{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "eval_steps": 200,
  "global_step": 1334,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0014992503748125937,
      "grad_norm": 1.2691674913964377,
      "learning_rate": 9.999986134743949e-06,
      "loss": 0.1815,
      "step": 1
    },
    {
      "epoch": 0.0029985007496251873,
      "grad_norm": 1.0228596277688027,
      "learning_rate": 9.99994453905269e-06,
      "loss": 0.1522,
      "step": 2
    },
    {
      "epoch": 0.004497751124437781,
      "grad_norm": 1.3372662388968108,
      "learning_rate": 9.999875213156919e-06,
      "loss": 0.1512,
      "step": 3
    },
    {
      "epoch": 0.005997001499250375,
      "grad_norm": 1.170935118226643,
      "learning_rate": 9.999778157441126e-06,
      "loss": 0.1234,
      "step": 4
    },
    {
      "epoch": 0.0074962518740629685,
      "grad_norm": 1.1571631597945553,
      "learning_rate": 9.99965337244359e-06,
      "loss": 0.1407,
      "step": 5
    },
    {
      "epoch": 0.008995502248875561,
      "grad_norm": 1.0302614494871871,
      "learning_rate": 9.999500858856382e-06,
      "loss": 0.1337,
      "step": 6
    },
    {
      "epoch": 0.010494752623688156,
      "grad_norm": 1.1254705772786355,
      "learning_rate": 9.999320617525356e-06,
      "loss": 0.1433,
      "step": 7
    },
    {
      "epoch": 0.01199400299850075,
      "grad_norm": 1.1473146169985184,
      "learning_rate": 9.999112649450154e-06,
      "loss": 0.1229,
      "step": 8
    },
    {
      "epoch": 0.013493253373313344,
      "grad_norm": 0.7085689723862112,
      "learning_rate": 9.998876955784183e-06,
      "loss": 0.0837,
      "step": 9
    },
    {
      "epoch": 0.014992503748125937,
      "grad_norm": 1.0944986908315535,
      "learning_rate": 9.998613537834625e-06,
      "loss": 0.1307,
      "step": 10
    },
    {
      "epoch": 0.01649175412293853,
      "grad_norm": 0.9520319519473738,
      "learning_rate": 9.998322397062426e-06,
      "loss": 0.1173,
      "step": 11
    },
    {
      "epoch": 0.017991004497751123,
      "grad_norm": 0.7702486899625681,
      "learning_rate": 9.99800353508228e-06,
      "loss": 0.0904,
      "step": 12
    },
    {
      "epoch": 0.019490254872563718,
      "grad_norm": 0.7475780162538579,
      "learning_rate": 9.997656953662627e-06,
      "loss": 0.0886,
      "step": 13
    },
    {
      "epoch": 0.020989505247376312,
      "grad_norm": 0.9731395975111904,
      "learning_rate": 9.997282654725645e-06,
      "loss": 0.1157,
      "step": 14
    },
    {
      "epoch": 0.022488755622188907,
      "grad_norm": 1.0016461831408086,
      "learning_rate": 9.996880640347234e-06,
      "loss": 0.0995,
      "step": 15
    },
    {
      "epoch": 0.0239880059970015,
      "grad_norm": 0.7410427008533226,
      "learning_rate": 9.99645091275701e-06,
      "loss": 0.0875,
      "step": 16
    },
    {
      "epoch": 0.025487256371814093,
      "grad_norm": 0.8929664450622328,
      "learning_rate": 9.99599347433828e-06,
      "loss": 0.1071,
      "step": 17
    },
    {
      "epoch": 0.026986506746626688,
      "grad_norm": 0.9775601185819104,
      "learning_rate": 9.99550832762805e-06,
      "loss": 0.1226,
      "step": 18
    },
    {
      "epoch": 0.02848575712143928,
      "grad_norm": 0.7616756047957358,
      "learning_rate": 9.99499547531699e-06,
      "loss": 0.1018,
      "step": 19
    },
    {
      "epoch": 0.029985007496251874,
      "grad_norm": 0.96154648237228,
      "learning_rate": 9.994454920249433e-06,
      "loss": 0.1138,
      "step": 20
    },
    {
      "epoch": 0.031484257871064465,
      "grad_norm": 0.7569937028955507,
      "learning_rate": 9.993886665423348e-06,
      "loss": 0.0905,
      "step": 21
    },
    {
      "epoch": 0.03298350824587706,
      "grad_norm": 0.799275423504696,
      "learning_rate": 9.993290713990343e-06,
      "loss": 0.1122,
      "step": 22
    },
    {
      "epoch": 0.034482758620689655,
      "grad_norm": 0.6384906148017058,
      "learning_rate": 9.99266706925562e-06,
      "loss": 0.0801,
      "step": 23
    },
    {
      "epoch": 0.035982008995502246,
      "grad_norm": 1.0084574284450645,
      "learning_rate": 9.992015734677979e-06,
      "loss": 0.1117,
      "step": 24
    },
    {
      "epoch": 0.037481259370314844,
      "grad_norm": 0.7724596644191922,
      "learning_rate": 9.991336713869785e-06,
      "loss": 0.1003,
      "step": 25
    },
    {
      "epoch": 0.038980509745127435,
      "grad_norm": 0.6541067868761109,
      "learning_rate": 9.99063001059696e-06,
      "loss": 0.09,
      "step": 26
    },
    {
      "epoch": 0.04047976011994003,
      "grad_norm": 1.0001625092622735,
      "learning_rate": 9.989895628778952e-06,
      "loss": 0.1248,
      "step": 27
    },
    {
      "epoch": 0.041979010494752625,
      "grad_norm": 0.7071244340991423,
      "learning_rate": 9.989133572488716e-06,
      "loss": 0.0778,
      "step": 28
    },
    {
      "epoch": 0.043478260869565216,
      "grad_norm": 0.8120594542537397,
      "learning_rate": 9.988343845952697e-06,
      "loss": 0.0777,
      "step": 29
    },
    {
      "epoch": 0.044977511244377814,
      "grad_norm": 0.8019604974306029,
      "learning_rate": 9.987526453550798e-06,
      "loss": 0.0881,
      "step": 30
    },
    {
      "epoch": 0.046476761619190406,
      "grad_norm": 0.6600532316036861,
      "learning_rate": 9.98668139981636e-06,
      "loss": 0.0775,
      "step": 31
    },
    {
      "epoch": 0.047976011994003,
      "grad_norm": 0.9350317968808954,
      "learning_rate": 9.98580868943614e-06,
      "loss": 0.1133,
      "step": 32
    },
    {
      "epoch": 0.049475262368815595,
      "grad_norm": 0.6024407464964036,
      "learning_rate": 9.984908327250278e-06,
      "loss": 0.069,
      "step": 33
    },
    {
      "epoch": 0.050974512743628186,
      "grad_norm": 0.6709253889525743,
      "learning_rate": 9.983980318252274e-06,
      "loss": 0.0623,
      "step": 34
    },
    {
      "epoch": 0.05247376311844078,
      "grad_norm": 0.6763013765907574,
      "learning_rate": 9.983024667588961e-06,
      "loss": 0.0888,
      "step": 35
    },
    {
      "epoch": 0.053973013493253376,
      "grad_norm": 0.9091221703629316,
      "learning_rate": 9.982041380560476e-06,
      "loss": 0.1081,
      "step": 36
    },
    {
      "epoch": 0.05547226386806597,
      "grad_norm": 0.8779043271136543,
      "learning_rate": 9.98103046262023e-06,
      "loss": 0.0939,
      "step": 37
    },
    {
      "epoch": 0.05697151424287856,
      "grad_norm": 1.0845619867641336,
      "learning_rate": 9.979991919374877e-06,
      "loss": 0.1216,
      "step": 38
    },
    {
      "epoch": 0.05847076461769116,
      "grad_norm": 0.8325754830032166,
      "learning_rate": 9.978925756584284e-06,
      "loss": 0.1078,
      "step": 39
    },
    {
      "epoch": 0.05997001499250375,
      "grad_norm": 0.6984280019699092,
      "learning_rate": 9.9778319801615e-06,
      "loss": 0.076,
      "step": 40
    },
    {
      "epoch": 0.06146926536731634,
      "grad_norm": 0.834759342491417,
      "learning_rate": 9.976710596172721e-06,
      "loss": 0.0945,
      "step": 41
    },
    {
      "epoch": 0.06296851574212893,
      "grad_norm": 0.8064825610227835,
      "learning_rate": 9.975561610837254e-06,
      "loss": 0.0956,
      "step": 42
    },
    {
      "epoch": 0.06446776611694154,
      "grad_norm": 0.6789080053064146,
      "learning_rate": 9.974385030527496e-06,
      "loss": 0.0765,
      "step": 43
    },
    {
      "epoch": 0.06596701649175413,
      "grad_norm": 0.8137652590157077,
      "learning_rate": 9.973180861768874e-06,
      "loss": 0.104,
      "step": 44
    },
    {
      "epoch": 0.06746626686656672,
      "grad_norm": 0.8867241541600891,
      "learning_rate": 9.971949111239838e-06,
      "loss": 0.105,
      "step": 45
    },
    {
      "epoch": 0.06896551724137931,
      "grad_norm": 0.5837616675643288,
      "learning_rate": 9.970689785771798e-06,
      "loss": 0.0676,
      "step": 46
    },
    {
      "epoch": 0.0704647676161919,
      "grad_norm": 0.6064516068058603,
      "learning_rate": 9.969402892349105e-06,
      "loss": 0.0663,
      "step": 47
    },
    {
      "epoch": 0.07196401799100449,
      "grad_norm": 0.8556259664919593,
      "learning_rate": 9.968088438109002e-06,
      "loss": 0.0744,
      "step": 48
    },
    {
      "epoch": 0.0734632683658171,
      "grad_norm": 0.7112221278855972,
      "learning_rate": 9.966746430341584e-06,
      "loss": 0.0698,
      "step": 49
    },
    {
      "epoch": 0.07496251874062969,
      "grad_norm": 1.1235562727393593,
      "learning_rate": 9.965376876489765e-06,
      "loss": 0.1176,
      "step": 50
    },
    {
      "epoch": 0.07646176911544228,
      "grad_norm": 0.7802202941284992,
      "learning_rate": 9.963979784149232e-06,
      "loss": 0.0951,
      "step": 51
    },
    {
      "epoch": 0.07796101949025487,
      "grad_norm": 1.024772195335334,
      "learning_rate": 9.962555161068401e-06,
      "loss": 0.0981,
      "step": 52
    },
    {
      "epoch": 0.07946026986506746,
      "grad_norm": 0.7808429764844336,
      "learning_rate": 9.961103015148376e-06,
      "loss": 0.0983,
      "step": 53
    },
    {
      "epoch": 0.08095952023988005,
      "grad_norm": 0.7891744443367578,
      "learning_rate": 9.95962335444291e-06,
      "loss": 0.0864,
      "step": 54
    },
    {
      "epoch": 0.08245877061469266,
      "grad_norm": 0.7254714472928101,
      "learning_rate": 9.958116187158351e-06,
      "loss": 0.0706,
      "step": 55
    },
    {
      "epoch": 0.08395802098950525,
      "grad_norm": 0.9676335312388703,
      "learning_rate": 9.956581521653604e-06,
      "loss": 0.1272,
      "step": 56
    },
    {
      "epoch": 0.08545727136431784,
      "grad_norm": 0.9110487109671824,
      "learning_rate": 9.955019366440082e-06,
      "loss": 0.1006,
      "step": 57
    },
    {
      "epoch": 0.08695652173913043,
      "grad_norm": 0.8643823527549352,
      "learning_rate": 9.953429730181653e-06,
      "loss": 0.0939,
      "step": 58
    },
    {
      "epoch": 0.08845577211394302,
      "grad_norm": 0.6109415076814978,
      "learning_rate": 9.95181262169461e-06,
      "loss": 0.0781,
      "step": 59
    },
    {
      "epoch": 0.08995502248875563,
      "grad_norm": 0.7921944994614851,
      "learning_rate": 9.950168049947597e-06,
      "loss": 0.0926,
      "step": 60
    },
    {
      "epoch": 0.09145427286356822,
      "grad_norm": 0.6100738113943155,
      "learning_rate": 9.948496024061577e-06,
      "loss": 0.0655,
      "step": 61
    },
    {
      "epoch": 0.09295352323838081,
      "grad_norm": 0.7370040640485737,
      "learning_rate": 9.94679655330978e-06,
      "loss": 0.1005,
      "step": 62
    },
    {
      "epoch": 0.0944527736131934,
      "grad_norm": 0.9869851324958635,
      "learning_rate": 9.945069647117645e-06,
      "loss": 0.0877,
      "step": 63
    },
    {
      "epoch": 0.095952023988006,
      "grad_norm": 0.7881190142822982,
      "learning_rate": 9.943315315062766e-06,
      "loss": 0.0844,
      "step": 64
    },
    {
      "epoch": 0.09745127436281859,
      "grad_norm": 0.5801361495993674,
      "learning_rate": 9.941533566874852e-06,
      "loss": 0.0568,
      "step": 65
    },
    {
      "epoch": 0.09895052473763119,
      "grad_norm": 0.9978953839299076,
      "learning_rate": 9.939724412435661e-06,
      "loss": 0.1097,
      "step": 66
    },
    {
      "epoch": 0.10044977511244378,
      "grad_norm": 0.6903924539096281,
      "learning_rate": 9.937887861778947e-06,
      "loss": 0.0849,
      "step": 67
    },
    {
      "epoch": 0.10194902548725637,
      "grad_norm": 1.2769739615391271,
      "learning_rate": 9.93602392509041e-06,
      "loss": 0.1104,
      "step": 68
    },
    {
      "epoch": 0.10344827586206896,
      "grad_norm": 0.7827829998547647,
      "learning_rate": 9.934132612707631e-06,
      "loss": 0.0813,
      "step": 69
    },
    {
      "epoch": 0.10494752623688156,
      "grad_norm": 0.7405994079043439,
      "learning_rate": 9.932213935120025e-06,
      "loss": 0.072,
      "step": 70
    },
    {
      "epoch": 0.10644677661169415,
      "grad_norm": 0.5289588902364146,
      "learning_rate": 9.930267902968774e-06,
      "loss": 0.0621,
      "step": 71
    },
    {
      "epoch": 0.10794602698650675,
      "grad_norm": 0.9458320628536533,
      "learning_rate": 9.928294527046771e-06,
      "loss": 0.1108,
      "step": 72
    },
    {
      "epoch": 0.10944527736131934,
      "grad_norm": 0.968391885456912,
      "learning_rate": 9.92629381829856e-06,
      "loss": 0.0983,
      "step": 73
    },
    {
      "epoch": 0.11094452773613193,
      "grad_norm": 1.1028306777426347,
      "learning_rate": 9.924265787820279e-06,
      "loss": 0.1213,
      "step": 74
    },
    {
      "epoch": 0.11244377811094453,
      "grad_norm": 0.7918948112001529,
      "learning_rate": 9.92221044685959e-06,
      "loss": 0.0937,
      "step": 75
    },
    {
      "epoch": 0.11394302848575712,
      "grad_norm": 0.7740242267247712,
      "learning_rate": 9.920127806815627e-06,
      "loss": 0.077,
      "step": 76
    },
    {
      "epoch": 0.11544227886056972,
      "grad_norm": 0.6916763160849486,
      "learning_rate": 9.918017879238922e-06,
      "loss": 0.0779,
      "step": 77
    },
    {
      "epoch": 0.11694152923538231,
      "grad_norm": 0.5481460858134668,
      "learning_rate": 9.915880675831352e-06,
      "loss": 0.053,
      "step": 78
    },
    {
      "epoch": 0.1184407796101949,
      "grad_norm": 0.8672567434970185,
      "learning_rate": 9.913716208446067e-06,
      "loss": 0.0861,
      "step": 79
    },
    {
      "epoch": 0.1199400299850075,
      "grad_norm": 0.7665095343786223,
      "learning_rate": 9.91152448908742e-06,
      "loss": 0.0833,
      "step": 80
    },
    {
      "epoch": 0.12143928035982009,
      "grad_norm": 0.9968401599957936,
      "learning_rate": 9.909305529910917e-06,
      "loss": 0.0982,
      "step": 81
    },
    {
      "epoch": 0.12293853073463268,
      "grad_norm": 0.7967827955029122,
      "learning_rate": 9.907059343223129e-06,
      "loss": 0.0925,
