.9005095199785464,
      "grad_norm": 0.28782405765399677,
      "learning_rate": 3.346694579140775e-08,
      "loss": 0.0141,
      "step": 10816
    },
    {
      "epoch": 2.900777688388308,
      "grad_norm": 0.22109900581821154,
      "learning_rate": 3.3286974315649426e-08,
      "loss": 0.0139,
      "step": 10817
    },
    {
      "epoch": 2.9010458567980693,
      "grad_norm": 0.27806366631288737,
      "learning_rate": 3.310748643555184e-08,
      "loss": 0.0177,
      "step": 10818
    },
    {
      "epoch": 2.9013140252078307,
      "grad_norm": 0.2773686658991523,
      "learning_rate": 3.292848216859101e-08,
      "loss": 0.0119,
      "step": 10819
    },
    {
      "epoch": 2.9015821936175916,
      "grad_norm": 0.1564384369181996,
      "learning_rate": 3.274996153219689e-08,
      "loss": 0.0069,
      "step": 10820
    },
    {
      "epoch": 2.901850362027353,
      "grad_norm": 0.272630106108349,
      "learning_rate": 3.257192454375113e-08,
      "loss": 0.0164,
      "step": 10821
    },
    {
      "epoch": 2.9021185304371144,
      "grad_norm": 0.22926231072659004,
      "learning_rate": 3.239437122058875e-08,
      "loss": 0.0149,
      "step": 10822
    },
    {
      "epoch": 2.902386698846876,
      "grad_norm": 0.17336560515504248,
      "learning_rate": 3.221730157999758e-08,
      "loss": 0.0087,
      "step": 10823
    },
    {
      "epoch": 2.9026548672566372,
      "grad_norm": 0.3225435136491405,
      "learning_rate": 3.204071563921885e-08,
      "loss": 0.0205,
      "step": 10824
    },
    {
      "epoch": 2.9029230356663986,
      "grad_norm": 0.2517478398980715,
      "learning_rate": 3.1864613415446e-08,
      "loss": 0.0127,
      "step": 10825
    },
    {
      "epoch": 2.90319120407616,
      "grad_norm": 0.2671185153747408,
      "learning_rate": 3.168899492582533e-08,
      "loss": 0.015,
      "step": 10826
    },
    {
      "epoch": 2.903459372485921,
      "grad_norm": 0.21847278256285665,
      "learning_rate": 3.1513860187457055e-08,
      "loss": 0.0119,
      "step": 10827
    },
    {
      "epoch": 2.9037275408956824,
      "grad_norm": 0.35113862020970865,
      "learning_rate": 3.133920921739253e-08,
      "loss": 0.0203,
      "step": 10828
    },
    {
      "epoch": 2.903995709305444,
      "grad_norm": 0.3425059752471659,
      "learning_rate": 3.1165042032638146e-08,
      "loss": 0.0202,
      "step": 10829
    },
    {
      "epoch": 2.9042638777152052,
      "grad_norm": 0.24174884084499454,
      "learning_rate": 3.099135865015146e-08,
      "loss": 0.0112,
      "step": 10830
    },
    {
      "epoch": 2.9045320461249666,
      "grad_norm": 0.23329382308367128,
      "learning_rate": 3.081815908684338e-08,
      "loss": 0.0113,
      "step": 10831
    },
    {
      "epoch": 2.9048002145347276,
      "grad_norm": 0.34952511048494866,
      "learning_rate": 3.0645443359578754e-08,
      "loss": 0.0192,
      "step": 10832
    },
    {
      "epoch": 2.905068382944489,
      "grad_norm": 0.24255461837975373,
      "learning_rate": 3.047321148517357e-08,
      "loss": 0.0097,
      "step": 10833
    },
    {
      "epoch": 2.9053365513542504,
      "grad_norm": 0.31755283098000875,
      "learning_rate": 3.030146348039831e-08,
      "loss": 0.027,
      "step": 10834
    },
    {
      "epoch": 2.905604719764012,
      "grad_norm": 0.2387300282402434,
      "learning_rate": 3.013019936197514e-08,
      "loss": 0.0131,
      "step": 10835
    },
    {
      "epoch": 2.905872888173773,
      "grad_norm": 0.29123761729236414,
      "learning_rate": 2.995941914657907e-08,
      "loss": 0.0204,
      "step": 10836
    },
    {
      "epoch": 2.9061410565835346,
      "grad_norm": 0.20668891821579455,
      "learning_rate": 2.9789122850840124e-08,
      "loss": 0.0088,
      "step": 10837
    },
    {
      "epoch": 2.906409224993296,
      "grad_norm": 0.19315548946213973,
      "learning_rate": 2.9619310491337814e-08,
      "loss": 0.0093,
      "step": 10838
    },
    {
      "epoch": 2.906677393403057,
      "grad_norm": 0.2291021266978299,
      "learning_rate": 2.9449982084607808e-08,
      "loss": 0.0119,
      "step": 10839
    },
    {
      "epoch": 2.9069455618128184,
      "grad_norm": 0.2940044090814101,
      "learning_rate": 2.9281137647135805e-08,
      "loss": 0.0158,
      "step": 10840
    },
    {
      "epoch": 2.90721373022258,
      "grad_norm": 0.2647232116255512,
      "learning_rate": 2.9112777195363096e-08,
      "loss": 0.0141,
      "step": 10841
    },
    {
      "epoch": 2.907481898632341,
      "grad_norm": 0.30093540365204746,
      "learning_rate": 2.894490074568157e-08,
      "loss": 0.0176,
      "step": 10842
    },
    {
      "epoch": 2.9077500670421026,
      "grad_norm": 0.5440335392178978,
      "learning_rate": 2.8777508314437043e-08,
      "loss": 0.0142,
      "step": 10843
    },
    {
      "epoch": 2.9080182354518636,
      "grad_norm": 0.27057289290393954,
      "learning_rate": 2.8610599917928694e-08,
      "loss": 0.0126,
      "step": 10844
    },
    {
      "epoch": 2.908286403861625,
      "grad_norm": 0.23844736325954108,
      "learning_rate": 2.8444175572406862e-08,
      "loss": 0.0111,
      "step": 10845
    },
    {
      "epoch": 2.9085545722713864,
      "grad_norm": 0.21932302395090356,
      "learning_rate": 2.8278235294076362e-08,
      "loss": 0.0114,
      "step": 10846
    },
    {
      "epoch": 2.9088227406811478,
      "grad_norm": 0.29483093832254936,
      "learning_rate": 2.8112779099094822e-08,
      "loss": 0.0138,
      "step": 10847
    },
    {
      "epoch": 2.909090909090909,
      "grad_norm": 0.3706939161842536,
      "learning_rate": 2.7947807003571582e-08,
      "loss": 0.0163,
      "step": 10848
    },
    {
      "epoch": 2.9093590775006706,
      "grad_norm": 0.23162637491168364,
      "learning_rate": 2.77833190235699e-08,
      "loss": 0.0148,
      "step": 10849
    },
    {
      "epoch": 2.909627245910432,
      "grad_norm": 0.21350993233796112,
      "learning_rate": 2.7619315175105856e-08,
      "loss": 0.0113,
      "step": 10850
    },
    {
      "epoch": 2.909895414320193,
      "grad_norm": 0.22518951800138112,
      "learning_rate": 2.7455795474147228e-08,
      "loss": 0.0131,
      "step": 10851
    },
    {
      "epoch": 2.9101635827299543,
      "grad_norm": 0.2963970757400302,
      "learning_rate": 2.7292759936615732e-08,
      "loss": 0.0173,
      "step": 10852
    },
    {
      "epoch": 2.9104317511397158,
      "grad_norm": 0.2506301399678004,
      "learning_rate": 2.713020857838644e-08,
      "loss": 0.0121,
      "step": 10853
    },
    {
      "epoch": 2.910699919549477,
      "grad_norm": 0.22642725002159597,
      "learning_rate": 2.696814141528614e-08,
      "loss": 0.0131,
      "step": 10854
    },
    {
      "epoch": 2.9109680879592386,
      "grad_norm": 0.2610499124594803,
      "learning_rate": 2.6806558463094433e-08,
      "loss": 0.0084,
      "step": 10855
    },
    {
      "epoch": 2.9112362563689995,
      "grad_norm": 0.23949194261336368,
      "learning_rate": 2.664545973754429e-08,
      "loss": 0.0121,
      "step": 10856
    },
    {
      "epoch": 2.911504424778761,
      "grad_norm": 0.21246806624092346,
      "learning_rate": 2.6484845254322045e-08,
      "loss": 0.0076,
      "step": 10857
    },
    {
      "epoch": 2.9117725931885223,
      "grad_norm": 0.20250600957057463,
      "learning_rate": 2.6324715029066307e-08,
      "loss": 0.0105,
      "step": 10858
    },
    {
      "epoch": 2.9120407615982837,
      "grad_norm": 0.23388056370842047,
      "learning_rate": 2.6165069077367932e-08,
      "loss": 0.0152,
      "step": 10859
    },
    {
      "epoch": 2.912308930008045,
      "grad_norm": 0.2653973236994618,
      "learning_rate": 2.6005907414771158e-08,
      "loss": 0.0123,
      "step": 10860
    },
    {
      "epoch": 2.9125770984178065,
      "grad_norm": 0.2628941638195506,
      "learning_rate": 2.584723005677414e-08,
      "loss": 0.0145,
      "step": 10861
    },
    {
      "epoch": 2.912845266827568,
      "grad_norm": 0.2157743822158722,
      "learning_rate": 2.5689037018825636e-08,
      "loss": 0.0118,
      "step": 10862
    },