      "step": 82
    },
    {
      "epoch": 0.12443778110944528,
      "grad_norm": 0.5311051435732562,
      "learning_rate": 9.904785941481638e-06,
      "loss": 0.0589,
      "step": 83
    },
    {
      "epoch": 0.12593703148425786,
      "grad_norm": 0.48686895941370767,
      "learning_rate": 9.902485337294965e-06,
      "loss": 0.0485,
      "step": 84
    },
    {
      "epoch": 0.12743628185907047,
      "grad_norm": 0.7107985058223918,
      "learning_rate": 9.900157543422493e-06,
      "loss": 0.0682,
      "step": 85
    },
    {
      "epoch": 0.12893553223388307,
      "grad_norm": 0.7549048070489338,
      "learning_rate": 9.897802572774407e-06,
      "loss": 0.0745,
      "step": 86
    },
    {
      "epoch": 0.13043478260869565,
      "grad_norm": 0.6483457097131894,
      "learning_rate": 9.895420438411616e-06,
      "loss": 0.0689,
      "step": 87
    },
    {
      "epoch": 0.13193403298350825,
      "grad_norm": 0.5860833625339595,
      "learning_rate": 9.893011153545679e-06,
      "loss": 0.0546,
      "step": 88
    },
    {
      "epoch": 0.13343328335832083,
      "grad_norm": 0.8832044093408553,
      "learning_rate": 9.89057473153874e-06,
      "loss": 0.0971,
      "step": 89
    },
    {
      "epoch": 0.13493253373313344,
      "grad_norm": 0.7568872136007466,
      "learning_rate": 9.888111185903442e-06,
      "loss": 0.0732,
      "step": 90
    },
    {
      "epoch": 0.136431784107946,
      "grad_norm": 0.7524167051813937,
      "learning_rate": 9.885620530302865e-06,
      "loss": 0.0734,
      "step": 91
    },
    {
      "epoch": 0.13793103448275862,
      "grad_norm": 0.9018568068469728,
      "learning_rate": 9.883102778550434e-06,
      "loss": 0.0965,
      "step": 92
    },
    {
      "epoch": 0.13943028485757122,
      "grad_norm": 0.6468758976461324,
      "learning_rate": 9.880557944609863e-06,
      "loss": 0.0713,
      "step": 93
    },
    {
      "epoch": 0.1409295352323838,
      "grad_norm": 0.8321167093425592,
      "learning_rate": 9.877986042595062e-06,
      "loss": 0.0903,
      "step": 94
    },
    {
      "epoch": 0.1424287856071964,
      "grad_norm": 0.9299884338810234,
      "learning_rate": 9.87538708677006e-06,
      "loss": 0.1056,
      "step": 95
    },
    {
      "epoch": 0.14392803598200898,
      "grad_norm": 0.8020921983978067,
      "learning_rate": 9.872761091548933e-06,
      "loss": 0.0831,
      "step": 96
    },
    {
      "epoch": 0.1454272863568216,
      "grad_norm": 1.0068296391475788,
      "learning_rate": 9.870108071495721e-06,
      "loss": 0.1095,
      "step": 97
    },
    {
      "epoch": 0.1469265367316342,
      "grad_norm": 0.6841888464042496,
      "learning_rate": 9.867428041324345e-06,
      "loss": 0.0767,
      "step": 98
    },
    {
      "epoch": 0.14842578710644677,
      "grad_norm": 0.7154653961981009,
      "learning_rate": 9.864721015898524e-06,
      "loss": 0.0783,
      "step": 99
    },
    {
      "epoch": 0.14992503748125938,
      "grad_norm": 0.893570718265095,
      "learning_rate": 9.861987010231701e-06,
      "loss": 0.0901,
      "step": 100
    },
    {
      "epoch": 0.15142428785607195,
      "grad_norm": 0.8394161517582679,
      "learning_rate": 9.85922603948695e-06,
      "loss": 0.0964,
      "step": 101
    },
    {
      "epoch": 0.15292353823088456,
      "grad_norm": 0.7452175545125396,
      "learning_rate": 9.856438118976899e-06,
      "loss": 0.0821,
      "step": 102
    },
    {
      "epoch": 0.15442278860569716,
      "grad_norm": 0.6761999722383197,
      "learning_rate": 9.853623264163638e-06,
      "loss": 0.0654,
      "step": 103
    },
    {
      "epoch": 0.15592203898050974,
      "grad_norm": 0.6941202184351634,
      "learning_rate": 9.850781490658643e-06,
      "loss": 0.0749,
      "step": 104
    },
    {
      "epoch": 0.15742128935532235,
      "grad_norm": 0.7632986720353027,
      "learning_rate": 9.84791281422268e-06,
      "loss": 0.0955,
      "step": 105
    },
    {
      "epoch": 0.15892053973013492,
      "grad_norm": 0.8453487512001985,
      "learning_rate": 9.845017250765721e-06,
      "loss": 0.0958,
      "step": 106
    },
    {
      "epoch": 0.16041979010494753,
      "grad_norm": 0.6450115676114526,
      "learning_rate": 9.84209481634686e-06,
      "loss": 0.0715,
      "step": 107
    },
    {
      "epoch": 0.1619190404797601,
      "grad_norm": 1.0120719606469422,
      "learning_rate": 9.839145527174216e-06,
      "loss": 0.1092,
      "step": 108
    },
    {
      "epoch": 0.1634182908545727,
      "grad_norm": 0.7153285784953192,
      "learning_rate": 9.836169399604846e-06,
      "loss": 0.0901,
      "step": 109
    },
    {
      "epoch": 0.16491754122938532,
      "grad_norm": 0.7482718906236362,
      "learning_rate": 9.833166450144665e-06,
      "loss": 0.0804,
      "step": 110
    },
    {
      "epoch": 0.1664167916041979,
      "grad_norm": 0.7650984841314135,
      "learning_rate": 9.830136695448334e-06,
      "loss": 0.077,
      "step": 111
    },
    {
      "epoch": 0.1679160419790105,
      "grad_norm": 0.7772028035894263,
      "learning_rate": 9.827080152319182e-06,
      "loss": 0.085,
      "step": 112
    },
    {
      "epoch": 0.16941529235382308,
      "grad_norm": 0.9452512974534755,
      "learning_rate": 9.823996837709114e-06,
      "loss": 0.0755,
      "step": 113
    },
    {
      "epoch": 0.17091454272863568,
      "grad_norm": 1.0590668005193353,
      "learning_rate": 9.820886768718503e-06,
      "loss": 0.0981,
      "step": 114
    },
    {
      "epoch": 0.1724137931034483,
      "grad_norm": 0.6157560718459545,
      "learning_rate": 9.817749962596115e-06,
      "loss": 0.0721,
      "step": 115
    },
    {
      "epoch": 0.17391304347826086,
      "grad_norm": 1.1864738019955636,
      "learning_rate": 9.814586436738998e-06,
      "loss": 0.1824,
      "step": 116
    },
    {
      "epoch": 0.17541229385307347,
      "grad_norm": 0.7708987225969971,
      "learning_rate": 9.811396208692387e-06,
      "loss": 0.0843,
      "step": 117
    },
    {
      "epoch": 0.17691154422788605,
      "grad_norm": 0.9398056370647403,
      "learning_rate": 9.808179296149616e-06,
      "loss": 0.1074,
      "step": 118
    },
    {
      "epoch": 0.17841079460269865,
      "grad_norm": 0.9246627172537241,
      "learning_rate": 9.804935716952011e-06,
      "loss": 0.1004,
      "step": 119
    },
    {
      "epoch": 0.17991004497751126,
      "grad_norm": 0.6907734213095442,
      "learning_rate": 9.801665489088795e-06,
      "loss": 0.0721,
      "step": 120
    },
    {
      "epoch": 0.18140929535232383,
      "grad_norm": 0.9763401608523066,
      "learning_rate": 9.798368630696984e-06,
      "loss": 0.1042,
      "step": 121
    },
    {
      "epoch": 0.18290854572713644,
      "grad_norm": 0.8597543355517215,
      "learning_rate": 9.795045160061295e-06,
      "loss": 0.1023,
      "step": 122
    },
    {
      "epoch": 0.18440779610194902,
      "grad_norm": 0.8256871534861028,
      "learning_rate": 9.791695095614036e-06,
      "loss": 0.1035,
      "step": 123
    },
    {
      "epoch": 0.18590704647676162,
      "grad_norm": 0.6460761180261383,
      "learning_rate": 9.788318455935008e-06,
      "loss": 0.065,
      "step": 124
    },
    {
      "epoch": 0.1874062968515742,
      "grad_norm": 0.596741669575221,
      "learning_rate": 9.7849152597514e-06,
      "loss": 0.0597,
      "step": 125
    },
    {
      "epoch": 0.1889055472263868,
      "grad_norm": 0.6712790331862943,
      "learning_rate": 9.781485525937683e-06,
      "loss": 0.0658,
      "step": 126
    },
    {
      "epoch": 0.1904047976011994,
      "grad_norm": 0.6247456738876328,
      "learning_rate": 9.778029273515519e-06,
      "loss": 0.0732,
      "step": 127
    },
    {
      "epoch": 0.191904047976012,
      "grad_norm": 0.7816286449516086,
      "learning_rate": 9.774546521653633e-06,
      "loss": 0.0958,
      "step": 128
    },
    {
      "epoch": 0.1934032983508246,
      "grad_norm": 0.7626589087178943,
      "learning_rate": 9.771037289667726e-06,
      "loss": 0.0774,
      "step": 129
    },
    {
      "epoch": 0.19490254872563717,
      "grad_norm": 0.8096678150544202,
      "learning_rate": 9.767501597020357e-06,
      "loss": 0.0895,
      "step": 130
    },
    {
      "epoch": 0.19640179910044978,
      "grad_norm": 0.7899151630539276,
      "learning_rate": 9.76393946332084e-06,
      "loss": 0.0884,
      "step": 131
    },
    {
      "epoch": 0.19790104947526238,
      "grad_norm": 0.9611819328878667,
      "learning_rate": 9.760350908325131e-06,
      "loss": 0.0853,
      "step": 132
    },
    {
      "epoch": 0.19940029985007496,
      "grad_norm": 0.7256916832931272,
      "learning_rate": 9.756735951935725e-06,
      "loss": 0.094,
      "step": 133
    },
    {
      "epoch": 0.20089955022488756,
      "grad_norm": 0.6372103552076069,
      "learning_rate": 9.753094614201542e-06,
      "loss": 0.0622,
      "step": 134
    },
    {
      "epoch": 0.20239880059970014,
      "grad_norm": 0.7834639179979273,
      "learning_rate": 9.749426915317812e-06,
      "loss": 0.0896,
      "step": 135
    },
    {
      "epoch": 0.20389805097451275,
      "grad_norm": 0.6089861937161505,
      "learning_rate": 9.74573287562597e-06,
      "loss": 0.0569,
      "step": 136
    },
    {
      "epoch": 0.20539730134932535,
      "grad_norm": 0.7856372086314439,
      "learning_rate": 9.742012515613536e-06,
      "loss": 0.0778,
      "step": 137
    },
    {
      "epoch": 0.20689655172413793,
      "grad_norm": 0.6553986098650872,
      "learning_rate": 9.738265855914014e-06,
      "loss": 0.0829,
      "step": 138
    },
    {
      "epoch": 0.20839580209895053,
      "grad_norm": 0.7624145641561555,
      "learning_rate": 9.734492917306754e-06,
      "loss": 0.0859,
      "step": 139
    },
    {
      "epoch": 0.2098950524737631,
      "grad_norm": 0.8099846946150838,
      "learning_rate": 9.730693720716866e-06,
      "loss": 0.1017,
      "step": 140
    },
    {
      "epoch": 0.21139430284857572,
      "grad_norm": 0.7480184534829436,
      "learning_rate": 9.72686828721508e-06,
      "loss": 0.0776,
      "step": 141
    },
    {
      "epoch": 0.2128935532233883,
      "grad_norm": 1.0373745775396137,
      "learning_rate": 9.723016638017644e-06,
      "loss": 0.1213,
      "step": 142
    },
    {
      "epoch": 0.2143928035982009,
      "grad_norm": 0.8074585555995454,
      "learning_rate": 9.719138794486198e-06,
      "loss": 0.0755,
      "step": 143
    },
    {
      "epoch": 0.2158920539730135,
      "grad_norm": 0.6689001266671155,
      "learning_rate": 9.715234778127658e-06,
      "loss": 0.064,
      "step": 144
    },
    {
      "epoch": 0.21739130434782608,
      "grad_norm": 0.8333425823399304,
      "learning_rate": 9.711304610594104e-06,
      "loss": 0.0869,
      "step": 145
    },
    {
      "epoch": 0.21889055472263869,
      "grad_norm": 0.9514897312389551,
      "learning_rate": 9.70734831368264e-06,
      "loss": 0.1115,
      "step": 146
    },
    {
      "epoch": 0.22038980509745126,
      "grad_norm": 0.7033544735769441,
      "learning_rate": 9.7033659093353e-06,
      "loss": 0.0706,
      "step": 147
    },
    {
      "epoch": 0.22188905547226387,
      "grad_norm": 0.7858067561706285,
      "learning_rate": 9.699357419638904e-06,
      "loss": 0.0743,
      "step": 148
    },
    {
      "epoch": 0.22338830584707647,
      "grad_norm": 0.8559634154877926,
      "learning_rate": 9.695322866824948e-06,
      "loss": 0.086,
      "step": 149
    },
    {
      "epoch": 0.22488755622188905,
      "grad_norm": 0.78537706232561,
      "learning_rate": 9.691262273269472e-06,
      "loss": 0.0829,
      "step": 150
    },
    {
      "epoch": 0.22638680659670166,
      "grad_norm": 0.576492742874246,
      "learning_rate": 9.687175661492944e-06,
      "loss": 0.0566,
      "step": 151
    },
    {
      "epoch": 0.22788605697151423,
      "grad_norm": 0.6133133978976767,
      "learning_rate": 9.683063054160136e-06,
      "loss": 0.0542,
      "step": 152
    },
    {
      "epoch": 0.22938530734632684,
      "grad_norm": 0.7362888860049434,
      "learning_rate": 9.678924474079986e-06,
      "loss": 0.0702,
      "step": 153
    },
    {
      "epoch": 0.23088455772113944,
      "grad_norm": 0.691481530642289,
      "learning_rate": 9.67475994420548e-06,
      "loss": 0.0632,
      "step": 154
    },
    {
      "epoch": 0.23238380809595202,
      "grad_norm": 0.7645196565994039,
      "learning_rate": 9.670569487633534e-06,
      "loss": 0.0821,
      "step": 155
    },
    {
      "epoch": 0.23388305847076463,
      "grad_norm": 0.8541143440535891,
      "learning_rate": 9.666353127604845e-06,
      "loss": 0.0997,
      "step": 156
    },
    {
      "epoch": 0.2353823088455772,
      "grad_norm": 0.5985842395487234,
      "learning_rate": 9.66211088750378e-06,
      "loss": 0.0703,
      "step": 157
    },
    {
      "epoch": 0.2368815592203898,
      "grad_norm": 0.7585590157432635,
      "learning_rate": 9.657842790858235e-06,
      "loss": 0.0756,
      "step": 158
    },
    {
      "epoch": 0.2383808095952024,
      "grad_norm": 1.3117551704594625,
      "learning_rate": 9.65354886133951e-06,
      "loss": 0.1469,
      "step": 159
    },
    {
      "epoch": 0.239880059970015,
      "grad_norm": 0.5891126317516739,
      "learning_rate": 9.64922912276218e-06,
      "loss": 0.0561,
      "step": 160
    },
    {
      "epoch": 0.2413793103448276,
      "grad_norm": 0.8142327402084952,
      "learning_rate": 9.644883599083959e-06,
      "loss": 0.0764,
      "step": 161
    },
    {
      "epoch": 0.24287856071964017,
      "grad_norm": 0.7378990931484766,
      "learning_rate": 9.640512314405563e-06,
      "loss": 0.0869,
      "step": 162
    },
    {
      "epoch": 0.24437781109445278,
      "grad_norm": 0.6718356112284394,
      "learning_rate": 9.636115292970587e-06,
      "loss": 0.0698,
      "step": 163
    },
    {
      "epoch": 0.24587706146926536,
      "grad_norm": 0.7637294361842925,
      "learning_rate": 9.63169255916536e-06,
      "loss": 0.0746,
      "step": 164
    },
    {
      "epoch": 0.24737631184407796,
      "grad_norm": 0.7111589577816126,
      "learning_rate": 9.627244137518821e-06,
      "loss": 0.0673,
      "step": 165
    },
    {
      "epoch": 0.24887556221889057,
      "grad_norm": 0.9077064440829609,
      "learning_rate": 9.622770052702366e-06,
      "loss": 0.0893,
      "step": 166
    },
    {
      "epoch": 0.25037481259370314,
      "grad_norm": 0.683684381059125,
      "learning_rate": 9.618270329529734e-06,
      "loss": 0.0647,
      "step": 167
    },
    {
      "epoch": 0.2518740629685157,
      "grad_norm": 0.5130623327688315,
      "learning_rate": 9.613744992956844e-06,
      "loss": 0.056,
      "step": 168
    },
    {
      "epoch": 0.25337331334332835,
      "grad_norm": 0.6080909153076831,
      "learning_rate": 9.609194068081682e-06,
      "loss": 0.0696,
      "step": 169
    },
    {
      "epoch": 0.25487256371814093,
      "grad_norm": 0.7098532927522136,
      "learning_rate": 9.60461758014414e-06,
      "loss": 0.0672,
      "step": 170
    },
    {
      "epoch": 0.2563718140929535,
      "grad_norm": 0.7789289784345176,
      "learning_rate": 9.60001555452589e-06,
      "loss": 0.0717,
      "step": 171
    },
    {
      "epoch": 0.25787106446776614,
      "grad_norm": 0.7711670098841029,
      "learning_rate": 9.595388016750236e-06,
      "loss": 0.0695,
      "step": 172
    },
    {
      "epoch": 0.2593703148425787,
      "grad_norm": 0.7922583843130238,
      "learning_rate": 9.590734992481978e-06,
      "loss": 0.0861,
      "step": 173
    },
    {
      "epoch": 0.2608695652173913,
      "grad_norm": 0.8091488821269872,
      "learning_rate": 9.586056507527266e-06,
      "loss": 0.0802,
      "step": 174
    },
    {
      "epoch": 0.2623688155922039,
      "grad_norm": 0.8271632291845011,
      "learning_rate": 9.581352587833455e-06,
      "loss": 0.0829,
      "step": 175
    },
    {
      "epoch": 0.2638680659670165,
      "grad_norm": 1.0912302842544632,
      "learning_rate": 9.576623259488966e-06,
      "loss": 0.1077,
      "step": 176
    },
    {
      "epoch": 0.2653673163418291,
      "grad_norm": 0.9212864757727234,
      "learning_rate": 9.571868548723137e-06,
      "loss": 0.1241,
      "step": 177
    },
    {
      "epoch": 0.26686656671664166,
      "grad_norm": 0.7262731271355038,
      "learning_rate": 9.567088481906084e-06,
      "loss": 0.0668,
      "step": 178
    },
    {
      "epoch": 0.2683658170914543,
      "grad_norm": 0.61401634840511,
      "learning_rate": 9.562283085548546e-06,
      "loss": 0.0647,
      "step": 179
    },
    {
      "epoch": 0.2698650674662669,
      "grad_norm": 1.077743414353555,
      "learning_rate": 9.55745238630174e-06,
      "loss": 0.0986,
      "step": 180
    },
    {
      "epoch": 0.27136431784107945,
      "grad_norm": 0.8689396489859607,
      "learning_rate": 9.552596410957224e-06,
      "loss": 0.1077,
      "step": 181
    },
    {
      "epoch": 0.272863568215892,
      "grad_norm": 0.8574001808695332,
      "learning_rate": 9.547715186446732e-06,
      "loss": 0.0826,
      "step": 182
    },
    {
      "epoch": 0.27436281859070466,
      "grad_norm": 0.743365378147186,
      "learning_rate": 9.542808739842034e-06,
      "loss": 0.0893,
      "step": 183
    },
    {
      "epoch": 0.27586206896551724,
      "grad_norm": 0.7058428351212276,
      "learning_rate": 9.537877098354787e-06,
      "loss": 0.0671,
      "step": 184
    },
    {
      "epoch": 0.2773613193403298,
      "grad_norm": 0.7094423822855045,
      "learning_rate": 9.532920289336378e-06,
      "loss": 0.081,
      "step": 185
    },
    {
      "epoch": 0.27886056971514245,
      "grad_norm": 0.681163364793603,
      "learning_rate": 9.52793834027778e-06,
      "loss": 0.072,
      "step": 186
    },
    {
      "epoch": 0.280359820089955,
      "grad_norm": 0.682661471118478,
      "learning_rate": 9.522931278809393e-06,
      "loss": 0.0802,
      "step": 187
    },
    {
      "epoch": 0.2818590704647676,
      "grad_norm": 0.708064826097952,
      "learning_rate": 9.517899132700889e-06,
      "loss": 0.0765,
      "step": 188
    },
    {
      "epoch": 0.28335832083958024,
      "grad_norm": 0.6664573890160361,
      "learning_rate": 9.512841929861069e-06,
      "loss": 0.0707,
      "step": 189
    },
    {
      "epoch": 0.2848575712143928,
      "grad_norm": 0.6878421774380495,
      "learning_rate": 9.507759698337698e-06,
      "loss": 0.0697,
      "step": 190
    },
    {
      "epoch": 0.2863568215892054,
      "grad_norm": 0.8002411166888339,
      "learning_rate": 9.50265246631735e-06,
      "loss": 0.0861,
      "step": 191
    },
    {
      "epoch": 0.28785607196401797,
      "grad_norm": 0.6601318661148838,
      "learning_rate": 9.49752026212526e-06,
      "loss": 0.0769,
      "step": 192
    },
    {
      "epoch": 0.2893553223388306,
      "grad_norm": 0.7784634786636896,
      "learning_rate": 9.492363114225156e-06,
      "loss": 0.0852,
      "step": 193
    },
    {
      "epoch": 0.2908545727136432,
      "grad_norm": 0.536852818314202,
      "learning_rate": 9.487181051219107e-06,
      "loss": 0.0618,
      "step": 194
    },
    {
      "epoch": 0.29235382308845576,
      "grad_norm": 0.6787532418818419,
      "learning_rate": 9.481974101847371e-06,
      "loss": 0.0738,
      "step": 195
    },
    {
      "epoch": 0.2938530734632684,
      "grad_norm": 0.6745862778791568,
      "learning_rate": 9.476742294988214e-06,
      "loss": 0.0547,
      "step": 196
    },
    {
      "epoch": 0.29535232383808097,
      "grad_norm": 0.7813999924479829,
      "learning_rate": 9.471485659657782e-06,
      "loss": 0.0882,
      "step": 197
    },
    {
      "epoch": 0.29685157421289354,
      "grad_norm": 0.7692296080318167,
      "learning_rate": 9.466204225009905e-06,
      "loss": 0.0776,
      "step": 198
    },
    {
      "epoch": 0.2983508245877061,
      "grad_norm": 0.7242739478490604,
      "learning_rate": 9.460898020335964e-06,
      "loss": 0.0767,
      "step": 199
    },
    {
      "epoch": 0.29985007496251875,
      "grad_norm": 0.5712732682485686,
      "learning_rate": 9.455567075064715e-06,
      "loss": 0.053,
      "step": 200
    },
    {
      "epoch": 0.29985007496251875,
      "eval_loss": 0.09502046555280685,
      "eval_runtime": 2.2162,
      "eval_samples_per_second": 24.366,
      "eval_steps_per_second": 6.317,
      "step": 200
    },
    {
      "epoch": 0.30134932533733133,
      "grad_norm": 0.578589834156033,
      "learning_rate": 9.450211418762123e-06,
      "loss": 0.0536,
      "step": 201
    },
    {
      "epoch": 0.3028485757121439,
      "grad_norm": 0.7795412644735149,
      "learning_rate": 9.444831081131209e-06,
      "loss": 0.0702,
      "step": 202
    },
    {
      "epoch": 0.30434782608695654,
      "grad_norm": 0.896049138341486,
      "learning_rate": 9.439426092011877e-06,
      "loss": 0.0789,
      "step": 203
    },
    {
      "epoch": 0.3058470764617691,
      "grad_norm": 0.5987443493602058,
      "learning_rate": 9.433996481380747e-06,
      "loss": 0.06,
      "step": 204
    },
    {
      "epoch": 0.3073463268365817,
      "grad_norm": 0.943621301313886,
      "learning_rate": 9.428542279351e-06,
      "loss": 0.0948,
      "step": 205
    },
    {
      "epoch": 0.30884557721139433,
      "grad_norm": 0.9241577071267497,
      "learning_rate": 9.423063516172195e-06,
      "loss": 0.0857,
      "step": 206
    },
    {
      "epoch": 0.3103448275862069,
      "grad_norm": 0.6934410660397378,
      "learning_rate": 9.417560222230115e-06,
      "loss": 0.0789,
      "step": 207
    },
    {
      "epoch": 0.3118440779610195,
      "grad_norm": 0.8231926877772439,
      "learning_rate": 9.412032428046594e-06,
      "loss": 0.0782,
      "step": 208
    },
    {
      "epoch": 0.31334332833583206,
      "grad_norm": 0.8389211120135854,
      "learning_rate": 9.40648016427934e-06,
      "loss": 0.0971,
      "step": 209
    },
    {
      "epoch": 0.3148425787106447,
      "grad_norm": 0.7646789808558371,
      "learning_rate": 9.400903461721783e-06,
      "loss": 0.0698,
      "step": 210
    },
    {
      "epoch": 0.31634182908545727,
      "grad_norm": 0.7695622158959841,
      "learning_rate": 9.395302351302881e-06,
      "loss": 0.0764,
      "step": 211
    },
    {
      "epoch": 0.31784107946026985,
      "grad_norm": 0.9728763416248213,
      "learning_rate": 9.38967686408697e-06,
      "loss": 0.1028,
      "step": 212
    },
    {
      "epoch": 0.3193403298350825,
      "grad_norm": 0.8580040406501014,
      "learning_rate": 9.384027031273575e-06,
      "loss": 0.0845,
      "step": 213
    },
    {
      "epoch": 0.32083958020989506,
      "grad_norm": 0.8789545356248069,
      "learning_rate": 9.37835288419725e-06,
      "loss": 0.0903,
      "step": 214
    },
    {
      "epoch": 0.32233883058470764,
      "grad_norm": 0.8051478279643431,
      "learning_rate": 9.372654454327394e-06,
      "loss": 0.1009,
      "step": 215
    },
    {
      "epoch": 0.3238380809595202,
      "grad_norm": 0.674708589638157,
      "learning_rate": 9.366931773268083e-06,
      "loss": 0.0762,
      "step": 216
    },
    {
      "epoch": 0.32533733133433285,
      "grad_norm": 0.8488717992189935,
      "learning_rate": 9.361184872757894e-06,
      "loss": 0.086,
      "step": 217
    },
    {
      "epoch": 0.3268365817091454,
      "grad_norm": 0.7755167526187594,
      "learning_rate": 9.355413784669722e-06,
      "loss": 0.0727,
      "step": 218
    },
    {
      "epoch": 0.328335832083958,
      "grad_norm": 0.7973480354973929,
      "learning_rate": 9.349618541010616e-06,
      "loss": 0.0736,
      "step": 219
    },
    {
      "epoch": 0.32983508245877063,
      "grad_norm": 0.9555885260754625,
      "learning_rate": 9.343799173921591e-06,
      "loss": 0.1031,
      "step": 220
    },
    {
      "epoch": 0.3313343328335832,
      "grad_norm": 0.6617621517932768,
      "learning_rate": 9.337955715677452e-06,
      "loss": 0.0685,
      "step": 221
    },
    {
      "epoch": 0.3328335832083958,
      "grad_norm": 0.8934358896501111,
      "learning_rate": 9.332088198686618e-06,
      "loss": 0.108,
      "step": 222
    },
    {
      "epoch": 0.3343328335832084,
      "grad_norm": 0.7061412115999932,
      "learning_rate": 9.326196655490935e-06,
      "loss": 0.0731,
      "step": 223
    },
    {
      "epoch": 0.335832083958021,
      "grad_norm": 0.8695043884423107,
      "learning_rate": 9.32028111876551e-06,
      "loss": 0.0907,
      "step": 224
    },
    {
      "epoch": 0.3373313343328336,
      "grad_norm": 0.6551837940815775,
      "learning_rate": 9.314341621318512e-06,
      "loss": 0.0849,
      "step": 225
    },
    {
      "epoch": 0.33883058470764615,
      "grad_norm": 0.873086794655443,
      "learning_rate": 9.308378196091006e-06,
      "loss": 0.111,
      "step": 226
    },
    {
      "epoch": 0.3403298350824588,
      "grad_norm": 0.7871278014457453,
      "learning_rate": 9.302390876156756e-06,
      "loss": 0.076,
      "step": 227
    },
    {
      "epoch": 0.34182908545727136,
      "grad_norm": 0.8298853385706422,
      "learning_rate": 9.296379694722051e-06,
      "loss": 0.0948,
      "step": 228
    },
    {
      "epoch": 0.34332833583208394,
      "grad_norm": 0.7205422991420991,
      "learning_rate": 9.29034468512552e-06,
      "loss": 0.078,
      "step": 229
    },
    {
      "epoch": 0.3448275862068966,
      "grad_norm": 0.6601213660598948,
      "learning_rate": 9.284285880837947e-06,
      "loss": 0.0696,
      "step": 230
    },
    {
      "epoch": 0.34632683658170915,
      "grad_norm": 0.7058218065083917,
      "learning_rate": 9.278203315462078e-06,
      "loss": 0.0642,
      "step": 231
    },
    {
      "epoch": 0.34782608695652173,
      "grad_norm": 0.6304791814288884,
      "learning_rate": 9.272097022732444e-06,
      "loss": 0.0716,
      "step": 232
    },
    {
      "epoch": 0.3493253373313343,
      "grad_norm": 0.7088130596590738,
      "learning_rate": 9.26596703651517e-06,
      "loss": 0.0724,
      "step": 233
    },
    {
      "epoch": 0.35082458770614694,
      "grad_norm": 0.9189616350177195,
      "learning_rate": 9.259813390807788e-06,
      "loss": 0.1057,
      "step": 234
    },
    {
      "epoch": 0.3523238380809595,
      "grad_norm": 0.817149981054797,
      "learning_rate": 9.253636119739046e-06,
      "loss": 0.0778,
      "step": 235
    },
    {
      "epoch": 0.3538230884557721,
      "grad_norm": 0.9250132186904606,
      "learning_rate": 9.247435257568724e-06,
      "loss": 0.087,
      "step": 236
    },
    {
      "epoch": 0.3553223388305847,
      "grad_norm": 0.8056621498259231,
      "learning_rate": 9.241210838687438e-06,
      "loss": 0.0903,
      "step": 237
    },
    {
      "epoch": 0.3568215892053973,
      "grad_norm": 0.7462435401453914,
      "learning_rate": 9.23496289761645e-06,
      "loss": 0.0775,
      "step": 238
    },
    {
      "epoch": 0.3583208395802099,
      "grad_norm": 0.9139048645172341,
      "learning_rate": 9.228691469007487e-06,
      "loss": 0.1075,
      "step": 239
    },
    {
      "epoch": 0.3598200899550225,
      "grad_norm": 0.6711487951354117,
      "learning_rate": 9.222396587642528e-06,
      "loss": 0.061,
      "step": 240
    },
    {
      "epoch": 0.3613193403298351,
      "grad_norm": 0.6826520743837196,
      "learning_rate": 9.216078288433632e-06,
      "loss": 0.0781,
      "step": 241
    },
    {
      "epoch": 0.36281859070464767,
      "grad_norm": 0.821923841879159,
      "learning_rate": 9.209736606422736e-06,
      "loss": 0.0873,
      "step": 242
    },
    {
      "epoch": 0.36431784107946025,
      "grad_norm": 0.771043260221293,
      "learning_rate": 9.203371576781457e-06,
      "loss": 0.0818,
      "step": 243
    },
    {
      "epoch": 0.3658170914542729,
      "grad_norm": 0.5801927109255428,
      "learning_rate": 9.1969832348109e-06,
      "loss": 0.0571,
      "step": 244
    },
    {
      "epoch": 0.36731634182908546,
      "grad_norm": 0.7196709987685058,
      "learning_rate": 9.190571615941462e-06,
      "loss": 0.0783,
      "step": 245
    },
    {
      "epoch": 0.36881559220389803,
      "grad_norm": 0.7246665800079929,