    {
      "epoch": 2.913113435237329,
      "grad_norm": 0.25503115763105455,
      "learning_rate": 2.5531328316328875e-08,
      "loss": 0.0135,
      "step": 10863
    },
    {
      "epoch": 2.9133816036470903,
      "grad_norm": 0.24056010411234718,
      "learning_rate": 2.5374103964640464e-08,
      "loss": 0.0109,
      "step": 10864
    },
    {
      "epoch": 2.9136497720568517,
      "grad_norm": 0.26611635378582005,
      "learning_rate": 2.52173639790676e-08,
      "loss": 0.0162,
      "step": 10865
    },
    {
      "epoch": 2.913917940466613,
      "grad_norm": 0.6677466696572626,
      "learning_rate": 2.5061108374872523e-08,
      "loss": 0.0093,
      "step": 10866
    },
    {
      "epoch": 2.9141861088763745,
      "grad_norm": 0.24808190944493533,
      "learning_rate": 2.4905337167269173e-08,
      "loss": 0.0139,
      "step": 10867
    },
    {
      "epoch": 2.9144542772861355,
      "grad_norm": 0.27476715797803086,
      "learning_rate": 2.4750050371424304e-08,
      "loss": 0.0143,
      "step": 10868
    },
    {
      "epoch": 2.914722445695897,
      "grad_norm": 0.34318240794436666,
      "learning_rate": 2.459524800245805e-08,
      "loss": 0.019,
      "step": 10869
    },
    {
      "epoch": 2.9149906141056583,
      "grad_norm": 0.2696767831988475,
      "learning_rate": 2.4440930075443346e-08,
      "loss": 0.0159,
      "step": 10870
    },
    {
      "epoch": 2.9152587825154197,
      "grad_norm": 0.24835967669051537,
      "learning_rate": 2.4287096605404846e-08,
      "loss": 0.0144,
      "step": 10871
    },
    {
      "epoch": 2.915526950925181,
      "grad_norm": 0.26940980209753834,
      "learning_rate": 2.413374760732168e-08,
      "loss": 0.0121,
      "step": 10872
    },
    {
      "epoch": 2.9157951193349425,
      "grad_norm": 0.29826746230982826,
      "learning_rate": 2.398088309612523e-08,
      "loss": 0.0181,
      "step": 10873
    },
    {
      "epoch": 2.916063287744704,
      "grad_norm": 0.25385356638676215,
      "learning_rate": 2.3828503086698595e-08,
      "loss": 0.0134,
      "step": 10874
    },
    {
      "epoch": 2.916331456154465,
      "grad_norm": 0.22594750749665615,
      "learning_rate": 2.367660759387935e-08,
      "loss": 0.0102,
      "step": 10875
    },
    {
      "epoch": 2.9165996245642263,
      "grad_norm": 0.20442983103087703,
      "learning_rate": 2.3525196632457337e-08,
      "loss": 0.013,
      "step": 10876
    },
    {
      "epoch": 2.9168677929739877,
      "grad_norm": 0.19229739889544645,
      "learning_rate": 2.3374270217174087e-08,
      "loss": 0.0133,
      "step": 10877
    },
    {
      "epoch": 2.917135961383749,
      "grad_norm": 0.26002567860369435,
      "learning_rate": 2.3223828362725632e-08,
      "loss": 0.013,
      "step": 10878
    },
    {
      "epoch": 2.9174041297935105,
      "grad_norm": 0.24462049493545407,
      "learning_rate": 2.3073871083760247e-08,
      "loss": 0.0134,
      "step": 10879
    },
    {
      "epoch": 2.9176722982032715,
      "grad_norm": 0.2036830402708642,
      "learning_rate": 2.292439839487792e-08,
      "loss": 0.0115,
      "step": 10880
    },
    {
      "epoch": 2.917940466613033,
      "grad_norm": 0.26515560051268705,
      "learning_rate": 2.2775410310633682e-08,
      "loss": 0.0142,
      "step": 10881
    },
    {
      "epoch": 2.9182086350227943,
      "grad_norm": 0.25299142825611,
      "learning_rate": 2.2626906845533148e-08,
      "loss": 0.0124,
      "step": 10882
    },
    {
      "epoch": 2.9184768034325557,
      "grad_norm": 0.17581443722947993,
      "learning_rate": 2.2478888014035862e-08,
      "loss": 0.0119,
      "step": 10883
    },
    {
      "epoch": 2.918744971842317,
      "grad_norm": 0.40795838706052545,
      "learning_rate": 2.2331353830554183e-08,
      "loss": 0.0137,
      "step": 10884
    },
    {
      "epoch": 2.919013140252078,
      "grad_norm": 0.18418322698913572,
      "learning_rate": 2.2184304309453286e-08,
      "loss": 0.0088,
      "step": 10885
    },
    {
      "epoch": 2.91928130866184,
      "grad_norm": 0.27709397402148384,
      "learning_rate": 2.2037739465051166e-08,
      "loss": 0.0098,
      "step": 10886
    },
    {
      "epoch": 2.919549477071601,
      "grad_norm": 0.2802932302015498,
      "learning_rate": 2.189165931161752e-08,
      "loss": 0.0147,
      "step": 10887
    },
    {
      "epoch": 2.9198176454813622,
      "grad_norm": 0.179176681958024,
      "learning_rate": 2.1746063863377075e-08,
      "loss": 0.0086,
      "step": 10888
    },
    {
      "epoch": 2.9200858138911236,
      "grad_norm": 0.24416693066820558,
      "learning_rate": 2.160095313450461e-08,
      "loss": 0.012,
      "step": 10889
    },
    {
      "epoch": 2.920353982300885,
      "grad_norm": 0.18674075151576883,
      "learning_rate": 2.1456327139130484e-08,
      "loss": 0.0095,
      "step": 10890
    },
    {
      "epoch": 2.9206221507106465,
      "grad_norm": 0.23776588630764686,
      "learning_rate": 2.1312185891335658e-08,
      "loss": 0.0139,
      "step": 10891
    },
    {
      "epoch": 2.9208903191204074,
      "grad_norm": 0.2751537833052374,
      "learning_rate": 2.116852940515557e-08,
      "loss": 0.0187,
      "step": 10892
    },
    {
      "epoch": 2.921158487530169,
      "grad_norm": 0.21451154460290375,
      "learning_rate": 2.102535769457681e-08,
      "loss": 0.0118,
      "step": 10893
    },
    {
      "epoch": 2.9214266559399302,
      "grad_norm": 0.4785150139906985,
      "learning_rate": 2.0882670773540448e-08,
      "loss": 0.0188,
      "step": 10894
    },
    {
      "epoch": 2.9216948243496916,
      "grad_norm": 0.19762954433430227,
      "learning_rate": 2.0740468655938707e-08,
      "loss": 0.0102,
      "step": 10895
    },
    {
      "epoch": 2.921962992759453,
      "grad_norm": 0.2849525008582507,
      "learning_rate": 2.0598751355618284e-08,
      "loss": 0.0135,
      "step": 10896
    },
    {
      "epoch": 2.922231161169214,
      "grad_norm": 0.1794722805759241,
      "learning_rate": 2.0457518886377038e-08,
      "loss": 0.009,
      "step": 10897
    },
    {
      "epoch": 2.922499329578976,
      "grad_norm": 0.2756025572020325,
      "learning_rate": 2.031677126196674e-08,
      "loss": 0.0135,
      "step": 10898
    },
    {
      "epoch": 2.922767497988737,
      "grad_norm": 0.20799350811844905,
      "learning_rate": 2.017650849609143e-08,
      "loss": 0.0098,
      "step": 10899
    },
    {
      "epoch": 2.923035666398498,
      "grad_norm": 0.22144873139322496,
      "learning_rate": 2.0036730602408518e-08,
      "loss": 0.0155,
      "step": 10900
    },
    {
      "epoch": 2.9233038348082596,
      "grad_norm": 0.2513534599302159,
      "learning_rate": 1.9897437594528223e-08,
      "loss": 0.009,
      "step": 10901
    },
    {
      "epoch": 2.923572003218021,
      "grad_norm": 0.24681839379200368,
      "learning_rate": 1.975862948601137e-08,
      "loss": 0.0161,
      "step": 10902
    },
    {
      "epoch": 2.9238401716277824,
      "grad_norm": 0.2671882114915564,
      "learning_rate": 1.962030629037548e-08,
      "loss": 0.0122,
      "step": 10903
    },
    {
      "epoch": 2.9241083400375434,
      "grad_norm": 0.2627288788966217,
      "learning_rate": 1.9482468021087553e-08,
      "loss": 0.0124,
      "step": 10904
    },
    {
      "epoch": 2.924376508447305,
      "grad_norm": 0.2660958897987377,
      "learning_rate": 1.9345114691567966e-08,
      "loss": 0.0124,
      "step": 10905
    },
    {
      "epoch": 2.924644676857066,
      "grad_norm": 0.23112443387305495,
      "learning_rate": 1.9208246315192135e-08,
      "loss": 0.0148,
      "step": 10906
    },
    {
      "epoch": 2.9249128452668276,
      "grad_norm": 0.24064417362208038,
      "learning_rate": 1.9071862905284954e-08,
      "loss": 0.0134,
      "step": 10907
    },
    {
      "epoch": 2.925181013676589,
      "grad_norm": 0.2940959148897896,
      "learning_rate": 1.8935964475126912e-08,
      "loss": 0.0124,
      "step": 10908
    },
    {
      "epoch": 2.92544918208635,
      "grad_norm": 0.19432312501761773,
      "learning_rate": 1.8800551037949088e-08,
      "loss": 0.0079,