      "learning_rate": 9.18413675573264e-06,
      "loss": 0.0615,
      "step": 246
    },
    {
      "epoch": 0.37031484257871067,
      "grad_norm": 0.9384672924000713,
      "learning_rate": 9.177678689872831e-06,
      "loss": 0.1175,
      "step": 247
    },
    {
      "epoch": 0.37181409295352325,
      "grad_norm": 0.6406943217584374,
      "learning_rate": 9.171197454179124e-06,
      "loss": 0.065,
      "step": 248
    },
    {
      "epoch": 0.3733133433283358,
      "grad_norm": 0.73528322901661,
      "learning_rate": 9.16469308459712e-06,
      "loss": 0.0745,
      "step": 249
    },
    {
      "epoch": 0.3748125937031484,
      "grad_norm": 0.6793755112988502,
      "learning_rate": 9.158165617200717e-06,
      "loss": 0.0619,
      "step": 250
    },
    {
      "epoch": 0.37631184407796103,
      "grad_norm": 0.8782683002729995,
      "learning_rate": 9.151615088191918e-06,
      "loss": 0.0782,
      "step": 251
    },
    {
      "epoch": 0.3778110944527736,
      "grad_norm": 0.5388534428374847,
      "learning_rate": 9.14504153390063e-06,
      "loss": 0.0611,
      "step": 252
    },
    {
      "epoch": 0.3793103448275862,
      "grad_norm": 0.8568768013761957,
      "learning_rate": 9.138444990784455e-06,
      "loss": 0.0989,
      "step": 253
    },
    {
      "epoch": 0.3808095952023988,
      "grad_norm": 0.7085521317689302,
      "learning_rate": 9.131825495428496e-06,
      "loss": 0.0676,
      "step": 254
    },
    {
      "epoch": 0.3823088455772114,
      "grad_norm": 0.755863732822612,
      "learning_rate": 9.125183084545158e-06,
      "loss": 0.0888,
      "step": 255
    },
    {
      "epoch": 0.383808095952024,
      "grad_norm": 0.834325854025348,
      "learning_rate": 9.118517794973925e-06,
      "loss": 0.0896,
      "step": 256
    },
    {
      "epoch": 0.3853073463268366,
      "grad_norm": 0.7735994509367796,
      "learning_rate": 9.111829663681182e-06,
      "loss": 0.0827,
      "step": 257
    },
    {
      "epoch": 0.3868065967016492,
      "grad_norm": 0.8935713034371032,
      "learning_rate": 9.105118727759984e-06,
      "loss": 0.0864,
      "step": 258
    },
    {
      "epoch": 0.38830584707646176,
      "grad_norm": 0.9831033044846298,
      "learning_rate": 9.098385024429875e-06,
      "loss": 0.1002,
      "step": 259
    },
    {
      "epoch": 0.38980509745127434,
      "grad_norm": 0.6041240524948254,
      "learning_rate": 9.09162859103666e-06,
      "loss": 0.0624,
      "step": 260
    },
    {
      "epoch": 0.391304347826087,
      "grad_norm": 0.8785292457479368,
      "learning_rate": 9.08484946505221e-06,
      "loss": 0.0883,
      "step": 261
    },
    {
      "epoch": 0.39280359820089955,
      "grad_norm": 0.8512256759655724,
      "learning_rate": 9.078047684074254e-06,
      "loss": 0.0881,
      "step": 262
    },
    {
      "epoch": 0.39430284857571213,
      "grad_norm": 0.7718531609263276,
      "learning_rate": 9.071223285826166e-06,
      "loss": 0.0844,
      "step": 263
    },
    {
      "epoch": 0.39580209895052476,
      "grad_norm": 0.6904846259570785,
      "learning_rate": 9.064376308156754e-06,
      "loss": 0.0709,
      "step": 264
    },
    {
      "epoch": 0.39730134932533734,
      "grad_norm": 0.8253332596642009,
      "learning_rate": 9.057506789040063e-06,
      "loss": 0.0897,
      "step": 265
    },
    {
      "epoch": 0.3988005997001499,
      "grad_norm": 0.8665307879480542,
      "learning_rate": 9.050614766575147e-06,
      "loss": 0.0963,
      "step": 266
    },
    {
      "epoch": 0.4002998500749625,
      "grad_norm": 0.6710349936016295,
      "learning_rate": 9.043700278985867e-06,
      "loss": 0.0657,
      "step": 267
    },
    {
      "epoch": 0.4017991004497751,
      "grad_norm": 0.9014592529969518,
      "learning_rate": 9.03676336462068e-06,
      "loss": 0.0983,
      "step": 268
    },
    {
      "epoch": 0.4032983508245877,
      "grad_norm": 0.7361385879105462,
      "learning_rate": 9.029804061952426e-06,
      "loss": 0.0827,
      "step": 269
    },
    {
      "epoch": 0.4047976011994003,
      "grad_norm": 0.5979082045621211,
      "learning_rate": 9.022822409578106e-06,
      "loss": 0.0623,
      "step": 270
    },
    {
      "epoch": 0.4062968515742129,
      "grad_norm": 0.5950562398215975,
      "learning_rate": 9.015818446218683e-06,
      "loss": 0.0544,
      "step": 271
    },
    {
      "epoch": 0.4077961019490255,
      "grad_norm": 0.85431295365093,
      "learning_rate": 9.008792210718854e-06,
      "loss": 0.097,
      "step": 272
    },
    {
      "epoch": 0.40929535232383807,
      "grad_norm": 1.1577675086258126,
      "learning_rate": 9.00174374204684e-06,
      "loss": 0.1347,
      "step": 273
    },
    {
      "epoch": 0.4107946026986507,
      "grad_norm": 0.8072100067535496,
      "learning_rate": 8.994673079294171e-06,
      "loss": 0.0868,
      "step": 274
    },
    {
      "epoch": 0.4122938530734633,
      "grad_norm": 0.7273464527227016,
      "learning_rate": 8.987580261675466e-06,
      "loss": 0.0641,
      "step": 275
    },
    {
      "epoch": 0.41379310344827586,
      "grad_norm": 0.8830597842742653,
      "learning_rate": 8.98046532852822e-06,
      "loss": 0.0946,
      "step": 276
    },
    {
      "epoch": 0.41529235382308843,
      "grad_norm": 0.7023383756227846,
      "learning_rate": 8.973328319312577e-06,
      "loss": 0.0638,
      "step": 277
    },
    {
      "epoch": 0.41679160419790107,
      "grad_norm": 0.6487198314996288,
      "learning_rate": 8.966169273611125e-06,
      "loss": 0.0715,
      "step": 278
    },
    {
      "epoch": 0.41829085457271364,
      "grad_norm": 0.9332418353934671,
      "learning_rate": 8.958988231128665e-06,
      "loss": 0.0989,
      "step": 279
    },
    {
      "epoch": 0.4197901049475262,
      "grad_norm": 0.7176519611350742,
      "learning_rate": 8.95178523169199e-06,
      "loss": 0.0834,
      "step": 280
    },
    {
      "epoch": 0.42128935532233885,
      "grad_norm": 0.6903363160336365,
      "learning_rate": 8.944560315249676e-06,
      "loss": 0.0708,
      "step": 281
    },
    {
      "epoch": 0.42278860569715143,
      "grad_norm": 0.7425281502230455,
      "learning_rate": 8.937313521871846e-06,
      "loss": 0.0845,
      "step": 282
    },
    {
      "epoch": 0.424287856071964,
      "grad_norm": 0.7162201845019355,
      "learning_rate": 8.930044891749962e-06,
      "loss": 0.0747,
      "step": 283
    },
    {
      "epoch": 0.4257871064467766,
      "grad_norm": 0.8952937283216326,
      "learning_rate": 8.922754465196591e-06,
      "loss": 0.108,
      "step": 284
    },
    {
      "epoch": 0.4272863568215892,
      "grad_norm": 0.5601686537780921,
      "learning_rate": 8.915442282645183e-06,
      "loss": 0.0624,
      "step": 285
    },
    {
      "epoch": 0.4287856071964018,
      "grad_norm": 0.7170929018955168,
      "learning_rate": 8.908108384649856e-06,
      "loss": 0.08,
      "step": 286
    },
    {
      "epoch": 0.4302848575712144,
      "grad_norm": 0.8400689256948733,
      "learning_rate": 8.900752811885152e-06,
      "loss": 0.1011,
      "step": 287
    },
    {
      "epoch": 0.431784107946027,
      "grad_norm": 0.5672607097631437,
      "learning_rate": 8.893375605145837e-06,
      "loss": 0.053,
      "step": 288
    },
    {
      "epoch": 0.4332833583208396,
      "grad_norm": 0.8241483701021035,
      "learning_rate": 8.885976805346651e-06,
      "loss": 0.1067,
      "step": 289
    },
    {
      "epoch": 0.43478260869565216,
      "grad_norm": 0.7467703708241894,
      "learning_rate": 8.8785564535221e-06,
      "loss": 0.0752,
      "step": 290
    },
    {
      "epoch": 0.4362818590704648,
      "grad_norm": 0.6572153596660257,
      "learning_rate": 8.871114590826211e-06,
      "loss": 0.0757,
      "step": 291
    },
    {
      "epoch": 0.43778110944527737,
      "grad_norm": 1.1389395507345843,
      "learning_rate": 8.86365125853232e-06,
      "loss": 0.0984,
      "step": 292
    },
    {
      "epoch": 0.43928035982008995,
      "grad_norm": 0.7579028829145666,
      "learning_rate": 8.85616649803283e-06,
      "loss": 0.0768,
      "step": 293
    },
    {
      "epoch": 0.4407796101949025,
      "grad_norm": 0.558160964984709,
      "learning_rate": 8.84866035083899e-06,
      "loss": 0.0619,
      "step": 294
    },
    {
      "epoch": 0.44227886056971516,
      "grad_norm": 0.9450442112355608,
      "learning_rate": 8.841132858580661e-06,
      "loss": 0.1077,
      "step": 295
    },
    {
      "epoch": 0.44377811094452774,
      "grad_norm": 0.7534521372624636,
      "learning_rate": 8.833584063006088e-06,
      "loss": 0.0888,
      "step": 296
    },
    {
      "epoch": 0.4452773613193403,
      "grad_norm": 0.7604220287530142,
      "learning_rate": 8.826014005981662e-06,
      "loss": 0.0733,
      "step": 297
    },
    {
      "epoch": 0.44677661169415295,
      "grad_norm": 0.7324521371188807,
      "learning_rate": 8.818422729491693e-06,
      "loss": 0.0804,
      "step": 298
    },
    {
      "epoch": 0.4482758620689655,
      "grad_norm": 0.7478260621987316,
      "learning_rate": 8.810810275638183e-06,
      "loss": 0.0914,
      "step": 299
    },
    {
      "epoch": 0.4497751124437781,
      "grad_norm": 0.649311935991274,
      "learning_rate": 8.803176686640577e-06,
      "loss": 0.0707,
      "step": 300
    },
    {
      "epoch": 0.4512743628185907,
      "grad_norm": 0.9969458490393408,
      "learning_rate": 8.795522004835543e-06,
      "loss": 0.1178,
      "step": 301
    },
    {
      "epoch": 0.4527736131934033,
      "grad_norm": 0.6982889858677304,
      "learning_rate": 8.787846272676728e-06,
      "loss": 0.0675,
      "step": 302
    },
    {
      "epoch": 0.4542728635682159,
      "grad_norm": 0.658758633783918,
      "learning_rate": 8.780149532734531e-06,
      "loss": 0.066,
      "step": 303
    },
    {
      "epoch": 0.45577211394302847,
      "grad_norm": 0.7663198597374845,
      "learning_rate": 8.772431827695862e-06,
      "loss": 0.0885,
      "step": 304
    },
    {
      "epoch": 0.4572713643178411,
      "grad_norm": 0.8575190120874159,
      "learning_rate": 8.764693200363897e-06,
      "loss": 0.0843,
      "step": 305
    },
    {
      "epoch": 0.4587706146926537,
      "grad_norm": 0.7038218132010158,
      "learning_rate": 8.756933693657863e-06,
      "loss": 0.0658,
      "step": 306
    },
    {
      "epoch": 0.46026986506746626,
      "grad_norm": 0.7229519530353599,
      "learning_rate": 8.749153350612774e-06,
      "loss": 0.0772,
      "step": 307
    },
    {
      "epoch": 0.4617691154422789,
      "grad_norm": 0.8590316027345603,
      "learning_rate": 8.74135221437921e-06,
      "loss": 0.0926,
      "step": 308
    },
    {
      "epoch": 0.46326836581709147,
      "grad_norm": 0.6804530148363489,
      "learning_rate": 8.733530328223076e-06,
      "loss": 0.0645,
      "step": 309
    },
    {
      "epoch": 0.46476761619190404,
      "grad_norm": 0.645899395322459,
      "learning_rate": 8.725687735525347e-06,
      "loss": 0.0665,
      "step": 310
    },
    {
      "epoch": 0.4662668665667166,
      "grad_norm": 0.9715716517167565,
      "learning_rate": 8.71782447978185e-06,
      "loss": 0.1014,
      "step": 311
    },
    {
      "epoch": 0.46776611694152925,
      "grad_norm": 0.8500232439094785,
      "learning_rate": 8.709940604603006e-06,
      "loss": 0.1048,
      "step": 312
    },
    {
      "epoch": 0.46926536731634183,
      "grad_norm": 0.6915587541611212,
      "learning_rate": 8.702036153713594e-06,
      "loss": 0.075,
      "step": 313
    },
    {
      "epoch": 0.4707646176911544,
      "grad_norm": 1.007945801754584,
      "learning_rate": 8.694111170952508e-06,
      "loss": 0.1114,
      "step": 314
    },
    {
      "epoch": 0.47226386806596704,
      "grad_norm": 0.6044273086971291,
      "learning_rate": 8.686165700272513e-06,
      "loss": 0.0742,
      "step": 315
    },
    {
      "epoch": 0.4737631184407796,
      "grad_norm": 0.7508422468613741,
      "learning_rate": 8.678199785740003e-06,
      "loss": 0.0813,
      "step": 316
    },
    {
      "epoch": 0.4752623688155922,
      "grad_norm": 0.7055230941601637,
      "learning_rate": 8.670213471534759e-06,
      "loss": 0.0894,
      "step": 317
    },
    {
      "epoch": 0.4767616191904048,
      "grad_norm": 0.7006081479583219,
      "learning_rate": 8.662206801949694e-06,
      "loss": 0.0716,
      "step": 318
    },
    {
      "epoch": 0.4782608695652174,
      "grad_norm": 0.6517492341534331,
      "learning_rate": 8.65417982139062e-06,
      "loss": 0.0603,
      "step": 319
    },
    {
      "epoch": 0.47976011994003,
      "grad_norm": 0.8639212460521969,
      "learning_rate": 8.646132574375994e-06,
      "loss": 0.0804,
      "step": 320
    },
    {
      "epoch": 0.48125937031484256,
      "grad_norm": 1.1192636832452907,
      "learning_rate": 8.638065105536669e-06,
      "loss": 0.1276,
      "step": 321
    },
    {
      "epoch": 0.4827586206896552,
      "grad_norm": 0.7636209489333018,
      "learning_rate": 8.629977459615655e-06,
      "loss": 0.0724,
      "step": 322
    },
    {
      "epoch": 0.48425787106446777,
      "grad_norm": 0.8251074465355053,
      "learning_rate": 8.621869681467865e-06,
      "loss": 0.0858,
      "step": 323
    },
    {
      "epoch": 0.48575712143928035,
      "grad_norm": 0.6389357966837655,
      "learning_rate": 8.613741816059867e-06,
      "loss": 0.0606,
      "step": 324
    },
    {
      "epoch": 0.487256371814093,
      "grad_norm": 2.373516846634749,
      "learning_rate": 8.605593908469635e-06,
      "loss": 0.0882,
      "step": 325
    },
    {
      "epoch": 0.48875562218890556,
      "grad_norm": 0.6240583503623685,
      "learning_rate": 8.597426003886295e-06,
      "loss": 0.0694,
      "step": 326
    },
    {
      "epoch": 0.49025487256371814,
      "grad_norm": 0.8713202884931421,
      "learning_rate": 8.58923814760989e-06,
      "loss": 0.0851,
      "step": 327
    },
    {
      "epoch": 0.4917541229385307,
      "grad_norm": 0.6532252627850518,
      "learning_rate": 8.581030385051105e-06,
      "loss": 0.0682,
      "step": 328