      "step": 10909
    },
    {
      "epoch": 2.925717350496112,
      "grad_norm": 0.2884075534348664,
      "learning_rate": 1.8665622606936497e-08,
      "loss": 0.0143,
      "step": 10910
    },
    {
      "epoch": 2.9259855189058728,
      "grad_norm": 0.22942145243531906,
      "learning_rate": 1.8531179195227512e-08,
      "loss": 0.0085,
      "step": 10911
    },
    {
      "epoch": 2.926253687315634,
      "grad_norm": 0.21690642565860213,
      "learning_rate": 1.839722081591222e-08,
      "loss": 0.0141,
      "step": 10912
    },
    {
      "epoch": 2.9265218557253956,
      "grad_norm": 0.24019350793914943,
      "learning_rate": 1.8263747482033522e-08,
      "loss": 0.0143,
      "step": 10913
    },
    {
      "epoch": 2.926790024135157,
      "grad_norm": 0.23615444280569983,
      "learning_rate": 1.8130759206587135e-08,
      "loss": 0.0128,
      "step": 10914
    },
    {
      "epoch": 2.9270581925449184,
      "grad_norm": 0.21250310265643144,
      "learning_rate": 1.7998256002522695e-08,
      "loss": 0.0128,
      "step": 10915
    },
    {
      "epoch": 2.9273263609546794,
      "grad_norm": 0.4299236312907545,
      "learning_rate": 1.7866237882740444e-08,
      "loss": 0.0198,
      "step": 10916
    },
    {
      "epoch": 2.9275945293644408,
      "grad_norm": 0.334468239435671,
      "learning_rate": 1.7734704860095652e-08,
      "loss": 0.019,
      "step": 10917
    },
    {
      "epoch": 2.927862697774202,
      "grad_norm": 0.23012294254895008,
      "learning_rate": 1.7603656947395297e-08,
      "loss": 0.0139,
      "step": 10918
    },
    {
      "epoch": 2.9281308661839636,
      "grad_norm": 0.2256914893620224,
      "learning_rate": 1.747309415739806e-08,
      "loss": 0.0147,
      "step": 10919
    },
    {
      "epoch": 2.928399034593725,
      "grad_norm": 0.3197902013077106,
      "learning_rate": 1.7343016502817667e-08,
      "loss": 0.0141,
      "step": 10920
    },
    {
      "epoch": 2.928667203003486,
      "grad_norm": 0.22949222162755775,
      "learning_rate": 1.7213423996319532e-08,
      "loss": 0.0094,
      "step": 10921
    },
    {
      "epoch": 2.928935371413248,
      "grad_norm": 0.2601219426063491,
      "learning_rate": 1.7084316650520795e-08,
      "loss": 0.015,
      "step": 10922
    },
    {
      "epoch": 2.9292035398230087,
      "grad_norm": 0.27709697234096875,
      "learning_rate": 1.6955694477993055e-08,
      "loss": 0.015,
      "step": 10923
    },
    {
      "epoch": 2.92947170823277,
      "grad_norm": 0.35969542260521226,
      "learning_rate": 1.6827557491259083e-08,
      "loss": 0.0136,
      "step": 10924
    },
    {
      "epoch": 2.9297398766425315,
      "grad_norm": 0.2777014455834459,
      "learning_rate": 1.6699905702796114e-08,
      "loss": 0.0183,
      "step": 10925
    },
    {
      "epoch": 2.930008045052293,
      "grad_norm": 0.2942994911345897,
      "learning_rate": 1.65727391250331e-08,
      "loss": 0.0203,
      "step": 10926
    },
    {
      "epoch": 2.9302762134620544,
      "grad_norm": 0.24350234493024012,
      "learning_rate": 1.6446057770351798e-08,
      "loss": 0.0113,
      "step": 10927
    },
    {
      "epoch": 2.9305443818718153,
      "grad_norm": 0.25727247506285983,
      "learning_rate": 1.6319861651086233e-08,
      "loss": 0.015,
      "step": 10928
    },
    {
      "epoch": 2.9308125502815767,
      "grad_norm": 0.21657136194624046,
      "learning_rate": 1.6194150779524353e-08,
      "loss": 0.0105,
      "step": 10929
    },
    {
      "epoch": 2.931080718691338,
      "grad_norm": 0.23798903100143778,
      "learning_rate": 1.6068925167906923e-08,
      "loss": 0.0127,
      "step": 10930
    },
    {
      "epoch": 2.9313488871010995,
      "grad_norm": 0.23611045165197303,
      "learning_rate": 1.5944184828425856e-08,
      "loss": 0.0133,
      "step": 10931
    },
    {
      "epoch": 2.931617055510861,
      "grad_norm": 0.28977469091433855,
      "learning_rate": 1.581992977322644e-08,
      "loss": 0.0132,
      "step": 10932
    },
    {
      "epoch": 2.931885223920622,
      "grad_norm": 0.3281394587405311,
      "learning_rate": 1.5696160014408436e-08,
      "loss": 0.0149,
      "step": 10933
    },
    {
      "epoch": 2.9321533923303837,
      "grad_norm": 0.34621639839528223,
      "learning_rate": 1.5572875564022206e-08,
      "loss": 0.0129,
      "step": 10934
    },
    {
      "epoch": 2.9324215607401447,
      "grad_norm": 0.20540521067679937,
      "learning_rate": 1.545007643407148e-08,
      "loss": 0.008,
      "step": 10935
    },
    {
      "epoch": 2.932689729149906,
      "grad_norm": 0.26270790769104435,
      "learning_rate": 1.5327762636513367e-08,
      "loss": 0.0118,
      "step": 10936
    },
    {
      "epoch": 2.9329578975596675,
      "grad_norm": 0.20646058657565122,
      "learning_rate": 1.5205934183256666e-08,
      "loss": 0.01,
      "step": 10937
    },
    {
      "epoch": 2.933226065969429,
      "grad_norm": 0.2165060735920053,
      "learning_rate": 1.5084591086164113e-08,
      "loss": 0.01,
      "step": 10938
    },
    {
      "epoch": 2.9334942343791903,
      "grad_norm": 0.2731529205126739,
      "learning_rate": 1.4963733357050147e-08,
      "loss": 0.012,
      "step": 10939
    },
    {
      "epoch": 2.9337624027889513,
      "grad_norm": 0.24638938825315598,
      "learning_rate": 1.4843361007682022e-08,
      "loss": 0.0154,
      "step": 10940
    },
    {
      "epoch": 2.9340305711987127,
      "grad_norm": 0.21061866867507942,
      "learning_rate": 1.4723474049780917e-08,
      "loss": 0.013,
      "step": 10941
    },
    {
      "epoch": 2.934298739608474,
      "grad_norm": 0.23984387495920148,
      "learning_rate": 1.4604072495019161e-08,
      "loss": 0.0112,
      "step": 10942
    },
    {
      "epoch": 2.9345669080182355,
      "grad_norm": 0.24114268966738003,
      "learning_rate": 1.4485156355023567e-08,
      "loss": 0.0127,
      "step": 10943
    },
    {
      "epoch": 2.934835076427997,
      "grad_norm": 0.30193331098843473,
      "learning_rate": 1.4366725641370982e-08,
      "loss": 0.0182,
      "step": 10944
    },
    {
      "epoch": 2.935103244837758,
      "grad_norm": 0.29895661705761783,
      "learning_rate": 1.4248780365594405e-08,
      "loss": 0.0167,
      "step": 10945
    },
    {
      "epoch": 2.9353714132475197,
      "grad_norm": 0.3266412504164508,
      "learning_rate": 1.4131320539177428e-08,
      "loss": 0.0146,
      "step": 10946
    },
    {
      "epoch": 2.9356395816572807,
      "grad_norm": 0.41820298922911775,
      "learning_rate": 1.4014346173555904e-08,
      "loss": 0.0244,
      "step": 10947
    },
    {
      "epoch": 2.935907750067042,
      "grad_norm": 0.25497468860597533,
      "learning_rate": 1.3897857280120164e-08,
      "loss": 0.0137,
      "step": 10948
    },
    {
      "epoch": 2.9361759184768035,
      "grad_norm": 0.26745734399606585,
      "learning_rate": 1.3781853870212247e-08,
      "loss": 0.013,
      "step": 10949
    },
    {
      "epoch": 2.936444086886565,
      "grad_norm": 0.25223436541181116,
      "learning_rate": 1.3666335955127008e-08,
      "loss": 0.0136,
      "step": 10950
    },
    {
      "epoch": 2.9367122552963263,
      "grad_norm": 0.21549469422500683,
      "learning_rate": 1.3551303546112671e-08,
      "loss": 0.0125,
      "step": 10951
    },
    {
      "epoch": 2.9369804237060873,
      "grad_norm": 0.5091830238050425,
      "learning_rate": 1.343675665436861e-08,
      "loss": 0.0169,
      "step": 10952
    },
    {
      "epoch": 2.9372485921158487,
      "grad_norm": 0.2150053677097419,
      "learning_rate": 1.3322695291048682e-08,
      "loss": 0.0106,
      "step": 10953
    },
    {
      "epoch": 2.93751676052561,
      "grad_norm": 0.24212300712390375,
      "learning_rate": 1.3209119467258447e-08,
      "loss": 0.0125,
      "step": 10954
    },
    {
      "epoch": 2.9377849289353715,
      "grad_norm": 0.24003109370612688,
      "learning_rate": 1.3096029194056836e-08,
      "loss": 0.0125,
      "step": 10955
    },
    {
      "epoch": 2.938053097345133,
      "grad_norm": 0.29986676349867614,
      "learning_rate": 1.2983424482454488e-08,