    },
    {
      "epoch": 0.49325337331334335,
      "grad_norm": 0.741611101115977,
      "learning_rate": 8.572802761731031e-06,
      "loss": 0.0835,
      "step": 329
    },
    {
      "epoch": 0.4947526236881559,
      "grad_norm": 1.0973441351896815,
      "learning_rate": 8.564555323280913e-06,
      "loss": 0.1136,
      "step": 330
    },
    {
      "epoch": 0.4962518740629685,
      "grad_norm": 0.7238407003523619,
      "learning_rate": 8.556288115441887e-06,
      "loss": 0.0718,
      "step": 331
    },
    {
      "epoch": 0.49775112443778113,
      "grad_norm": 0.7509583422509074,
      "learning_rate": 8.548001184064733e-06,
      "loss": 0.0795,
      "step": 332
    },
    {
      "epoch": 0.4992503748125937,
      "grad_norm": 0.6388142646334699,
      "learning_rate": 8.539694575109626e-06,
      "loss": 0.0585,
      "step": 333
    },
    {
      "epoch": 0.5007496251874063,
      "grad_norm": 0.6488380572895145,
      "learning_rate": 8.531368334645865e-06,
      "loss": 0.07,
      "step": 334
    },
    {
      "epoch": 0.5022488755622189,
      "grad_norm": 0.7310206713484633,
      "learning_rate": 8.523022508851634e-06,
      "loss": 0.0951,
      "step": 335
    },
    {
      "epoch": 0.5037481259370314,
      "grad_norm": 0.6720227418681202,
      "learning_rate": 8.514657144013738e-06,
      "loss": 0.0671,
      "step": 336
    },
    {
      "epoch": 0.5052473763118441,
      "grad_norm": 0.7348300745775271,
      "learning_rate": 8.506272286527346e-06,
      "loss": 0.0757,
      "step": 337
    },
    {
      "epoch": 0.5067466266866567,
      "grad_norm": 0.6456937543323911,
      "learning_rate": 8.497867982895741e-06,
      "loss": 0.0583,
      "step": 338
    },
    {
      "epoch": 0.5082458770614693,
      "grad_norm": 0.6639276492188844,
      "learning_rate": 8.489444279730046e-06,
      "loss": 0.0753,
      "step": 339
    },
    {
      "epoch": 0.5097451274362819,
      "grad_norm": 0.6744581580711936,
      "learning_rate": 8.481001223748986e-06,
      "loss": 0.0557,
      "step": 340
    },
    {
      "epoch": 0.5112443778110944,
      "grad_norm": 0.8565585037246568,
      "learning_rate": 8.47253886177861e-06,
      "loss": 0.1016,
      "step": 341
    },
    {
      "epoch": 0.512743628185907,
      "grad_norm": 1.0265515335676028,
      "learning_rate": 8.464057240752046e-06,
      "loss": 0.0808,
      "step": 342
    },
    {
      "epoch": 0.5142428785607196,
      "grad_norm": 0.7127437427807087,
      "learning_rate": 8.455556407709235e-06,
      "loss": 0.0722,
      "step": 343
    },
    {
      "epoch": 0.5157421289355323,
      "grad_norm": 0.6986054224251383,
      "learning_rate": 8.447036409796663e-06,
      "loss": 0.0793,
      "step": 344
    },
    {
      "epoch": 0.5172413793103449,
      "grad_norm": 0.8020535407080561,
      "learning_rate": 8.438497294267117e-06,
      "loss": 0.0857,
      "step": 345
    },
    {
      "epoch": 0.5187406296851574,
      "grad_norm": 0.7957704355578628,
      "learning_rate": 8.429939108479403e-06,
      "loss": 0.0754,
      "step": 346
    },
    {
      "epoch": 0.52023988005997,
      "grad_norm": 0.8212675674821511,
      "learning_rate": 8.421361899898095e-06,
      "loss": 0.0747,
      "step": 347
    },
    {
      "epoch": 0.5217391304347826,
      "grad_norm": 0.6632812085173838,
      "learning_rate": 8.412765716093273e-06,
      "loss": 0.0738,
      "step": 348
    },
    {
      "epoch": 0.5232383808095952,
      "grad_norm": 0.91075775549987,
      "learning_rate": 8.404150604740248e-06,
      "loss": 0.1049,
      "step": 349
    },
    {
      "epoch": 0.5247376311844077,
      "grad_norm": 0.7259947879715208,
      "learning_rate": 8.395516613619315e-06,
      "loss": 0.0628,
      "step": 350
    },
    {
      "epoch": 0.5262368815592204,
      "grad_norm": 0.5902748233869028,
      "learning_rate": 8.386863790615472e-06,
      "loss": 0.0588,
      "step": 351
    },
    {
      "epoch": 0.527736131934033,
      "grad_norm": 0.7602940355359913,
      "learning_rate": 8.378192183718158e-06,
      "loss": 0.0815,
      "step": 352
    },
    {
      "epoch": 0.5292353823088456,
      "grad_norm": 0.9450097933050396,
      "learning_rate": 8.369501841021e-06,
      "loss": 0.1104,
      "step": 353
    },
    {
      "epoch": 0.5307346326836582,
      "grad_norm": 0.6334439171356193,
      "learning_rate": 8.360792810721522e-06,
      "loss": 0.0754,
      "step": 354
    },
    {
      "epoch": 0.5322338830584707,
      "grad_norm": 0.8143753544727452,
      "learning_rate": 8.352065141120902e-06,
      "loss": 0.0901,
      "step": 355
    },
    {
      "epoch": 0.5337331334332833,
      "grad_norm": 0.9492987126751995,
      "learning_rate": 8.343318880623688e-06,
      "loss": 0.085,
      "step": 356
    },
    {
      "epoch": 0.5352323838080959,
      "grad_norm": 0.7933684074296237,
      "learning_rate": 8.334554077737535e-06,
      "loss": 0.0816,
      "step": 357
    },
    {
      "epoch": 0.5367316341829086,
      "grad_norm": 0.8431590900929863,
      "learning_rate": 8.325770781072939e-06,
      "loss": 0.0808,
      "step": 358
    },
    {
      "epoch": 0.5382308845577212,
      "grad_norm": 0.6852702622093223,
      "learning_rate": 8.316969039342963e-06,
      "loss": 0.0767,
      "step": 359
    },
    {
      "epoch": 0.5397301349325337,
      "grad_norm": 0.7740447814473573,
      "learning_rate": 8.30814890136297e-06,
      "loss": 0.1003,
      "step": 360
    },
    {
      "epoch": 0.5412293853073463,
      "grad_norm": 0.8992601633057014,
      "learning_rate": 8.299310416050345e-06,
      "loss": 0.1139,
      "step": 361
    },
    {
      "epoch": 0.5427286356821589,
      "grad_norm": 0.6846052707268697,
      "learning_rate": 8.290453632424236e-06,
      "loss": 0.0776,
      "step": 362
    },
    {
      "epoch": 0.5442278860569715,
      "grad_norm": 0.9358750900284801,
      "learning_rate": 8.281578599605269e-06,
      "loss": 0.0886,
      "step": 363
    },
    {
      "epoch": 0.545727136431784,
      "grad_norm": 0.9506859994014473,
      "learning_rate": 8.272685366815287e-06,
      "loss": 0.1021,
      "step": 364
    },
    {
      "epoch": 0.5472263868065967,
      "grad_norm": 0.8567211837209996,
      "learning_rate": 8.26377398337707e-06,
      "loss": 0.0806,
      "step": 365
    },
    {
      "epoch": 0.5487256371814093,
      "grad_norm": 0.694799780941582,
      "learning_rate": 8.254844498714063e-06,
      "loss": 0.0782,
      "step": 366
    },
    {
      "epoch": 0.5502248875562219,
      "grad_norm": 0.8504533846661089,
      "learning_rate": 8.2458969623501e-06,
      "loss": 0.0989,
      "step": 367
    },
    {
      "epoch": 0.5517241379310345,
      "grad_norm": 0.7266391525233931,
      "learning_rate": 8.23693142390914e-06,
      "loss": 0.0818,
      "step": 368
    },
    {
      "epoch": 0.553223388305847,
      "grad_norm": 0.7341200018143592,
      "learning_rate": 8.227947933114971e-06,
      "loss": 0.0873,
      "step": 369
    },
    {
      "epoch": 0.5547226386806596,
      "grad_norm": 0.7706948723699811,
      "learning_rate": 8.218946539790957e-06,
      "loss": 0.0828,
      "step": 370
    },
    {
      "epoch": 0.5562218890554723,
      "grad_norm": 0.7124547400210968,
      "learning_rate": 8.209927293859746e-06,
      "loss": 0.0777,
      "step": 371
    },
    {
      "epoch": 0.5577211394302849,
      "grad_norm": 0.7673330749326083,
      "learning_rate": 8.200890245342999e-06,
      "loss": 0.0849,
      "step": 372
    },
    {
      "epoch": 0.5592203898050975,
      "grad_norm": 1.077269246282627,
      "learning_rate": 8.191835444361113e-06,
      "loss": 0.1294,
      "step": 373
    },
    {
      "epoch": 0.56071964017991,
      "grad_norm": 0.6641097826169041,
      "learning_rate": 8.182762941132944e-06,
      "loss": 0.063,
      "step": 374
    },
    {
      "epoch": 0.5622188905547226,
      "grad_norm": 0.6870076682864237,
      "learning_rate": 8.173672785975522e-06,
      "loss": 0.0613,
      "step": 375
    },
    {
      "epoch": 0.5637181409295352,
      "grad_norm": 0.7570209436428715,
      "learning_rate": 8.16456502930378e-06,
      "loss": 0.0855,
      "step": 376
    },
    {
      "epoch": 0.5652173913043478,
      "grad_norm": 0.7825079054087174,
      "learning_rate": 8.155439721630265e-06,
      "loss": 0.0934,
      "step": 377
    },
    {
      "epoch": 0.5667166416791605,
      "grad_norm": 0.6440996553536563,
      "learning_rate": 8.146296913564872e-06,
      "loss": 0.0671,
      "step": 378
    },
    {
      "epoch": 0.568215892053973,
      "grad_norm": 0.7287055111343522,
      "learning_rate": 8.13713665581455e-06,
      "loss": 0.0791,
      "step": 379
    },
    {
      "epoch": 0.5697151424287856,
      "grad_norm": 0.6443138601410697,
      "learning_rate": 8.127958999183027e-06,
      "loss": 0.0707,
      "step": 380
    },
    {
      "epoch": 0.5712143928035982,
      "grad_norm": 0.6330782809596815,
      "learning_rate": 8.118763994570528e-06,
      "loss": 0.0613,
      "step": 381
    },
    {
      "epoch": 0.5727136431784108,
      "grad_norm": 0.7341629966252481,
      "learning_rate": 8.109551692973487e-06,
      "loss": 0.0788,
      "step": 382
    },
    {
      "epoch": 0.5742128935532234,
      "grad_norm": 0.9159527879260371,
      "learning_rate": 8.100322145484275e-06,
      "loss": 0.0987,
      "step": 383
    },
    {
      "epoch": 0.5757121439280359,
      "grad_norm": 0.7978650300642782,
      "learning_rate": 8.091075403290905e-06,
      "loss": 0.0891,
      "step": 384
    },
    {
      "epoch": 0.5772113943028486,
      "grad_norm": 0.747592624000814,
      "learning_rate": 8.081811517676759e-06,
      "loss": 0.0928,
      "step": 385
    },
    {
      "epoch": 0.5787106446776612,
      "grad_norm": 0.6989142250360894,
      "learning_rate": 8.072530540020294e-06,
      "loss": 0.078,
      "step": 386
    },
    {
      "epoch": 0.5802098950524738,
      "grad_norm": 0.7727284204403324,
      "learning_rate": 8.063232521794762e-06,
      "loss": 0.0875,
      "step": 387
    },
    {
      "epoch": 0.5817091454272864,
      "grad_norm": 0.5669492870847148,
      "learning_rate": 8.053917514567927e-06,
      "loss": 0.0685,
      "step": 388
    },
    {
      "epoch": 0.5832083958020989,
      "grad_norm": 0.7831285273183912,
      "learning_rate": 8.04458557000177e-06,
      "loss": 0.0844,
      "step": 389
    },
    {
      "epoch": 0.5847076461769115,
      "grad_norm": 0.9291755660694857,
      "learning_rate": 8.035236739852214e-06,
      "loss": 0.1137,
      "step": 390
    },
    {
      "epoch": 0.5862068965517241,
      "grad_norm": 0.6208201532981111,
      "learning_rate": 8.025871075968828e-06,
      "loss": 0.0569,
      "step": 391
    },
    {
      "epoch": 0.5877061469265368,
      "grad_norm": 0.6751986830876698,
      "learning_rate": 8.016488630294539e-06,
      "loss": 0.0678,
      "step": 392
    },
    {
      "epoch": 0.5892053973013494,
      "grad_norm": 0.856023261629115,
      "learning_rate": 8.007089454865358e-06,
      "loss": 0.0848,
      "step": 393
    },
    {
      "epoch": 0.5907046476761619,
      "grad_norm": 0.7750431443323113,
      "learning_rate": 7.997673601810071e-06,
      "loss": 0.0819,
      "step": 394
    },
    {
      "epoch": 0.5922038980509745,
      "grad_norm": 0.6248737031285564,
      "learning_rate": 7.988241123349965e-06,
      "loss": 0.0618,
      "step": 395
    },
    {
      "epoch": 0.5937031484257871,
      "grad_norm": 0.9975286169456027,
      "learning_rate": 7.97879207179853e-06,
      "loss": 0.1029,
      "step": 396
    },
    {
      "epoch": 0.5952023988005997,
      "grad_norm": 0.7246983873837445,
      "learning_rate": 7.969326499561173e-06,
      "loss": 0.0821,
      "step": 397
    },
    {
      "epoch": 0.5967016491754122,
      "grad_norm": 0.7287094342897744,
      "learning_rate": 7.95984445913493e-06,
      "loss": 0.0701,
      "step": 398
    },
    {
      "epoch": 0.5982008995502249,
      "grad_norm": 0.629329765819991,
      "learning_rate": 7.950346003108167e-06,
      "loss": 0.0691,
      "step": 399
    },
    {
      "epoch": 0.5997001499250375,
      "grad_norm": 0.629161627538197,
      "learning_rate": 7.940831184160294e-06,
      "loss": 0.0633,
      "step": 400
    },
    {
      "epoch": 0.5997001499250375,
      "eval_loss": 0.09152090549468994,
      "eval_runtime": 2.1873,
      "eval_samples_per_second": 24.688,
      "eval_steps_per_second": 6.401,
      "step": 400
    },
    {
      "epoch": 0.6011994002998501,
      "grad_norm": 0.5798935579036949,
      "learning_rate": 7.93130005506147e-06,
      "loss": 0.0575,
      "step": 401
    },
    {
      "epoch": 0.6026986506746627,
      "grad_norm": 0.6464065903488889,
      "learning_rate": 7.921752668672316e-06,
      "loss": 0.0864,
      "step": 402
    },
    {
      "epoch": 0.6041979010494752,
      "grad_norm": 0.7815362566143813,
      "learning_rate": 7.912189077943613e-06,
      "loss": 0.0787,
      "step": 403
    },
    {
      "epoch": 0.6056971514242878,
      "grad_norm": 0.7928929835156217,
      "learning_rate": 7.902609335916015e-06,
      "loss": 0.0743,
      "step": 404
    },
    {
      "epoch": 0.6071964017991005,
      "grad_norm": 0.815663635435752,
      "learning_rate": 7.893013495719752e-06,
      "loss": 0.089,
      "step": 405
    },
    {
      "epoch": 0.6086956521739131,
      "grad_norm": 0.9546555701097286,
      "learning_rate": 7.883401610574338e-06,
      "loss": 0.0961,
      "step": 406
    },
    {
      "epoch": 0.6101949025487257,
      "grad_norm": 0.8546473442859561,
      "learning_rate": 7.873773733788268e-06,
      "loss": 0.1007,
      "step": 407
    },
    {
      "epoch": 0.6116941529235382,
      "grad_norm": 1.1008251181263533,
      "learning_rate": 7.864129918758738e-06,
      "loss": 0.1181,
      "step": 408
    },
    {
      "epoch": 0.6131934032983508,
      "grad_norm": 0.7260120567436672,
      "learning_rate": 7.854470218971333e-06,
      "loss": 0.0824,
      "step": 409
    },
    {
      "epoch": 0.6146926536731634,