      "loss": 0.0212,
      "step": 10956
    },
    {
      "epoch": 2.938321265754894,
      "grad_norm": 0.23085016194647184,
      "learning_rate": 1.2871305343415964e-08,
      "loss": 0.0105,
      "step": 10957
    },
    {
      "epoch": 2.9385894341646557,
      "grad_norm": 0.22637526893915144,
      "learning_rate": 1.275967178785753e-08,
      "loss": 0.0124,
      "step": 10958
    },
    {
      "epoch": 2.9388576025744166,
      "grad_norm": 0.2821899239517504,
      "learning_rate": 1.2648523826649384e-08,
      "loss": 0.014,
      "step": 10959
    },
    {
      "epoch": 2.939125770984178,
      "grad_norm": 0.22929435166813625,
      "learning_rate": 1.2537861470613421e-08,
      "loss": 0.012,
      "step": 10960
    },
    {
      "epoch": 2.9393939393939394,
      "grad_norm": 0.22523382883723653,
      "learning_rate": 1.2427684730524358e-08,
      "loss": 0.0123,
      "step": 10961
    },
    {
      "epoch": 2.939662107803701,
      "grad_norm": 0.23359015417517265,
      "learning_rate": 1.2317993617109724e-08,
      "loss": 0.0106,
      "step": 10962
    },
    {
      "epoch": 2.9399302762134623,
      "grad_norm": 0.3021731993023363,
      "learning_rate": 1.2208788141049866e-08,
      "loss": 0.0157,
      "step": 10963
    },
    {
      "epoch": 2.940198444623223,
      "grad_norm": 0.20889239751945007,
      "learning_rate": 1.2100068312978496e-08,
      "loss": 0.0143,
      "step": 10964
    },
    {
      "epoch": 2.9404666130329846,
      "grad_norm": 0.30111804872513614,
      "learning_rate": 1.1991834143480485e-08,
      "loss": 0.0181,
      "step": 10965
    },
    {
      "epoch": 2.940734781442746,
      "grad_norm": 0.3000648374294634,
      "learning_rate": 1.1884085643094623e-08,
      "loss": 0.0187,
      "step": 10966
    },
    {
      "epoch": 2.9410029498525074,
      "grad_norm": 0.2225868173738058,
      "learning_rate": 1.1776822822312517e-08,
      "loss": 0.0097,
      "step": 10967
    },
    {
      "epoch": 2.941271118262269,
      "grad_norm": 0.21643714942560044,
      "learning_rate": 1.1670045691577481e-08,
      "loss": 0.0129,
      "step": 10968
    },
    {
      "epoch": 2.94153928667203,
      "grad_norm": 0.2557999911459049,
      "learning_rate": 1.1563754261286198e-08,
      "loss": 0.0165,
      "step": 10969
    },
    {
      "epoch": 2.941807455081791,
      "grad_norm": 0.2465186903940971,
      "learning_rate": 1.1457948541788166e-08,
      "loss": 0.0118,
      "step": 10970
    },
    {
      "epoch": 2.9420756234915526,
      "grad_norm": 0.252442814789678,
      "learning_rate": 1.1352628543385702e-08,
      "loss": 0.0126,
      "step": 10971
    },
    {
      "epoch": 2.942343791901314,
      "grad_norm": 0.1883836758101972,
      "learning_rate": 1.1247794276332824e-08,
      "loss": 0.0091,
      "step": 10972
    },
    {
      "epoch": 2.9426119603110754,
      "grad_norm": 0.19540019219023644,
      "learning_rate": 1.1143445750837479e-08,
      "loss": 0.01,
      "step": 10973
    },
    {
      "epoch": 2.942880128720837,
      "grad_norm": 0.19583168607066823,
      "learning_rate": 1.103958297705987e-08,
      "loss": 0.0075,
      "step": 10974
    },
    {
      "epoch": 2.9431482971305982,
      "grad_norm": 0.23055118245359696,
      "learning_rate": 1.0936205965111913e-08,
      "loss": 0.0096,
      "step": 10975
    },
    {
      "epoch": 2.943416465540359,
      "grad_norm": 0.19322218084880466,
      "learning_rate": 1.0833314725060551e-08,
      "loss": 0.0111,
      "step": 10976
    },
    {
      "epoch": 2.9436846339501206,
      "grad_norm": 0.23140029168761792,
      "learning_rate": 1.0730909266923328e-08,
      "loss": 0.0154,
      "step": 10977
    },
    {
      "epoch": 2.943952802359882,
      "grad_norm": 0.2212576074103909,
      "learning_rate": 1.0628989600671158e-08,
      "loss": 0.012,
      "step": 10978
    },
    {
      "epoch": 2.9442209707696434,
      "grad_norm": 0.27311872120108416,
      "learning_rate": 1.0527555736227213e-08,
      "loss": 0.015,
      "step": 10979
    },
    {
      "epoch": 2.944489139179405,
      "grad_norm": 0.2643609744650085,
      "learning_rate": 1.0426607683469148e-08,
      "loss": 0.0155,
      "step": 10980
    },
    {
      "epoch": 2.9447573075891658,
      "grad_norm": 0.16364578439557606,
      "learning_rate": 1.0326145452224656e-08,
      "loss": 0.0066,
      "step": 10981
    },
    {
      "epoch": 2.945025475998927,
      "grad_norm": 0.2218760926935012,
      "learning_rate": 1.0226169052276469e-08,
      "loss": 0.0151,
      "step": 10982
    },
    {
      "epoch": 2.9452936444086886,
      "grad_norm": 0.21742735518511308,
      "learning_rate": 1.0126678493358466e-08,
      "loss": 0.0118,
      "step": 10983
    },
    {
      "epoch": 2.94556181281845,
      "grad_norm": 0.2341012132103654,
      "learning_rate": 1.0027673785157899e-08,
      "loss": 0.0118,
      "step": 10984
    },
    {
      "epoch": 2.9458299812282114,
      "grad_norm": 0.29852071709328143,
      "learning_rate": 9.929154937314278e-09,
      "loss": 0.0143,
      "step": 10985
    },
    {
      "epoch": 2.946098149637973,
      "grad_norm": 0.26944244971715664,
      "learning_rate": 9.831121959421042e-09,
      "loss": 0.012,
      "step": 10986
    },
    {
      "epoch": 2.946366318047734,
      "grad_norm": 0.268586108013912,
      "learning_rate": 9.733574861022222e-09,
      "loss": 0.0149,
      "step": 10987
    },
    {
      "epoch": 2.946634486457495,
      "grad_norm": 0.2593955399371402,
      "learning_rate": 9.636513651616885e-09,
      "loss": 0.0197,
      "step": 10988
    },
    {
      "epoch": 2.9469026548672566,
      "grad_norm": 0.22545787927317099,
      "learning_rate": 9.539938340654698e-09,
      "loss": 0.0134,
      "step": 10989
    },
    {
      "epoch": 2.947170823277018,
      "grad_norm": 0.2043549721156079,
      "learning_rate": 9.443848937539246e-09,
      "loss": 0.012,
      "step": 10990
    },
    {
      "epoch": 2.9474389916867794,
      "grad_norm": 0.39302996078005364,
      "learning_rate": 9.348245451626935e-09,
      "loss": 0.0093,
      "step": 10991
    },
    {
      "epoch": 2.9477071600965408,
      "grad_norm": 0.29771201896700183,
      "learning_rate": 9.253127892225877e-09,
      "loss": 0.0121,
      "step": 10992
    },
    {
      "epoch": 2.9479753285063017,
      "grad_norm": 0.20806591628705107,
      "learning_rate": 9.15849626859755e-09,
      "loss": 0.0107,
      "step": 10993
    },
    {
      "epoch": 2.948243496916063,
      "grad_norm": 0.2608632504168459,
      "learning_rate": 9.064350589955695e-09,
      "loss": 0.0121,
      "step": 10994
    },
    {
      "epoch": 2.9485116653258245,
      "grad_norm": 0.21976282886729137,
      "learning_rate": 8.97069086546798e-09,
      "loss": 0.0125,
      "step": 10995
    },
    {
      "epoch": 2.948779833735586,
      "grad_norm": 0.2243083650198348,
      "learning_rate": 8.877517104252665e-09,
      "loss": 0.0129,
      "step": 10996
    },
    {
      "epoch": 2.9490480021453473,
      "grad_norm": 0.3116338752155259,
      "learning_rate": 8.784829315382493e-09,
      "loss": 0.0172,
      "step": 10997
    },
    {
      "epoch": 2.9493161705551088,
      "grad_norm": 0.1813790332835123,
      "learning_rate": 8.692627507881912e-09,
      "loss": 0.0084,
      "step": 10998
    },
    {
      "epoch": 2.94958433896487,
      "grad_norm": 0.20252986270411516,
      "learning_rate": 8.600911690728742e-09,
      "loss": 0.0123,
      "step": 10999
    },
    {
      "epoch": 2.949852507374631,
      "grad_norm": 0.28292403732637544,
      "learning_rate": 8.50968187285306e-09,
      "loss": 0.014,
      "step": 11000
    },
    {
      "epoch": 2.9501206757843925,
      "grad_norm": 0.2920487919435454,
      "learning_rate": 8.418938063137205e-09,
      "loss": 0.0118,
      "step": 11001
    },
    {
      "epoch": 2.950388844194154,
      "grad_norm": 0.32111722081115657,
      "learning_rate": 8.328680270416888e-09,
      "loss": 0.0126,
      "step": 11002
    },
    {
      "epoch": 2.9506570126039153,
      "grad_norm": 0.26180813662921765,
      "learning_rate": 8.238908503480636e-09,
      "loss": 0.0139,
      "step": 11003
    },
    {
      "epoch": 2.9509251810136767,
      "grad_norm": 0.3055862399151533,
      "learning_rate": 8.149622771068677e-09,
      "loss": 0.0139,
      "step": 11004
    },
    {
      "epoch": 2.9511933494234377,
      "grad_norm": 0.18615916869483817,
      "learning_rate": 8.06082308187517e-09,
      "loss": 0.0102,
      "step": 11005
    },
    {
      "epoch": 2.951461517833199,
      "grad_norm": 0.18563356252299068,
      "learning_rate": 7.972509444545973e-09,
      "loss": 0.0091,
      "step": 11006
    },
    {
      "epoch": 2.9517296862429605,
      "grad_norm": 0.23022551686459858,
      "learning_rate": 7.884681867679766e-09,
      "loss": 0.0134,
      "step": 11007
    },
    {
      "epoch": 2.951997854652722,
      "grad_norm": 0.21354139216914916,
      "learning_rate": 7.797340359828043e-09,
      "loss": 0.0181,
      "step": 11008
    },
    {
      "epoch": 2.9522660230624833,
      "grad_norm": 0.24139668495791866,
      "learning_rate": 7.710484929495666e-09,
      "loss": 0.0139,
      "step": 11009
    },
    {
      "epoch": 2.9525341914722447,
      "grad_norm": 0.22657338054194212,
      "learning_rate": 7.624115585139202e-09,
      "loss": 0.011,
      "step": 11010
    },
    {
      "epoch": 2.952802359882006,
      "grad_norm": 0.29407794008679367,
      "learning_rate": 7.538232335167483e-09,
      "loss": 0.012,
      "step": 11011
    },
    {
      "epoch": 2.953070528291767,
      "grad_norm": 0.23532974648279006,
      "learning_rate": 7.452835187943819e-09,
      "loss": 0.0116,
      "step": 11012
    },
    {
      "epoch": 2.9533386967015285,
      "grad_norm": 0.23972060940016243,
      "learning_rate": 7.3679241517826685e-09,
      "loss": 0.0075,
      "step": 11013
    },
    {
      "epoch": 2.95360686511129,
      "grad_norm": 0.26350688155962315,
      "learning_rate": 7.2834992349513075e-09,
      "loss": 0.0142,
      "step": 11014
    },
    {
      "epoch": 2.9538750335210513,
      "grad_norm": 0.29370899842408016,
      "learning_rate": 7.199560445670384e-09,
      "loss": 0.0141,
      "step": 11015
    },
    {
      "epoch": 2.9541432019308127,
      "grad_norm": 0.3539727969652111,
      "learning_rate": 7.116107792112803e-09,
      "loss": 0.0188,
      "step": 11016
    },
    {
      "epoch": 2.9544113703405737,
      "grad_norm": 0.2068094474418927,
      "learning_rate": 7.033141282403733e-09,
      "loss": 0.0109,
      "step": 11017
    },
    {
      "epoch": 2.954679538750335,
      "grad_norm": 0.27576145864803947,
      "learning_rate": 6.950660924621156e-09,
      "loss": 0.0167,
      "step": 11018
    },
    {
      "epoch": 2.9549477071600965,
      "grad_norm": 0.2496861556096474,
      "learning_rate": 6.86866672679698e-09,
      "loss": 0.014,
      "step": 11019
    },
    {
      "epoch": 2.955215875569858,
      "grad_norm": 0.261069895226711,
      "learning_rate": 6.787158696914265e-09,
      "loss": 0.0145,
      "step": 11020
    },
    {
      "epoch": 2.9554840439796193,
      "grad_norm": 0.1885906194524536,
      "learning_rate": 6.706136842909439e-09,
      "loss": 0.0087,
      "step": 11021
    },
    {
      "epoch": 2.9557522123893807,
      "grad_norm": 0.2320540735531457,
      "learning_rate": 6.625601172670637e-09,
      "loss": 0.0102,
      "step": 11022
    },
    {
      "epoch": 2.956020380799142,
      "grad_norm": 0.3001526439293878,
      "learning_rate": 6.54555169403992e-09,
      "loss": 0.0179,
      "step": 11023
    },
    {
      "epoch": 2.956288549208903,
      "grad_norm": 0.29231508712076265,
      "learning_rate": 6.465988414811608e-09,
      "loss": 0.0164,
      "step": 11024
    },
    {
      "epoch": 2.9565567176186645,
      "grad_norm": 0.2089858622079921,
      "learning_rate": 6.386911342732283e-09,
      "loss": 0.0111,
      "step": 11025
    },
    {
      "epoch": 2.956824886028426,
      "grad_norm": 0.456812526477083,
      "learning_rate": 6.308320485501895e-09,
      "loss": 0.0122,
      "step": 11026
    },
    {
      "epoch": 2.9570930544381873,
      "grad_norm": 0.22100851311188455,
      "learning_rate": 6.230215850772103e-09,
      "loss": 0.0105,
      "step": 11027
    },
    {
      "epoch": 2.9573612228479487,
      "grad_norm": 0.2692792636774736,
      "learning_rate": 6.1525974461479345e-09,
      "loss": 0.0155,
      "step": 11028
    },
    {
      "epoch": 2.9576293912577096,
      "grad_norm": 0.16411345040323266,
      "learning_rate": 6.075465279187232e-09,
      "loss": 0.0077,
      "step": 11029
    },
    {
      "epoch": 2.957897559667471,
      "grad_norm": 0.22520368159005086,
      "learning_rate": 5.998819357399543e-09,
      "loss": 0.0129,
      "step": 11030
    },
    {
      "epoch": 2.9581657280772324,
      "grad_norm": 0.28934349194706493,
      "learning_rate": 5.9226596882483445e-09,
      "loss": 0.0164,
      "step": 11031
    },
    {
      "epoch": 2.958433896486994,
      "grad_norm": 0.2595500065451148,
      "learning_rate": 5.846986279148259e-09,
      "loss": 0.0142,
      "step": 11032
    },
    {
      "epoch": 2.9587020648967552,
      "grad_norm": 0.15325889819975147,
      "learning_rate": 5.771799137468392e-09,
      "loss": 0.0073,
      "step": 11033
    },
    {
      "epoch": 2.9589702333065167,
      "grad_norm": 0.2484839551954366,
      "learning_rate": 5.697098270529e-09,
      "loss": 0.0147,
      "step": 11034
    },
    {
      "epoch": 2.959238401716278,
      "grad_norm": 0.327708875680215,
      "learning_rate": 5.622883685603153e-09,
      "loss": 0.0367,
      "step": 11035
    },
    {
      "epoch": 2.959506570126039,
      "grad_norm": 0.30876315799378196,
      "learning_rate": 5.549155389917848e-09,
      "loss": 0.0153,
      "step": 11036
    },
    {
      "epoch": 2.9597747385358004,
      "grad_norm": 0.23619310520640383,
      "learning_rate": 5.475913390650678e-09,
      "loss": 0.0144,
      "step": 11037
    },
    {
      "epoch": 2.960042906945562,
      "grad_norm": 0.2035711117694188,
      "learning_rate": 5.403157694933714e-09,
      "loss": 0.0117,
      "step": 11038
    },
    {
      "epoch": 2.9603110753553232,
      "grad_norm": 0.26458460359750696,
      "learning_rate": 5.33088830985129e-09,
      "loss": 0.014,
      "step": 11039
    },
    {
      "epoch": 2.9605792437650846,
      "grad_norm": 0.2328340881568737,
      "learning_rate": 5.259105242439444e-09,
      "loss": 0.0104,
      "step": 11040
    },
    {
      "epoch": 2.9608474121748456,
      "grad_norm": 0.25267594743758226,
      "learning_rate": 5.1878084996875855e-09,
      "loss": 0.013,
      "step": 11041
    },
    {
      "epoch": 2.961115580584607,
      "grad_norm": 0.22293967678766832,
      "learning_rate": 5.116998088537939e-09,
      "loss": 0.0113,
      "step": 11042
    },
    {
      "epoch": 2.9613837489943684,
      "grad_norm": 0.21258786732186968,
      "learning_rate": 5.0466740158849895e-09,
      "loss": 0.013,
      "step": 11043
    },
    {
      "epoch": 2.96165191740413,
      "grad_norm": 0.2267627508780837,
      "learning_rate": 4.976836288576592e-09,
      "loss": 0.0094,
      "step": 11044
    },
    {
      "epoch": 2.961920085813891,
      "grad_norm": 0.2795388000207788,
      "learning_rate": 4.907484913411198e-09,
      "loss": 0.0129,
      "step": 11045
    },
    {
      "epoch": 2.9621882542236526,
      "grad_norm": 0.18654059103612475,
      "learning_rate": 4.838619897142849e-09,
      "loss": 0.0109,
      "step": 11046
    },
    {
      "epoch": 2.962456422633414,
      "grad_norm": 0.22663611981306914,
      "learning_rate": 4.7702412464761815e-09,
      "loss": 0.0099,
      "step": 11047
    },
    {
      "epoch": 2.962724591043175,
      "grad_norm": 0.23270388261886116,
      "learning_rate": 4.702348968069204e-09,
      "loss": 0.0144,
      "step": 11048
    },
    {
      "epoch": 2.9629927594529364,
      "grad_norm": 0.2219330575825363,
      "learning_rate": 4.634943068531627e-09,
      "loss": 0.0129,
      "step": 11049
    },
    {
      "epoch": 2.963260927862698,
      "grad_norm": 0.261189059244269,
      "learning_rate": 4.568023554427647e-09,