      "grad_norm": 0.7679123231991849,
      "learning_rate": 7.844794687999737e-06,
      "loss": 0.0793,
      "step": 410
    },
    {
      "epoch": 0.616191904047976,
      "grad_norm": 0.9011773063151945,
      "learning_rate": 7.835103379505433e-06,
      "loss": 0.1015,
      "step": 411
    },
    {
      "epoch": 0.6176911544227887,
      "grad_norm": 0.6036873455748664,
      "learning_rate": 7.825396347237413e-06,
      "loss": 0.0542,
      "step": 412
    },
    {
      "epoch": 0.6191904047976012,
      "grad_norm": 0.74148409427925,
      "learning_rate": 7.815673645031871e-06,
      "loss": 0.0754,
      "step": 413
    },
    {
      "epoch": 0.6206896551724138,
      "grad_norm": 0.8381382389777878,
      "learning_rate": 7.805935326811913e-06,
      "loss": 0.0677,
      "step": 414
    },
    {
      "epoch": 0.6221889055472264,
      "grad_norm": 0.6098455535842088,
      "learning_rate": 7.796181446587244e-06,
      "loss": 0.0562,
      "step": 415
    },
    {
      "epoch": 0.623688155922039,
      "grad_norm": 0.8526599565989879,
      "learning_rate": 7.786412058453886e-06,
      "loss": 0.0942,
      "step": 416
    },
    {
      "epoch": 0.6251874062968515,
      "grad_norm": 0.729366095624253,
      "learning_rate": 7.776627216593863e-06,
      "loss": 0.075,
      "step": 417
    },
    {
      "epoch": 0.6266866566716641,
      "grad_norm": 0.5560988813532235,
      "learning_rate": 7.766826975274916e-06,
      "loss": 0.0613,
      "step": 418
    },
    {
      "epoch": 0.6281859070464768,
      "grad_norm": 1.3742609246680018,
      "learning_rate": 7.75701138885018e-06,
      "loss": 0.139,
      "step": 419
    },
    {
      "epoch": 0.6296851574212894,
      "grad_norm": 0.8722686214062148,
      "learning_rate": 7.747180511757908e-06,
      "loss": 0.0812,
      "step": 420
    },
    {
      "epoch": 0.631184407796102,
      "grad_norm": 0.7608449077550606,
      "learning_rate": 7.737334398521149e-06,
      "loss": 0.0693,
      "step": 421
    },
    {
      "epoch": 0.6326836581709145,
      "grad_norm": 1.0227653644881336,
      "learning_rate": 7.727473103747456e-06,
      "loss": 0.1285,
      "step": 422
    },
    {
      "epoch": 0.6341829085457271,
      "grad_norm": 0.7218820740664142,
      "learning_rate": 7.717596682128578e-06,
      "loss": 0.071,
      "step": 423
    },
    {
      "epoch": 0.6356821589205397,
      "grad_norm": 0.7415913204624978,
      "learning_rate": 7.707705188440165e-06,
      "loss": 0.0691,
      "step": 424
    },
    {
      "epoch": 0.6371814092953523,
      "grad_norm": 0.9746879719295184,
      "learning_rate": 7.697798677541448e-06,
      "loss": 0.1157,
      "step": 425
    },
    {
      "epoch": 0.638680659670165,
      "grad_norm": 0.7871359396960251,
      "learning_rate": 7.687877204374957e-06,
      "loss": 0.0876,
      "step": 426
    },
    {
      "epoch": 0.6401799100449775,
      "grad_norm": 0.7214771696181748,
      "learning_rate": 7.677940823966196e-06,
      "loss": 0.0735,
      "step": 427
    },
    {
      "epoch": 0.6416791604197901,
      "grad_norm": 0.8631700064188158,
      "learning_rate": 7.667989591423349e-06,
      "loss": 0.0843,
      "step": 428
    },
    {
      "epoch": 0.6431784107946027,
      "grad_norm": 0.6491024837241722,
      "learning_rate": 7.658023561936966e-06,
      "loss": 0.0735,
      "step": 429
    },
    {
      "epoch": 0.6446776611694153,
      "grad_norm": 0.704669499304359,
      "learning_rate": 7.648042790779677e-06,
      "loss": 0.0776,
      "step": 430
    },
    {
      "epoch": 0.6461769115442278,
      "grad_norm": 0.7282982211259483,
      "learning_rate": 7.638047333305853e-06,
      "loss": 0.0713,
      "step": 431
    },
    {
      "epoch": 0.6476761619190404,
      "grad_norm": 0.7518108889467108,
      "learning_rate": 7.628037244951328e-06,
      "loss": 0.0848,
      "step": 432
    },
    {
      "epoch": 0.6491754122938531,
      "grad_norm": 0.7660191608260162,
      "learning_rate": 7.618012581233076e-06,
      "loss": 0.0689,
      "step": 433
    },
    {
      "epoch": 0.6506746626686657,
      "grad_norm": 0.6244470935743375,
      "learning_rate": 7.607973397748909e-06,
      "loss": 0.0673,
      "step": 434
    },
    {
      "epoch": 0.6521739130434783,
      "grad_norm": 0.775469802599003,
      "learning_rate": 7.597919750177168e-06,
      "loss": 0.0884,
      "step": 435
    },
    {
      "epoch": 0.6536731634182908,
      "grad_norm": 0.7147760971688474,
      "learning_rate": 7.587851694276412e-06,
      "loss": 0.0685,
      "step": 436
    },
    {
      "epoch": 0.6551724137931034,
      "grad_norm": 0.7107685952836847,
      "learning_rate": 7.57776928588511e-06,
      "loss": 0.0649,
      "step": 437
    },
    {
      "epoch": 0.656671664167916,
      "grad_norm": 0.6073628432091066,
      "learning_rate": 7.56767258092133e-06,
      "loss": 0.0577,
      "step": 438
    },
    {
      "epoch": 0.6581709145427287,
      "grad_norm": 0.7819403967227397,
      "learning_rate": 7.557561635382433e-06,
      "loss": 0.0754,
      "step": 439
    },
    {
      "epoch": 0.6596701649175413,
      "grad_norm": 0.6998307935118903,
      "learning_rate": 7.54743650534476e-06,
      "loss": 0.0664,
      "step": 440
    },
    {
      "epoch": 0.6611694152923538,
      "grad_norm": 0.9091602425919721,
      "learning_rate": 7.537297246963316e-06,
      "loss": 0.1001,
      "step": 441
    },
    {
      "epoch": 0.6626686656671664,
      "grad_norm": 0.7114245305068242,
      "learning_rate": 7.5271439164714695e-06,
      "loss": 0.0728,
      "step": 442
    },
    {
      "epoch": 0.664167916041979,
      "grad_norm": 0.6357113859680864,
      "learning_rate": 7.5169765701806295e-06,
      "loss": 0.0634,
      "step": 443
    },
    {
      "epoch": 0.6656671664167916,
      "grad_norm": 0.5005147278030838,
      "learning_rate": 7.506795264479941e-06,
      "loss": 0.0541,
      "step": 444
    },
    {
      "epoch": 0.6671664167916042,
      "grad_norm": 0.7282982314379133,
      "learning_rate": 7.4966000558359675e-06,
      "loss": 0.0769,
      "step": 445
    },
    {
      "epoch": 0.6686656671664168,
      "grad_norm": 0.9286303281271938,
      "learning_rate": 7.486391000792379e-06,
      "loss": 0.0955,
      "step": 446
    },
    {
      "epoch": 0.6701649175412294,
      "grad_norm": 0.7927245374776729,
      "learning_rate": 7.476168155969643e-06,
      "loss": 0.0912,
      "step": 447
    },
    {
      "epoch": 0.671664167916042,
      "grad_norm": 0.7977387698915536,
      "learning_rate": 7.465931578064703e-06,
      "loss": 0.0835,
      "step": 448
    },
    {
      "epoch": 0.6731634182908546,
      "grad_norm": 0.7877848634537132,
      "learning_rate": 7.455681323850669e-06,
      "loss": 0.0768,
      "step": 449
    },
    {
      "epoch": 0.6746626686656672,
      "grad_norm": 0.5406928510330875,
      "learning_rate": 7.4454174501765e-06,
      "loss": 0.0522,
      "step": 450
    },
    {
      "epoch": 0.6761619190404797,
      "grad_norm": 0.7114225911370212,
      "learning_rate": 7.4351400139666894e-06,
      "loss": 0.0748,
      "step": 451
    },
    {
      "epoch": 0.6776611694152923,
      "grad_norm": 0.6677538537383216,
      "learning_rate": 7.424849072220953e-06,
      "loss": 0.0844,
      "step": 452
    },
    {
      "epoch": 0.679160419790105,
      "grad_norm": 0.6841641643025314,
      "learning_rate": 7.414544682013907e-06,
      "loss": 0.0602,
      "step": 453
    },
    {
      "epoch": 0.6806596701649176,
      "grad_norm": 0.7044203809265486,
      "learning_rate": 7.404226900494753e-06,
      "loss": 0.062,
      "step": 454
    },
    {
      "epoch": 0.6821589205397302,
      "grad_norm": 0.5793715909367938,
      "learning_rate": 7.3938957848869684e-06,
      "loss": 0.063,
      "step": 455
    },
    {
      "epoch": 0.6836581709145427,
      "grad_norm": 0.6150069280842929,
      "learning_rate": 7.3835513924879755e-06,
      "loss": 0.0521,
      "step": 456
    },
    {
      "epoch": 0.6851574212893553,
      "grad_norm": 0.6001111283385462,
      "learning_rate": 7.373193780668835e-06,
      "loss": 0.0611,
      "step": 457
    },
    {
      "epoch": 0.6866566716641679,
      "grad_norm": 0.9036394737032505,
      "learning_rate": 7.36282300687392e-06,
      "loss": 0.0941,
      "step": 458
    },
    {
      "epoch": 0.6881559220389805,
      "grad_norm": 0.7706254578045112,
      "learning_rate": 7.35243912862061e-06,
      "loss": 0.0733,
      "step": 459
    },
    {
      "epoch": 0.6896551724137931,
      "grad_norm": 0.8655297599237697,
      "learning_rate": 7.342042203498952e-06,
      "loss": 0.0809,
      "step": 460
    },
    {
      "epoch": 0.6911544227886057,
      "grad_norm": 0.9845040341400938,
      "learning_rate": 7.33163228917136e-06,
      "loss": 0.088,
      "step": 461
    },
    {
      "epoch": 0.6926536731634183,
      "grad_norm": 0.6964106756019035,
      "learning_rate": 7.321209443372284e-06,
      "loss": 0.0792,
      "step": 462
    },
    {
      "epoch": 0.6941529235382309,
      "grad_norm": 0.7441716717863236,
      "learning_rate": 7.310773723907895e-06,
      "loss": 0.0854,
      "step": 463
    },
    {
      "epoch": 0.6956521739130435,
      "grad_norm": 0.9467679900120367,
      "learning_rate": 7.300325188655762e-06,
      "loss": 0.0871,
      "step": 464
    },
    {
      "epoch": 0.697151424287856,
      "grad_norm": 0.8132338619814425,
      "learning_rate": 7.289863895564531e-06,
      "loss": 0.0927,
      "step": 465
    },
    {
      "epoch": 0.6986506746626686,
      "grad_norm": 0.6859990311452611,
      "learning_rate": 7.279389902653606e-06,
      "loss": 0.0778,
      "step": 466
    },
    {
      "epoch": 0.7001499250374813,
      "grad_norm": 0.5741245108042371,
      "learning_rate": 7.268903268012823e-06,
      "loss": 0.0642,
      "step": 467
    },
    {
      "epoch": 0.7016491754122939,
      "grad_norm": 0.6522044205577349,
      "learning_rate": 7.258404049802135e-06,
      "loss": 0.0682,
      "step": 468
    },
    {
      "epoch": 0.7031484257871065,
      "grad_norm": 0.5492475403135738,
      "learning_rate": 7.247892306251276e-06,
      "loss": 0.0495,
      "step": 469
    },
    {
      "epoch": 0.704647676161919,
      "grad_norm": 0.6499797471666889,
      "learning_rate": 7.237368095659459e-06,
      "loss": 0.0776,
      "step": 470
    },
    {
      "epoch": 0.7061469265367316,
      "grad_norm": 0.7344786126341518,
      "learning_rate": 7.226831476395028e-06,
      "loss": 0.072,
      "step": 471
    },
    {
      "epoch": 0.7076461769115442,
      "grad_norm": 0.7502444553729471,
      "learning_rate": 7.216282506895155e-06,
      "loss": 0.0687,
      "step": 472
    },
    {
      "epoch": 0.7091454272863568,
      "grad_norm": 0.7399787207806131,
      "learning_rate": 7.2057212456655055e-06,
      "loss": 0.0762,
      "step": 473
    },
    {
      "epoch": 0.7106446776611695,
      "grad_norm": 0.7781468042629904,
      "learning_rate": 7.195147751279915e-06,
      "loss": 0.0914,
      "step": 474
    },
    {
      "epoch": 0.712143928035982,
      "grad_norm": 0.9657179889685302,
      "learning_rate": 7.184562082380069e-06,
      "loss": 0.1177,
      "step": 475
    },
    {
      "epoch": 0.7136431784107946,
      "grad_norm": 0.8462806933944134,
      "learning_rate": 7.173964297675168e-06,
      "loss": 0.0878,
      "step": 476
    },
    {
      "epoch": 0.7151424287856072,
      "grad_norm": 0.9108780697563349,
      "learning_rate": 7.163354455941614e-06,
      "loss": 0.0897,
      "step": 477
    },
    {
      "epoch": 0.7166416791604198,
      "grad_norm": 0.737963173268101,
      "learning_rate": 7.152732616022675e-06,
      "loss": 0.0797,
      "step": 478
    },
    {
      "epoch": 0.7181409295352323,
      "grad_norm": 0.7853555043173104,
      "learning_rate": 7.142098836828162e-06,
      "loss": 0.0937,
      "step": 479
    },
    {
      "epoch": 0.719640179910045,
      "grad_norm": 0.9751022810950013,
      "learning_rate": 7.131453177334103e-06,
      "loss": 0.095,
      "step": 480
    },
    {
      "epoch": 0.7211394302848576,
      "grad_norm": 0.6127744434464046,
      "learning_rate": 7.120795696582419e-06,
      "loss": 0.064,
      "step": 481
    },
    {
      "epoch": 0.7226386806596702,
      "grad_norm": 0.8129995742022367,
      "learning_rate": 7.1101264536805885e-06,
      "loss": 0.085,
      "step": 482
    },
    {
      "epoch": 0.7241379310344828,
      "grad_norm": 0.6574983412452825,
      "learning_rate": 7.099445507801324e-06,
      "loss": 0.0603,
      "step": 483
    },
    {
      "epoch": 0.7256371814092953,
      "grad_norm": 0.7490159595866301,
      "learning_rate": 7.088752918182247e-06,
      "loss": 0.0717,
      "step": 484
    },
    {
      "epoch": 0.7271364317841079,
      "grad_norm": 0.9368502043717238,
      "learning_rate": 7.078048744125553e-06,
      "loss": 0.1118,
      "step": 485
    },
    {
      "epoch": 0.7286356821589205,
      "grad_norm": 0.7393632484665587,
      "learning_rate": 7.067333044997689e-06,
      "loss": 0.0887,
      "step": 486
    },
    {
      "epoch": 0.7301349325337332,
      "grad_norm": 1.0245914456450402,
      "learning_rate": 7.0566058802290196e-06,
      "loss": 0.1197,
      "step": 487
    },
    {
      "epoch": 0.7316341829085458,
      "grad_norm": 0.6465615347587207,
      "learning_rate": 7.045867309313499e-06,
      "loss": 0.0744,
      "step": 488
    },
    {
      "epoch": 0.7331334332833583,
      "grad_norm": 0.6703388376648135,
      "learning_rate": 7.035117391808341e-06,
      "loss": 0.0722,
      "step": 489
    },
    {
      "epoch": 0.7346326836581709,
      "grad_norm": 0.7204740609904797,
      "learning_rate": 7.024356187333692e-06,
      "loss": 0.0783,
      "step": 490
    },
    {
      "epoch": 0.7361319340329835,
      "grad_norm": 0.9323351198611449,
      "learning_rate": 7.01358375557229e-06,
      "loss": 0.1185,
      "step": 491
    },
    {
      "epoch": 0.7376311844077961,
      "grad_norm": 0.5906510039899294,
      "learning_rate": 7.0028001562691475e-06,
      "loss": 0.0655,