      "loss": 0.0154,
      "step": 11050
    },
    {
      "epoch": 2.963529096272459,
      "grad_norm": 0.25025273335962905,
      "learning_rate": 4.501590432272607e-09,
      "loss": 0.0133,
      "step": 11051
    },
    {
      "epoch": 2.9637972646822206,
      "grad_norm": 0.2554766721300398,
      "learning_rate": 4.435643708534665e-09,
      "loss": 0.014,
      "step": 11052
    },
    {
      "epoch": 2.9640654330919816,
      "grad_norm": 0.20598332245553708,
      "learning_rate": 4.370183389635352e-09,
      "loss": 0.0137,
      "step": 11053
    },
    {
      "epoch": 2.964333601501743,
      "grad_norm": 0.24057108564672777,
      "learning_rate": 4.305209481948458e-09,
      "loss": 0.0119,
      "step": 11054
    },
    {
      "epoch": 2.9646017699115044,
      "grad_norm": 0.19577140392189143,
      "learning_rate": 4.240721991799479e-09,
      "loss": 0.0116,
      "step": 11055
    },
    {
      "epoch": 2.9648699383212658,
      "grad_norm": 0.28393335385980295,
      "learning_rate": 4.176720925467837e-09,
      "loss": 0.0126,
      "step": 11056
    },
    {
      "epoch": 2.965138106731027,
      "grad_norm": 0.27124130629002036,
      "learning_rate": 4.113206289185767e-09,
      "loss": 0.0151,
      "step": 11057
    },
    {
      "epoch": 2.965406275140788,
      "grad_norm": 0.2548893394015401,
      "learning_rate": 4.050178089136658e-09,
      "loss": 0.0115,
      "step": 11058
    },
    {
      "epoch": 2.96567444355055,
      "grad_norm": 0.20846308078470122,
      "learning_rate": 3.987636331457822e-09,
      "loss": 0.0115,
      "step": 11059
    },
    {
      "epoch": 2.965942611960311,
      "grad_norm": 0.2724776133684185,
      "learning_rate": 3.925581022238279e-09,
      "loss": 0.0154,
      "step": 11060
    },
    {
      "epoch": 2.9662107803700724,
      "grad_norm": 0.22127555164722354,
      "learning_rate": 3.86401216752097e-09,
      "loss": 0.0119,
      "step": 11061
    },
    {
      "epoch": 2.9664789487798338,
      "grad_norm": 0.23543738384679128,
      "learning_rate": 3.802929773300546e-09,
      "loss": 0.0162,
      "step": 11062
    },
    {
      "epoch": 2.966747117189595,
      "grad_norm": 0.23469447189839013,
      "learning_rate": 3.742333845523916e-09,
      "loss": 0.012,
      "step": 11063
    },
    {
      "epoch": 2.9670152855993566,
      "grad_norm": 0.3869669479271844,
      "learning_rate": 3.68222439009136e-09,
      "loss": 0.0221,
      "step": 11064
    },
    {
      "epoch": 2.9672834540091175,
      "grad_norm": 0.3842178041331062,
      "learning_rate": 3.6226014128554198e-09,
      "loss": 0.0092,
      "step": 11065
    },
    {
      "epoch": 2.967551622418879,
      "grad_norm": 0.3102696991711149,
      "learning_rate": 3.563464919622006e-09,
      "loss": 0.0165,
      "step": 11066
    },
    {
      "epoch": 2.9678197908286403,
      "grad_norm": 0.26073251863966534,
      "learning_rate": 3.5048149161487356e-09,
      "loss": 0.0158,
      "step": 11067
    },
    {
      "epoch": 2.9680879592384017,
      "grad_norm": 0.25194678460377523,
      "learning_rate": 3.44665140814604e-09,
      "loss": 0.0139,
      "step": 11068
    },
    {
      "epoch": 2.968356127648163,
      "grad_norm": 0.23923012932064258,
      "learning_rate": 3.3889744012771676e-09,
      "loss": 0.011,
      "step": 11069
    },
    {
      "epoch": 2.968624296057924,
      "grad_norm": 0.2595424809639399,
      "learning_rate": 3.331783901158736e-09,
      "loss": 0.0148,
      "step": 11070
    },
    {
      "epoch": 2.968892464467686,
      "grad_norm": 0.21031684967431186,
      "learning_rate": 3.275079913357959e-09,
      "loss": 0.0107,
      "step": 11071
    },
    {
      "epoch": 2.969160632877447,
      "grad_norm": 0.23175539703954384,
      "learning_rate": 3.218862443397086e-09,
      "loss": 0.0091,
      "step": 11072
    },
    {
      "epoch": 2.9694288012872083,
      "grad_norm": 0.280362620482897,
      "learning_rate": 3.163131496748961e-09,
      "loss": 0.0106,
      "step": 11073
    },
    {
      "epoch": 2.9696969696969697,
      "grad_norm": 0.3304426194496598,
      "learning_rate": 3.1078870788403547e-09,
      "loss": 0.0177,
      "step": 11074
    },
    {
      "epoch": 2.969965138106731,
      "grad_norm": 0.23580989752216705,
      "learning_rate": 3.053129195050852e-09,
      "loss": 0.0161,
      "step": 11075
    },
    {
      "epoch": 2.9702333065164925,
      "grad_norm": 0.21336601368907335,
      "learning_rate": 2.99885785071119e-09,
      "loss": 0.0077,
      "step": 11076
    },
    {
      "epoch": 2.9705014749262535,
      "grad_norm": 0.3206062621438976,
      "learning_rate": 2.9450730511060288e-09,
      "loss": 0.0171,
      "step": 11077
    },
    {
      "epoch": 2.970769643336015,
      "grad_norm": 0.2933543724305954,
      "learning_rate": 2.891774801471736e-09,
      "loss": 0.0136,
      "step": 11078
    },
    {
      "epoch": 2.9710378117457763,
      "grad_norm": 0.2986064974373111,
      "learning_rate": 2.8389631069986044e-09,
      "loss": 0.0174,
      "step": 11079
    },
    {
      "epoch": 2.9713059801555377,
      "grad_norm": 0.2198478181512586,
      "learning_rate": 2.786637972828632e-09,
      "loss": 0.012,
      "step": 11080
    },
    {
      "epoch": 2.971574148565299,
      "grad_norm": 0.23523966564909787,
      "learning_rate": 2.7347994040560767e-09,
      "loss": 0.0127,
      "step": 11081
    },
    {
      "epoch": 2.97184231697506,
      "grad_norm": 0.24498553157980932,
      "learning_rate": 2.6834474057285676e-09,
      "loss": 0.0187,
      "step": 11082
    },
    {
      "epoch": 2.972110485384822,
      "grad_norm": 0.2637130222516912,
      "learning_rate": 2.6325819828465495e-09,
      "loss": 0.0152,
      "step": 11083
    },
    {
      "epoch": 2.972378653794583,
      "grad_norm": 0.2110131468880044,
      "learning_rate": 2.5822031403621715e-09,
      "loss": 0.0109,
      "step": 11084
    },
    {
      "epoch": 2.9726468222043443,
      "grad_norm": 0.2585096548134443,
      "learning_rate": 2.5323108831804e-09,
      "loss": 0.0097,
      "step": 11085
    },
    {
      "epoch": 2.9729149906141057,
      "grad_norm": 0.30902473013626225,
      "learning_rate": 2.4829052161601253e-09,
      "loss": 0.014,
      "step": 11086
    },
    {
      "epoch": 2.973183159023867,
      "grad_norm": 0.3101559214986143,
      "learning_rate": 2.4339861441108337e-09,
      "loss": 0.018,
      "step": 11087
    },
    {
      "epoch": 2.9734513274336285,
      "grad_norm": 0.2098092126192229,
      "learning_rate": 2.3855536717959372e-09,
      "loss": 0.0116,
      "step": 11088
    },
    {
      "epoch": 2.9737194958433895,
      "grad_norm": 0.20397358814268166,
      "learning_rate": 2.337607803931108e-09,
      "loss": 0.009,
      "step": 11089
    },
    {
      "epoch": 2.973987664253151,
      "grad_norm": 0.23881556088138067,
      "learning_rate": 2.290148545185389e-09,
      "loss": 0.0119,
      "step": 11090
    },
    {
      "epoch": 2.9742558326629123,
      "grad_norm": 0.18642913363217348,
      "learning_rate": 2.2431759001789734e-09,
      "loss": 0.0085,
      "step": 11091
    },
    {
      "epoch": 2.9745240010726737,
      "grad_norm": 0.24466709529192943,
      "learning_rate": 2.1966898734854246e-09,
      "loss": 0.0113,
      "step": 11092
    },
    {
      "epoch": 2.974792169482435,
      "grad_norm": 0.23209169646385097,
      "learning_rate": 2.1506904696316775e-09,
      "loss": 0.0173,
      "step": 11093
    },
    {
      "epoch": 2.975060337892196,
      "grad_norm": 0.24040530436982996,
      "learning_rate": 2.1051776930963718e-09,
      "loss": 0.0109,
      "step": 11094
    },
    {
      "epoch": 2.975328506301958,
      "grad_norm": 0.2215895372181935,
      "learning_rate": 2.0601515483104073e-09,
      "loss": 0.0125,
      "step": 11095
    },
    {
      "epoch": 2.975596674711719,
      "grad_norm": 0.219227633530607,
      "learning_rate": 2.015612039658055e-09,
      "loss": 0.0106,
      "step": 11096
    },
    {
      "epoch": 2.9758648431214803,
      "grad_norm": 0.3057921284664526,
      "learning_rate": 1.9715591714764003e-09,