      "step": 492
    },
    {
      "epoch": 0.7391304347826086,
      "grad_norm": 0.6985561366078488,
      "learning_rate": 6.9920054492312086e-06,
      "loss": 0.0854,
      "step": 493
    },
    {
      "epoch": 0.7406296851574213,
      "grad_norm": 0.8839287283231729,
      "learning_rate": 6.981199694327024e-06,
      "loss": 0.0789,
      "step": 494
    },
    {
      "epoch": 0.7421289355322339,
      "grad_norm": 0.6717191078333004,
      "learning_rate": 6.97038295148642e-06,
      "loss": 0.0718,
      "step": 495
    },
    {
      "epoch": 0.7436281859070465,
      "grad_norm": 0.5574788738776045,
      "learning_rate": 6.959555280700162e-06,
      "loss": 0.0584,
      "step": 496
    },
    {
      "epoch": 0.7451274362818591,
      "grad_norm": 0.8611984677748481,
      "learning_rate": 6.948716742019616e-06,
      "loss": 0.0825,
      "step": 497
    },
    {
      "epoch": 0.7466266866566716,
      "grad_norm": 1.0098999647859446,
      "learning_rate": 6.937867395556428e-06,
      "loss": 0.0953,
      "step": 498
    },
    {
      "epoch": 0.7481259370314842,
      "grad_norm": 0.5043623582669988,
      "learning_rate": 6.927007301482187e-06,
      "loss": 0.0456,
      "step": 499
    },
    {
      "epoch": 0.7496251874062968,
      "grad_norm": 0.9542175517326643,
      "learning_rate": 6.916136520028087e-06,
      "loss": 0.1108,
      "step": 500
    },
    {
      "epoch": 0.7511244377811095,
      "grad_norm": 0.750898346926149,
      "learning_rate": 6.905255111484592e-06,
      "loss": 0.0767,
      "step": 501
    },
    {
      "epoch": 0.7526236881559221,
      "grad_norm": 0.8845678444933264,
      "learning_rate": 6.894363136201114e-06,
      "loss": 0.0973,
      "step": 502
    },
    {
      "epoch": 0.7541229385307346,
      "grad_norm": 0.7563252493770662,
      "learning_rate": 6.88346065458566e-06,
      "loss": 0.0926,
      "step": 503
    },
    {
      "epoch": 0.7556221889055472,
      "grad_norm": 0.8000719920488322,
      "learning_rate": 6.8725477271045085e-06,
      "loss": 0.082,
      "step": 504
    },
    {
      "epoch": 0.7571214392803598,
      "grad_norm": 0.770285410855956,
      "learning_rate": 6.861624414281875e-06,
      "loss": 0.0753,
      "step": 505
    },
    {
      "epoch": 0.7586206896551724,
      "grad_norm": 0.7169952279246674,
      "learning_rate": 6.850690776699574e-06,
      "loss": 0.0757,
      "step": 506
    },
    {
      "epoch": 0.760119940029985,
      "grad_norm": 0.9210942971752456,
      "learning_rate": 6.8397468749966735e-06,
      "loss": 0.0755,
      "step": 507
    },
    {
      "epoch": 0.7616191904047976,
      "grad_norm": 0.7293893019619603,
      "learning_rate": 6.8287927698691745e-06,
      "loss": 0.0727,
      "step": 508
    },
    {
      "epoch": 0.7631184407796102,
      "grad_norm": 0.731114614323025,
      "learning_rate": 6.8178285220696686e-06,
      "loss": 0.0881,
      "step": 509
    },
    {
      "epoch": 0.7646176911544228,
      "grad_norm": 0.5638844985017472,
      "learning_rate": 6.806854192406995e-06,
      "loss": 0.053,
      "step": 510
    },
    {
      "epoch": 0.7661169415292354,
      "grad_norm": 0.7867431341261079,
      "learning_rate": 6.795869841745912e-06,
      "loss": 0.0899,
      "step": 511
    },
    {
      "epoch": 0.767616191904048,
      "grad_norm": 0.6091909771572415,
      "learning_rate": 6.784875531006751e-06,
      "loss": 0.0585,
      "step": 512
    },
    {
      "epoch": 0.7691154422788605,
      "grad_norm": 1.0482333792071226,
      "learning_rate": 6.7738713211650885e-06,
      "loss": 0.1184,
      "step": 513
    },
    {
      "epoch": 0.7706146926536732,
      "grad_norm": 0.6427340293081968,
      "learning_rate": 6.762857273251396e-06,
      "loss": 0.067,
      "step": 514
    },
    {
      "epoch": 0.7721139430284858,
      "grad_norm": 0.8515793403739592,
      "learning_rate": 6.751833448350713e-06,
      "loss": 0.0807,
      "step": 515
    },
    {
      "epoch": 0.7736131934032984,
      "grad_norm": 1.0278000679048025,
      "learning_rate": 6.740799907602302e-06,
      "loss": 0.1011,
      "step": 516
    },
    {
      "epoch": 0.775112443778111,
      "grad_norm": 0.6297328693853753,
      "learning_rate": 6.729756712199309e-06,
      "loss": 0.058,
      "step": 517
    },
    {
      "epoch": 0.7766116941529235,
      "grad_norm": 0.5944775321482884,
      "learning_rate": 6.718703923388427e-06,
      "loss": 0.0592,
      "step": 518
    },
    {
      "epoch": 0.7781109445277361,
      "grad_norm": 0.7231384584592381,
      "learning_rate": 6.707641602469554e-06,
      "loss": 0.0685,
      "step": 519
    },
    {
      "epoch": 0.7796101949025487,
      "grad_norm": 0.8361819256037644,
      "learning_rate": 6.696569810795455e-06,
      "loss": 0.086,
      "step": 520
    },
    {
      "epoch": 0.7811094452773614,
      "grad_norm": 0.6068761218696832,
      "learning_rate": 6.685488609771422e-06,
      "loss": 0.0576,
      "step": 521
    },
    {
      "epoch": 0.782608695652174,
      "grad_norm": 0.7082214082637094,
      "learning_rate": 6.674398060854931e-06,
      "loss": 0.0698,
      "step": 522
    },
    {
      "epoch": 0.7841079460269865,
      "grad_norm": 0.7276574655995693,
      "learning_rate": 6.6632982255553004e-06,
      "loss": 0.0776,
      "step": 523
    },
    {
      "epoch": 0.7856071964017991,
      "grad_norm": 1.0486482679057727,
      "learning_rate": 6.652189165433356e-06,
      "loss": 0.0986,
      "step": 524
    },
    {
      "epoch": 0.7871064467766117,
      "grad_norm": 0.6703677182844063,
      "learning_rate": 6.64107094210108e-06,
      "loss": 0.077,
      "step": 525
    },
    {
      "epoch": 0.7886056971514243,
      "grad_norm": 0.7531943464037067,
      "learning_rate": 6.62994361722128e-06,
      "loss": 0.0698,
      "step": 526
    },
    {
      "epoch": 0.7901049475262368,
      "grad_norm": 0.7238557122724278,
      "learning_rate": 6.618807252507238e-06,
      "loss": 0.0697,
      "step": 527
    },
    {
      "epoch": 0.7916041979010495,
      "grad_norm": 1.0346167190432127,
      "learning_rate": 6.6076619097223735e-06,
      "loss": 0.1187,
      "step": 528
    },
    {
      "epoch": 0.7931034482758621,
      "grad_norm": 1.0577295588629734,
      "learning_rate": 6.5965076506799e-06,
      "loss": 0.1111,
      "step": 529
    },
    {
      "epoch": 0.7946026986506747,
      "grad_norm": 0.7843118463332387,
      "learning_rate": 6.5853445372424805e-06,
      "loss": 0.0868,
      "step": 530
    },
    {
      "epoch": 0.7961019490254873,
      "grad_norm": 0.6507017761305778,
      "learning_rate": 6.574172631321885e-06,
      "loss": 0.064,
      "step": 531
    },
    {
      "epoch": 0.7976011994002998,
      "grad_norm": 0.8052774438019565,
      "learning_rate": 6.562991994878649e-06,
      "loss": 0.0821,
      "step": 532
    },
    {
      "epoch": 0.7991004497751124,
      "grad_norm": 0.7867410653289352,
      "learning_rate": 6.551802689921726e-06,
      "loss": 0.0898,
      "step": 533
    },
    {
      "epoch": 0.800599700149925,
      "grad_norm": 0.7988308528842235,
      "learning_rate": 6.5406047785081485e-06,
      "loss": 0.0965,
      "step": 534
    },
    {
      "epoch": 0.8020989505247377,
      "grad_norm": 0.6220343655801845,
      "learning_rate": 6.529398322742677e-06,
      "loss": 0.0631,
      "step": 535
    },
    {
      "epoch": 0.8035982008995503,
      "grad_norm": 0.5823816888031392,
      "learning_rate": 6.518183384777468e-06,
      "loss": 0.0603,
      "step": 536
    },
    {
      "epoch": 0.8050974512743628,
      "grad_norm": 1.0044064379263997,
      "learning_rate": 6.506960026811712e-06,
      "loss": 0.1001,
      "step": 537
    },
    {
      "epoch": 0.8065967016491754,
      "grad_norm": 0.8347035674400577,
      "learning_rate": 6.495728311091303e-06,
      "loss": 0.1052,
      "step": 538
    },
    {
      "epoch": 0.808095952023988,
      "grad_norm": 0.698048032974808,
      "learning_rate": 6.484488299908487e-06,
      "loss": 0.0671,
      "step": 539
    },
    {
      "epoch": 0.8095952023988006,
      "grad_norm": 0.7400042491383872,
      "learning_rate": 6.473240055601517e-06,
      "loss": 0.0769,
      "step": 540
    },
    {
      "epoch": 0.8110944527736131,
      "grad_norm": 0.7434291769442241,
      "learning_rate": 6.46198364055431e-06,
      "loss": 0.0767,
      "step": 541
    },
    {
      "epoch": 0.8125937031484258,
      "grad_norm": 1.121470341672876,
      "learning_rate": 6.450719117196094e-06,
      "loss": 0.101,
      "step": 542
    },
    {
      "epoch": 0.8140929535232384,
      "grad_norm": 0.8937393171823191,
      "learning_rate": 6.439446548001069e-06,
      "loss": 0.0729,
      "step": 543
    },
    {
      "epoch": 0.815592203898051,
      "grad_norm": 1.0145195075201001,
      "learning_rate": 6.4281659954880605e-06,
      "loss": 0.0951,
      "step": 544
    },
    {
      "epoch": 0.8170914542728636,
      "grad_norm": 0.654307346580361,
      "learning_rate": 6.416877522220167e-06,
      "loss": 0.0629,
      "step": 545
    },
    {
      "epoch": 0.8185907046476761,
      "grad_norm": 0.8483803265271206,
      "learning_rate": 6.405581190804418e-06,
      "loss": 0.0729,
      "step": 546
    },
    {
      "epoch": 0.8200899550224887,
      "grad_norm": 0.5983900686870395,
      "learning_rate": 6.394277063891422e-06,
      "loss": 0.0601,
      "step": 547
    },
    {
      "epoch": 0.8215892053973014,
      "grad_norm": 0.8243416146149315,
      "learning_rate": 6.382965204175027e-06,
      "loss": 0.11,
      "step": 548
    },
    {
      "epoch": 0.823088455772114,
      "grad_norm": 0.6432359950470236,
      "learning_rate": 6.371645674391967e-06,
      "loss": 0.0655,
      "step": 549
    },
    {
      "epoch": 0.8245877061469266,
      "grad_norm": 0.9033901277556133,
      "learning_rate": 6.3603185373215105e-06,
      "loss": 0.0815,
      "step": 550
    },
    {
      "epoch": 0.8260869565217391,
      "grad_norm": 0.8418380660451029,
      "learning_rate": 6.348983855785122e-06,
      "loss": 0.1098,
      "step": 551
    },
    {
      "epoch": 0.8275862068965517,
      "grad_norm": 0.84117954791311,
      "learning_rate": 6.337641692646106e-06,
      "loss": 0.0884,
      "step": 552
    },
    {
      "epoch": 0.8290854572713643,
      "grad_norm": 0.572853211516459,
      "learning_rate": 6.326292110809258e-06,
      "loss": 0.0617,
      "step": 553
    },
    {
      "epoch": 0.8305847076461769,
      "grad_norm": 0.541348618504373,
      "learning_rate": 6.314935173220524e-06,
      "loss": 0.0634,
      "step": 554
    },
    {
      "epoch": 0.8320839580209896,
      "grad_norm": 0.9409173033659952,
      "learning_rate": 6.303570942866643e-06,
      "loss": 0.1213,
      "step": 555
    },
    {
      "epoch": 0.8335832083958021,
      "grad_norm": 0.7189799611610367,
      "learning_rate": 6.2921994827748e-06,
      "loss": 0.0735,
      "step": 556
    },
    {
      "epoch": 0.8350824587706147,
      "grad_norm": 0.8179884642234975,
      "learning_rate": 6.280820856012277e-06,
      "loss": 0.0707,
      "step": 557
    },
    {
      "epoch": 0.8365817091454273,
      "grad_norm": 0.5359174499716282,
      "learning_rate": 6.269435125686105e-06,
      "loss": 0.0531,
      "step": 558
    },
    {
      "epoch": 0.8380809595202399,
      "grad_norm": 0.7712647972998709,
      "learning_rate": 6.258042354942708e-06,
      "loss": 0.0802,
      "step": 559
    },
    {
      "epoch": 0.8395802098950524,
      "grad_norm": 0.6689914898423012,
      "learning_rate": 6.2466426069675626e-06,
      "loss": 0.0689,
      "step": 560
    },
    {
      "epoch": 0.841079460269865,
      "grad_norm": 0.6945642851444355,
      "learning_rate": 6.235235944984835e-06,
      "loss": 0.08,
      "step": 561
    },
    {
      "epoch": 0.8425787106446777,
      "grad_norm": 0.7181392901268854,
      "learning_rate": 6.223822432257043e-06,
      "loss": 0.0652,
      "step": 562
    },
    {
      "epoch": 0.8440779610194903,
      "grad_norm": 0.915671684307425,
      "learning_rate": 6.212402132084697e-06,
      "loss": 0.0968,
      "step": 563
    },
    {
      "epoch": 0.8455772113943029,
      "grad_norm": 0.7913771089789317,
      "learning_rate": 6.200975107805951e-06,
      "loss": 0.0669,
      "step": 564
    },
    {
      "epoch": 0.8470764617691154,
      "grad_norm": 0.7314238627193941,
      "learning_rate": 6.189541422796254e-06,
      "loss": 0.0751,
      "step": 565
    },
    {
      "epoch": 0.848575712143928,
      "grad_norm": 0.6445313810434284,
      "learning_rate": 6.1781011404679905e-06,
      "loss": 0.0724,
      "step": 566
    },
    {
      "epoch": 0.8500749625187406,
      "grad_norm": 0.6728632223116919,
      "learning_rate": 6.16665432427014e-06,
      "loss": 0.0642,
      "step": 567
    },
    {
      "epoch": 0.8515742128935532,
      "grad_norm": 0.7646871322608257,
      "learning_rate": 6.155201037687917e-06,
      "loss": 0.1063,
      "step": 568
    },
    {
      "epoch": 0.8530734632683659,
      "grad_norm": 0.7083163206811307,
      "learning_rate": 6.1437413442424236e-06,
      "loss": 0.0648,
      "step": 569
    },
    {
      "epoch": 0.8545727136431784,
      "grad_norm": 0.6760757464996109,
      "learning_rate": 6.132275307490291e-06,
      "loss": 0.0705,
      "step": 570
    },
    {
      "epoch": 0.856071964017991,
      "grad_norm": 0.7273413483317993,
      "learning_rate": 6.120802991023334e-06,
      "loss": 0.0601,
      "step": 571
    },
    {
      "epoch": 0.8575712143928036,
      "grad_norm": 0.7358186945584324,
      "learning_rate": 6.109324458468198e-06,
      "loss": 0.0845,
      "step": 572
    },
    {
      "epoch": 0.8590704647676162,
      "grad_norm": 0.7460173945689503,
      "learning_rate": 6.097839773485995e-06,
      "loss": 0.0705,
      "step": 573
    },
    {
      "epoch": 0.8605697151424287,
      "grad_norm": 0.8588428151633754,
      "learning_rate": 6.086348999771967e-06,
      "loss": 0.0952,
      "step": 574
    },
    {
      "epoch": 0.8620689655172413,
      "grad_norm": 0.7004417023222904,