      "loss": 0.0139,
      "step": 11097
    },
    {
      "epoch": 2.9761330115312417,
      "grad_norm": 0.2557013325661048,
      "learning_rate": 1.927992948054791e-09,
      "loss": 0.0126,
      "step": 11098
    },
    {
      "epoch": 2.976401179941003,
      "grad_norm": 0.3541306906266848,
      "learning_rate": 1.8849133736342786e-09,
      "loss": 0.0136,
      "step": 11099
    },
    {
      "epoch": 2.9766693483507645,
      "grad_norm": 0.22237571207545903,
      "learning_rate": 1.8423204524103955e-09,
      "loss": 0.0157,
      "step": 11100
    },
    {
      "epoch": 2.9769375167605254,
      "grad_norm": 0.2434901625028527,
      "learning_rate": 1.8002141885298252e-09,
      "loss": 0.0103,
      "step": 11101
    },
    {
      "epoch": 2.977205685170287,
      "grad_norm": 0.21593467113912282,
      "learning_rate": 1.7585945860926213e-09,
      "loss": 0.0178,
      "step": 11102
    },
    {
      "epoch": 2.9774738535800482,
      "grad_norm": 0.26323112928104564,
      "learning_rate": 1.7174616491510975e-09,
      "loss": 0.016,
      "step": 11103
    },
    {
      "epoch": 2.9777420219898096,
      "grad_norm": 0.26082567390149924,
      "learning_rate": 1.6768153817098288e-09,
      "loss": 0.0095,
      "step": 11104
    },
    {
      "epoch": 2.978010190399571,
      "grad_norm": 0.22808575341026444,
      "learning_rate": 1.6366557877267597e-09,
      "loss": 0.0104,
      "step": 11105
    },
    {
      "epoch": 2.978278358809332,
      "grad_norm": 0.2130544878242677,
      "learning_rate": 1.5969828711120961e-09,
      "loss": 0.0086,
      "step": 11106
    },
    {
      "epoch": 2.978546527219094,
      "grad_norm": 0.2432737700232307,
      "learning_rate": 1.5577966357288588e-09,
      "loss": 0.0128,
      "step": 11107
    },
    {
      "epoch": 2.978814695628855,
      "grad_norm": 0.30025075504350274,
      "learning_rate": 1.5190970853923293e-09,
      "loss": 0.0137,
      "step": 11108
    },
    {
      "epoch": 2.979082864038616,
      "grad_norm": 0.2591173124550683,
      "learning_rate": 1.4808842238706045e-09,
      "loss": 0.015,
      "step": 11109
    },
    {
      "epoch": 2.9793510324483776,
      "grad_norm": 0.2618365627730751,
      "learning_rate": 1.4431580548840418e-09,
      "loss": 0.017,
      "step": 11110
    },
    {
      "epoch": 2.979619200858139,
      "grad_norm": 0.26814887701328494,
      "learning_rate": 1.4059185821069243e-09,
      "loss": 0.0115,
      "step": 11111
    },
    {
      "epoch": 2.9798873692679004,
      "grad_norm": 0.25802258121965493,
      "learning_rate": 1.3691658091641302e-09,
      "loss": 0.011,
      "step": 11112
    },
    {
      "epoch": 2.9801555376776614,
      "grad_norm": 0.2920627452502052,
      "learning_rate": 1.3328997396344634e-09,
      "loss": 0.0142,
      "step": 11113
    },
    {
      "epoch": 2.980423706087423,
      "grad_norm": 0.21314065808199437,
      "learning_rate": 1.2971203770489883e-09,
      "loss": 0.0105,
      "step": 11114
    },
    {
      "epoch": 2.980691874497184,
      "grad_norm": 0.2178413280274995,
      "learning_rate": 1.2618277248921397e-09,
      "loss": 0.0152,
      "step": 11115
    },
    {
      "epoch": 2.9809600429069456,
      "grad_norm": 0.21505661295541822,
      "learning_rate": 1.2270217865995026e-09,
      "loss": 0.0164,
      "step": 11116
    },
    {
      "epoch": 2.981228211316707,
      "grad_norm": 0.23383493051656698,
      "learning_rate": 1.1927025655600333e-09,
      "loss": 0.0093,
      "step": 11117
    },
    {
      "epoch": 2.981496379726468,
      "grad_norm": 0.3195386338478377,
      "learning_rate": 1.1588700651155027e-09,
      "loss": 0.0144,
      "step": 11118
    },
    {
      "epoch": 2.98176454813623,
      "grad_norm": 0.23018117007013592,
      "learning_rate": 1.1255242885604978e-09,
      "loss": 0.0137,
      "step": 11119
    },
    {
      "epoch": 2.982032716545991,
      "grad_norm": 0.26294794627346524,
      "learning_rate": 1.0926652391413106e-09,
      "loss": 0.0097,
      "step": 11120
    },
    {
      "epoch": 2.982300884955752,
      "grad_norm": 0.2206135189688444,
      "learning_rate": 1.060292920057604e-09,
      "loss": 0.013,
      "step": 11121
    },
    {
      "epoch": 2.9825690533655136,
      "grad_norm": 0.21675469562138897,
      "learning_rate": 1.0284073344607458e-09,
      "loss": 0.0088,
      "step": 11122
    },
    {
      "epoch": 2.982837221775275,
      "grad_norm": 0.18564725714428187,
      "learning_rate": 9.970084854565853e-10,
      "loss": 0.0098,
      "step": 11123
    },
    {
      "epoch": 2.9831053901850364,
      "grad_norm": 0.5889375437527694,
      "learning_rate": 9.660963761010112e-10,
      "loss": 0.0193,
      "step": 11124
    },
    {
      "epoch": 2.9833735585947974,
      "grad_norm": 0.2978600648543535,
      "learning_rate": 9.35671009404948e-10,
      "loss": 0.0129,
      "step": 11125
    },
    {
      "epoch": 2.9836417270045588,
      "grad_norm": 0.19947924721086488,
      "learning_rate": 9.057323883299163e-10,
      "loss": 0.0114,
      "step": 11126
    },
    {
      "epoch": 2.98390989541432,
      "grad_norm": 0.23424007650263012,
      "learning_rate": 8.762805157913612e-10,
      "loss": 0.0131,
      "step": 11127
    },
    {
      "epoch": 2.9841780638240816,
      "grad_norm": 0.19782068621730642,
      "learning_rate": 8.47315394656989e-10,
      "loss": 0.0096,
      "step": 11128
    },
    {
      "epoch": 2.984446232233843,
      "grad_norm": 0.23739222908975008,
      "learning_rate": 8.188370277473212e-10,
      "loss": 0.0149,
      "step": 11129
    },
    {
      "epoch": 2.984714400643604,
      "grad_norm": 0.2001292391251265,
      "learning_rate": 7.908454178345848e-10,
      "loss": 0.0094,
      "step": 11130
    },
    {
      "epoch": 2.984982569053366,
      "grad_norm": 0.23650733090340254,
      "learning_rate": 7.633405676449323e-10,
      "loss": 0.0142,
      "step": 11131
    },
    {
      "epoch": 2.9852507374631267,
      "grad_norm": 0.19164698734944594,
      "learning_rate": 7.363224798556668e-10,
      "loss": 0.0085,
      "step": 11132
    },
    {
      "epoch": 2.985518905872888,
      "grad_norm": 0.23934634512458913,
      "learning_rate": 7.097911570985717e-10,
      "loss": 0.0177,
      "step": 11133
    },
    {
      "epoch": 2.9857870742826496,
      "grad_norm": 0.24602476640544266,
      "learning_rate": 6.837466019554706e-10,
      "loss": 0.0131,
      "step": 11134
    },
    {
      "epoch": 2.986055242692411,
      "grad_norm": 0.2836286349639682,
      "learning_rate": 6.581888169637784e-10,
      "loss": 0.0196,
      "step": 11135
    },
    {
      "epoch": 2.9863234111021724,
      "grad_norm": 0.2358314743971647,
      "learning_rate": 6.331178046103948e-10,
      "loss": 0.0124,
      "step": 11136
    },
    {
      "epoch": 2.9865915795119333,
      "grad_norm": 0.25059738817762284,
      "learning_rate": 6.085335673378101e-10,
      "loss": 0.0113,
      "step": 11137
    },
    {
      "epoch": 2.9868597479216947,
      "grad_norm": 0.2377480780758983,
      "learning_rate": 5.844361075391103e-10,
      "loss": 0.0149,
      "step": 11138
    },
    {
      "epoch": 2.987127916331456,
      "grad_norm": 0.2823192037429361,
      "learning_rate": 5.608254275607516e-10,
      "loss": 0.0141,
      "step": 11139
    },
    {
      "epoch": 2.9873960847412175,
      "grad_norm": 0.2354067392717898,
      "learning_rate": 5.37701529701451e-10,
      "loss": 0.0163,
      "step": 11140
    },
    {
      "epoch": 2.987664253150979,
      "grad_norm": 0.27142116577533837,
      "learning_rate": 5.150644162127405e-10,
      "loss": 0.0145,
      "step": 11141
    },
    {
      "epoch": 2.98793242156074,
      "grad_norm": 0.23774416641152707,
      "learning_rate": 4.929140892989681e-10,
      "loss": 0.0176,
      "step": 11142
    },
    {
      "epoch": 2.9882005899705013,
      "grad_norm": 0.22655035337014107,
      "learning_rate": 4.71250551116742e-10,
      "loss": 0.0133,
      "step": 11143
    },
    {
      "epoch": 2.9884687583802627,
      "grad_norm": 0.24267125735223366,
      "learning_rate": 4.5007380377548594e-10,
      "loss": 0.0169,
      "step": 11144
    },
    {
      "epoch": 2.988736926790024,
      "grad_norm": 0.3861421649057795,
      "learning_rate": 4.2938384933688405e-10,
      "loss": 0.0123,
      "step": 11145
    },
    {
      "epoch": 2.9890050951997855,
      "grad_norm": 0.27172508949621915,
      "learning_rate": 4.09180689815436e-10,
      "loss": 0.0151,
      "step": 11146
    },
    {
      "epoch": 2.989273263609547,
      "grad_norm": 0.20851391388169851,
      "learning_rate": 3.8946432717845707e-10,
      "loss": 0.0097,
      "step": 11147
    },
    {
      "epoch": 2.9895414320193083,
      "grad_norm": 0.1904651841812917,
      "learning_rate": 3.7023476334552276e-10,
      "loss": 0.0116,
      "step": 11148
    },
    {
      "epoch": 2.9898096004290693,
      "grad_norm": 0.31894540289041023,
      "learning_rate": 3.514920001895794e-10,
      "loss": 0.0191,
      "step": 11149
    },
    {
      "epoch": 2.9900777688388307,
      "grad_norm": 0.24232342545378996,
      "learning_rate": 3.332360395341683e-10,
      "loss": 0.0163,
      "step": 11150
    },
    {
      "epoch": 2.990345937248592,
      "grad_norm": 0.27627238640933655,
      "learning_rate": 3.1546688315842177e-10,
      "loss": 0.0171,
      "step": 11151
    },
    {
      "epoch": 2.9906141056583535,
      "grad_norm": 0.2852730314315949,
      "learning_rate": 2.981845327915123e-10,
      "loss": 0.0172,
      "step": 11152
    },
    {
      "epoch": 2.990882274068115,
      "grad_norm": 0.19117218788983623,
      "learning_rate": 2.8138899011653785e-10,
      "loss": 0.0094,
      "step": 11153
    },
    {
      "epoch": 2.991150442477876,
      "grad_norm": 0.31070996350426516,
      "learning_rate": 2.6508025676885706e-10,
      "loss": 0.0178,
      "step": 11154
    },
    {
      "epoch": 2.9914186108876373,
      "grad_norm": 0.267172109353608,
      "learning_rate": 2.492583343360888e-10,
      "loss": 0.0125,
      "step": 11155
    },
    {
      "epoch": 2.9916867792973987,
      "grad_norm": 0.2784276436201992,
      "learning_rate": 2.3392322435866753e-10,
      "loss": 0.018,
      "step": 11156
    },
    {
      "epoch": 2.99195494770716,
      "grad_norm": 0.2577698577895211,
      "learning_rate": 2.1907492833039835e-10,
      "loss": 0.0147,
      "step": 11157
    },
    {
      "epoch": 2.9922231161169215,
      "grad_norm": 0.20364807102104018,
      "learning_rate": 2.0471344769623646e-10,
      "loss": 0.0141,
      "step": 11158
    },
    {
      "epoch": 2.992491284526683,
      "grad_norm": 0.20838888270026573,
      "learning_rate": 1.9083878385561806e-10,
      "loss": 0.0137,
      "step": 11159
    },
    {
      "epoch": 2.9927594529364443,
      "grad_norm": 0.23332824344910325,
      "learning_rate": 1.7745093815801916e-10,
      "loss": 0.0124,
      "step": 11160
    },
    {
      "epoch": 2.9930276213462053,
      "grad_norm": 0.2757713952020089,
      "learning_rate": 1.6454991190795188e-10,
      "loss": 0.0117,
      "step": 11161
    },
    {
      "epoch": 2.9932957897559667,
      "grad_norm": 0.2587955182055354,
      "learning_rate": 1.5213570636163356e-10,
      "loss": 0.0162,
      "step": 11162
    },
    {
      "epoch": 2.993563958165728,
      "grad_norm": 0.23051791135146446,
      "learning_rate": 1.4020832272754193e-10,
      "loss": 0.0103,
      "step": 11163
    },
    {
      "epoch": 2.9938321265754895,
      "grad_norm": 0.183293732768527,
      "learning_rate": 1.2876776216697028e-10,
      "loss": 0.0092,
      "step": 11164
    },
    {
      "epoch": 2.994100294985251,
      "grad_norm": 0.21153636109007778,
      "learning_rate": 1.1781402579347235e-10,
      "loss": 0.012,
      "step": 11165
    },
    {
      "epoch": 2.994368463395012,
      "grad_norm": 0.29251776124803863,
      "learning_rate": 1.0734711467452752e-10,
      "loss": 0.0138,
      "step": 11166
    },
    {
      "epoch": 2.9946366318047732,
      "grad_norm": 0.25994329029003777,
      "learning_rate": 9.736702982821033e-11,
      "loss": 0.0122,
      "step": 11167
    },
    {
      "epoch": 2.9949048002145346,
      "grad_norm": 0.27612524144044037,
      "learning_rate": 8.787377222707616e-11,
      "loss": 0.0172,
      "step": 11168
    },
    {
      "epoch": 2.995172968624296,
      "grad_norm": 0.3315308896947658,
      "learning_rate": 7.886734279538566e-11,
      "loss": 0.0312,
      "step": 11169
    },
    {
      "epoch": 2.9954411370340575,
      "grad_norm": 0.2510527447873307,
      "learning_rate": 7.034774240965992e-11,
      "loss": 0.0128,
      "step": 11170
    },
    {
      "epoch": 2.995709305443819,
      "grad_norm": 0.2284363134102265,
      "learning_rate": 6.231497189979063e-11,
      "loss": 0.0103,
      "step": 11171
    },
    {
      "epoch": 2.9959774738535803,
      "grad_norm": 0.3081636017108401,
      "learning_rate": 5.4769032047374823e-11,
      "loss": 0.0122,
      "step": 11172
    },
    {
      "epoch": 2.9962456422633412,
      "grad_norm": 0.25084693586701395,
      "learning_rate": 4.7709923587380136e-11,
      "loss": 0.0148,
      "step": 11173
    },
    {
      "epoch": 2.9965138106731026,
      "grad_norm": 0.21931056938745797,
      "learning_rate": 4.113764720758972e-11,
      "loss": 0.0115,
      "step": 11174
    },
    {
      "epoch": 2.996781979082864,
      "grad_norm": 0.24404822227757195,
      "learning_rate": 3.505220354749206e-11,
      "loss": 0.0112,
      "step": 11175
    },
    {
      "epoch": 2.9970501474926254,
      "grad_norm": 0.17974322241819385,
      "learning_rate": 2.9453593199946226e-11,
      "loss": 0.0108,
      "step": 11176
    },
    {
      "epoch": 2.997318315902387,
      "grad_norm": 0.34380747076878193,
      "learning_rate": 2.434181670951663e-11,
      "loss": 0.0164,
      "step": 11177
    },
    {
      "epoch": 2.997586484312148,
      "grad_norm": 0.23731527116957574,
      "learning_rate": 1.9716874574138288e-11,
      "loss": 0.0092,
      "step": 11178
    },
    {
      "epoch": 2.997854652721909,
      "grad_norm": 0.25534359720347416,
      "learning_rate": 1.5578767244561753e-11,
      "loss": 0.0119,
      "step": 11179
    },
    {
      "epoch": 2.9981228211316706,
      "grad_norm": 0.21976533655679964,
      "learning_rate": 1.1927495123242872e-11,
      "loss": 0.0121,
      "step": 11180
    },
    {
      "epoch": 2.998390989541432,
      "grad_norm": 0.21196187585790638,
      "learning_rate": 8.76305856600812e-12,
      "loss": 0.0146,
      "step": 11181
    },
    {
      "epoch": 2.9986591579511934,
      "grad_norm": 0.1956700201041784,
      "learning_rate": 6.085457880389278e-12,
      "loss": 0.0105,
      "step": 11182
    },
    {
      "epoch": 2.998927326360955,
      "grad_norm": 0.29091444988670095,
      "learning_rate": 3.8946933278438685e-12,
      "loss": 0.0146,
      "step": 11183
    },
    {
      "epoch": 2.9991954947707162,
      "grad_norm": 0.21513102373085805,
      "learning_rate": 2.19076512153471e-12,
      "loss": 0.018,
      "step": 11184
    },
    {
      "epoch": 2.999463663180477,
      "grad_norm": 0.6836501830867728,
      "learning_rate": 9.736734268850356e-13,
      "loss": 0.0203,
      "step": 11185
    },
    {
      "epoch": 2.9997318315902386,
      "grad_norm": 0.2272897283105736,
      "learning_rate": 2.4341836268870765e-13,
      "loss": 0.0096,
      "step": 11186
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.2688336217588715,
      "learning_rate": 0.0,
      "loss": 0.0116,
      "step": 11187
    },
    {
      "epoch": 3.0,
      "eval_loss": 0.02185099571943283,
      "eval_runtime": 292.629,
      "eval_samples_per_second": 85.846,
      "eval_steps_per_second": 1.343,
      "step": 11187
    },
    {
      "epoch": 3.0,
      "step": 11187,
      "total_flos": 4.3304182632184545e+18,
      "train_loss": 0.030125039980372262,
      "train_runtime": 67465.0545,
      "train_samples_per_second": 21.222,
      "train_steps_per_second": 0.166
    }
  ],
  "logging_steps": 1,
  "max_steps": 11187,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 4.3304182632184545e+18,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}