      "learning_rate": 6.074852201055121e-06,
      "loss": 0.0688,
      "step": 575
    },
    {
      "epoch": 0.863568215892054,
      "grad_norm": 0.7229990360380887,
      "learning_rate": 6.063349441097881e-06,
      "loss": 0.0736,
      "step": 576
    },
    {
      "epoch": 0.8650674662668666,
      "grad_norm": 0.755864470130181,
      "learning_rate": 6.051840783695731e-06,
      "loss": 0.0712,
      "step": 577
    },
    {
      "epoch": 0.8665667166416792,
      "grad_norm": 0.7711462534269377,
      "learning_rate": 6.040326292676865e-06,
      "loss": 0.0781,
      "step": 578
    },
    {
      "epoch": 0.8680659670164917,
      "grad_norm": 1.1979282371105484,
      "learning_rate": 6.028806031901829e-06,
      "loss": 0.0972,
      "step": 579
    },
    {
      "epoch": 0.8695652173913043,
      "grad_norm": 0.740276078676565,
      "learning_rate": 6.0172800652631706e-06,
      "loss": 0.0866,
      "step": 580
    },
    {
      "epoch": 0.8710644677661169,
      "grad_norm": 0.7630702982803514,
      "learning_rate": 6.005748456685077e-06,
      "loss": 0.0655,
      "step": 581
    },
    {
      "epoch": 0.8725637181409296,
      "grad_norm": 0.8662252887360015,
      "learning_rate": 5.994211270123034e-06,
      "loss": 0.0971,
      "step": 582
    },
    {
      "epoch": 0.8740629685157422,
      "grad_norm": 0.8556182360244335,
      "learning_rate": 5.9826685695634575e-06,
      "loss": 0.092,
      "step": 583
    },
    {
      "epoch": 0.8755622188905547,
      "grad_norm": 0.716276324301073,
      "learning_rate": 5.971120419023349e-06,
      "loss": 0.063,
      "step": 584
    },
    {
      "epoch": 0.8770614692653673,
      "grad_norm": 0.5868804253721288,
      "learning_rate": 5.959566882549936e-06,
      "loss": 0.0844,
      "step": 585
    },
    {
      "epoch": 0.8785607196401799,
      "grad_norm": 0.7384381708885581,
      "learning_rate": 5.948008024220311e-06,
      "loss": 0.0774,
      "step": 586
    },
    {
      "epoch": 0.8800599700149925,
      "grad_norm": 0.6395213342632895,
      "learning_rate": 5.936443908141088e-06,
      "loss": 0.063,
      "step": 587
    },
    {
      "epoch": 0.881559220389805,
      "grad_norm": 0.6901179125370542,
      "learning_rate": 5.924874598448038e-06,
      "loss": 0.0715,
      "step": 588
    },
    {
      "epoch": 0.8830584707646177,
      "grad_norm": 0.6119301494676177,
      "learning_rate": 5.913300159305741e-06,
      "loss": 0.0721,
      "step": 589
    },
    {
      "epoch": 0.8845577211394303,
      "grad_norm": 0.8472740001007733,
      "learning_rate": 5.901720654907217e-06,
      "loss": 0.0653,
      "step": 590
    },
    {
      "epoch": 0.8860569715142429,
      "grad_norm": 0.6822104860117685,
      "learning_rate": 5.8901361494735874e-06,
      "loss": 0.0722,
      "step": 591
    },
    {
      "epoch": 0.8875562218890555,
      "grad_norm": 0.7628250214068681,
      "learning_rate": 5.878546707253704e-06,
      "loss": 0.0794,
      "step": 592
    },
    {
      "epoch": 0.889055472263868,
      "grad_norm": 0.8466766906927604,
      "learning_rate": 5.8669523925238e-06,
      "loss": 0.0917,
      "step": 593
    },
    {
      "epoch": 0.8905547226386806,
      "grad_norm": 0.6491133919807952,
      "learning_rate": 5.855353269587134e-06,
      "loss": 0.0616,
      "step": 594
    },
    {
      "epoch": 0.8920539730134932,
      "grad_norm": 0.6281090020361754,
      "learning_rate": 5.843749402773629e-06,
      "loss": 0.062,
      "step": 595
    },
    {
      "epoch": 0.8935532233883059,
      "grad_norm": 0.686565177242158,
      "learning_rate": 5.8321408564395165e-06,
      "loss": 0.0598,
      "step": 596
    },
    {
      "epoch": 0.8950524737631185,
      "grad_norm": 0.7544759252592572,
      "learning_rate": 5.820527694966988e-06,
      "loss": 0.0721,
      "step": 597
    },
    {
      "epoch": 0.896551724137931,
      "grad_norm": 0.5533489221400043,
      "learning_rate": 5.808909982763825e-06,
      "loss": 0.0485,
      "step": 598
    },
    {
      "epoch": 0.8980509745127436,
      "grad_norm": 0.6071100335349233,
      "learning_rate": 5.797287784263047e-06,
      "loss": 0.0777,
      "step": 599
    },
    {
      "epoch": 0.8995502248875562,
      "grad_norm": 0.8133418151949829,
      "learning_rate": 5.785661163922558e-06,
      "loss": 0.0752,
      "step": 600
    },
    {
      "epoch": 0.8995502248875562,
      "eval_loss": 0.08782054483890533,
      "eval_runtime": 2.1925,
      "eval_samples_per_second": 24.63,
      "eval_steps_per_second": 6.386,
      "step": 600
    },
    {
      "epoch": 0.9010494752623688,
      "grad_norm": 0.8011942672109775,
      "learning_rate": 5.774030186224786e-06,
      "loss": 0.0747,
      "step": 601
    },
    {
      "epoch": 0.9025487256371814,
      "grad_norm": 0.5615213889587152,
      "learning_rate": 5.762394915676325e-06,
      "loss": 0.0525,
      "step": 602
    },
    {
      "epoch": 0.904047976011994,
      "grad_norm": 0.8656439666695872,
      "learning_rate": 5.750755416807575e-06,
      "loss": 0.1025,
      "step": 603
    },
    {
      "epoch": 0.9055472263868066,
      "grad_norm": 0.7125643487148012,
      "learning_rate": 5.7391117541723914e-06,
      "loss": 0.0699,
      "step": 604
    },
    {
      "epoch": 0.9070464767616192,
      "grad_norm": 0.8255902191044361,
      "learning_rate": 5.727463992347719e-06,
      "loss": 0.0823,
      "step": 605
    },
    {
      "epoch": 0.9085457271364318,
      "grad_norm": 0.7596448001463955,
      "learning_rate": 5.715812195933238e-06,
      "loss": 0.0686,
      "step": 606
    },
    {
      "epoch": 0.9100449775112444,
      "grad_norm": 0.7754770985801284,
      "learning_rate": 5.704156429551004e-06,
      "loss": 0.079,
      "step": 607
    },
    {
      "epoch": 0.9115442278860569,
      "grad_norm": 0.8013933118302163,
      "learning_rate": 5.692496757845092e-06,
      "loss": 0.0783,
      "step": 608
    },
    {
      "epoch": 0.9130434782608695,
      "grad_norm": 0.781970431977861,
      "learning_rate": 5.680833245481234e-06,
      "loss": 0.0776,
      "step": 609
    },
    {
      "epoch": 0.9145427286356822,
      "grad_norm": 0.97418597922222,
      "learning_rate": 5.6691659571464655e-06,
      "loss": 0.1237,
      "step": 610
    },
    {
      "epoch": 0.9160419790104948,
      "grad_norm": 0.6055781760021098,
      "learning_rate": 5.657494957548761e-06,
      "loss": 0.0696,
      "step": 611
    },
    {
      "epoch": 0.9175412293853074,
      "grad_norm": 0.6850588977343051,
      "learning_rate": 5.645820311416681e-06,
      "loss": 0.0584,
      "step": 612
    },
    {
      "epoch": 0.9190404797601199,
      "grad_norm": 0.9061867860747096,
      "learning_rate": 5.63414208349901e-06,
      "loss": 0.0948,
      "step": 613
    },
    {
      "epoch": 0.9205397301349325,
      "grad_norm": 0.7036414089775104,
      "learning_rate": 5.622460338564393e-06,
      "loss": 0.0763,
      "step": 614
    },
    {
      "epoch": 0.9220389805097451,
      "grad_norm": 0.6463181433540709,
      "learning_rate": 5.610775141400986e-06,
      "loss": 0.0577,
      "step": 615
    },
    {
      "epoch": 0.9235382308845578,
      "grad_norm": 0.7604148691666996,
      "learning_rate": 5.599086556816089e-06,
      "loss": 0.0813,
      "step": 616
    },
    {
      "epoch": 0.9250374812593704,
      "grad_norm": 0.8025299924321806,
      "learning_rate": 5.587394649635789e-06,
      "loss": 0.0953,
      "step": 617
    },
    {
      "epoch": 0.9265367316341829,
      "grad_norm": 0.743367420644983,
      "learning_rate": 5.575699484704599e-06,
      "loss": 0.0619,
      "step": 618
    },
    {
      "epoch": 0.9280359820089955,
      "grad_norm": 0.8618930460810412,
      "learning_rate": 5.564001126885106e-06,
      "loss": 0.086,
      "step": 619
    },
    {
      "epoch": 0.9295352323838081,
      "grad_norm": 0.8238842736732995,
      "learning_rate": 5.552299641057596e-06,
      "loss": 0.0897,
      "step": 620
    },
    {
      "epoch": 0.9310344827586207,
      "grad_norm": 0.7904042295559179,
      "learning_rate": 5.540595092119709e-06,
      "loss": 0.0796,
      "step": 621
    },
    {
      "epoch": 0.9325337331334332,
      "grad_norm": 0.7828668274241736,
      "learning_rate": 5.5288875449860745e-06,
      "loss": 0.0767,
      "step": 622
    },
    {
      "epoch": 0.9340329835082459,
      "grad_norm": 0.6091135699364111,
      "learning_rate": 5.517177064587945e-06,
      "loss": 0.0665,
      "step": 623
    },
    {
      "epoch": 0.9355322338830585,
      "grad_norm": 0.5819614211939534,
      "learning_rate": 5.505463715872846e-06,
      "loss": 0.0624,
      "step": 624
    },
    {
      "epoch": 0.9370314842578711,
      "grad_norm": 0.7444506896989295,
      "learning_rate": 5.493747563804211e-06,
      "loss": 0.0726,
      "step": 625
    },
    {
      "epoch": 0.9385307346326837,
      "grad_norm": 0.5117255189737073,
      "learning_rate": 5.482028673361015e-06,
      "loss": 0.0546,
      "step": 626
    },
    {
      "epoch": 0.9400299850074962,
      "grad_norm": 0.7792160565480374,
      "learning_rate": 5.470307109537427e-06,
      "loss": 0.0783,
      "step": 627
    },
    {
      "epoch": 0.9415292353823088,
      "grad_norm": 0.7358282722545803,
      "learning_rate": 5.45858293734244e-06,
      "loss": 0.0705,
      "step": 628
    },
    {
      "epoch": 0.9430284857571214,
      "grad_norm": 1.1001280948506074,
      "learning_rate": 5.446856221799515e-06,
      "loss": 0.0929,
      "step": 629
    },
    {
      "epoch": 0.9445277361319341,
      "grad_norm": 0.7691707275553036,
      "learning_rate": 5.435127027946215e-06,
      "loss": 0.0879,
      "step": 630
    },
    {
      "epoch": 0.9460269865067467,
      "grad_norm": 0.9791383536976564,
      "learning_rate": 5.423395420833853e-06,
      "loss": 0.0917,
      "step": 631
    },
    {
      "epoch": 0.9475262368815592,
      "grad_norm": 0.7873422126893124,
      "learning_rate": 5.411661465527123e-06,
      "loss": 0.0776,
      "step": 632
    },
    {
      "epoch": 0.9490254872563718,
      "grad_norm": 0.7106115517930557,
      "learning_rate": 5.39992522710374e-06,
      "loss": 0.0743,
      "step": 633
    },
    {
      "epoch": 0.9505247376311844,
      "grad_norm": 0.6497484636944706,
      "learning_rate": 5.38818677065409e-06,
      "loss": 0.0539,
      "step": 634
    },
    {
      "epoch": 0.952023988005997,
      "grad_norm": 0.7532480547016503,
      "learning_rate": 5.376446161280851e-06,
      "loss": 0.0722,
      "step": 635
    },
    {
      "epoch": 0.9535232383808095,
      "grad_norm": 0.8686532838292925,
      "learning_rate": 5.364703464098645e-06,
      "loss": 0.1011,
      "step": 636
    },
    {
      "epoch": 0.9550224887556222,
      "grad_norm": 0.7809090361201195,
      "learning_rate": 5.352958744233673e-06,
      "loss": 0.0707,
      "step": 637
    },
    {
      "epoch": 0.9565217391304348,
      "grad_norm": 0.7196793045078094,
      "learning_rate": 5.341212066823356e-06,
      "loss": 0.0694,
      "step": 638
    },
    {
      "epoch": 0.9580209895052474,
      "grad_norm": 0.680817389754364,
      "learning_rate": 5.329463497015969e-06,
      "loss": 0.0727,
      "step": 639
    },
    {
      "epoch": 0.95952023988006,
      "grad_norm": 0.6522445442811737,
      "learning_rate": 5.317713099970283e-06,
      "loss": 0.0624,
      "step": 640
    },
    {
      "epoch": 0.9610194902548725,
      "grad_norm": 0.7859780679579011,
      "learning_rate": 5.305960940855205e-06,
      "loss": 0.0933,
      "step": 641
    },
    {
      "epoch": 0.9625187406296851,
      "grad_norm": 0.8402232504591317,
      "learning_rate": 5.294207084849412e-06,
      "loss": 0.0818,
      "step": 642
    },
    {
      "epoch": 0.9640179910044977,
      "grad_norm": 0.5721239154590833,
      "learning_rate": 5.282451597140994e-06,
      "loss": 0.0573,
      "step": 643
    },
    {
      "epoch": 0.9655172413793104,
      "grad_norm": 0.8871700535529499,
      "learning_rate": 5.270694542927089e-06,
      "loss": 0.0919,
      "step": 644
    },
    {
      "epoch": 0.967016491754123,
      "grad_norm": 1.1016922062633026,
      "learning_rate": 5.258935987413524e-06,
      "loss": 0.1057,
      "step": 645
    },
    {
      "epoch": 0.9685157421289355,
      "grad_norm": 0.7277234831047426,
      "learning_rate": 5.247175995814452e-06,
      "loss": 0.0718,
      "step": 646
    },
    {
      "epoch": 0.9700149925037481,
      "grad_norm": 1.00110988152861,
      "learning_rate": 5.235414633351992e-06,
      "loss": 0.0783,
      "step": 647
    },
    {
      "epoch": 0.9715142428785607,
      "grad_norm": 0.8201257208338649,
      "learning_rate": 5.223651965255864e-06,
      "loss": 0.1007,
      "step": 648
    },
    {
      "epoch": 0.9730134932533733,
      "grad_norm": 0.6230657639693181,
      "learning_rate": 5.211888056763029e-06,
      "loss": 0.0556,
      "step": 649
    },
    {
      "epoch": 0.974512743628186,
      "grad_norm": 0.8226644389644732,
      "learning_rate": 5.20012297311733e-06,
      "loss": 0.087,
      "step": 650
    },
    {
      "epoch": 0.9760119940029985,
      "grad_norm": 0.5770766584407886,
      "learning_rate": 5.188356779569125e-06,
      "loss": 0.0676,
      "step": 651
    },
    {
      "epoch": 0.9775112443778111,
      "grad_norm": 0.8516236487457829,
      "learning_rate": 5.176589541374929e-06,
      "loss": 0.0797,
      "step": 652
    },
    {
      "epoch": 0.9790104947526237,
      "grad_norm": 0.974285679985352,
      "learning_rate": 5.164821323797051e-06,
      "loss": 0.0841,
      "step": 653
    },
    {
      "epoch": 0.9805097451274363,
      "grad_norm": 0.8244399613839225,
      "learning_rate": 5.1530521921032305e-06,
      "loss": 0.0724,
      "step": 654
    },
    {
      "epoch": 0.9820089955022488,
      "grad_norm": 0.7637946595176912,
      "learning_rate": 5.141282211566276e-06,
      "loss": 0.0836,
      "step": 655
    },
    {
      "epoch": 0.9835082458770614,
      "grad_norm": 0.7082765281588762,
      "learning_rate": 5.129511447463705e-06,
      "loss": 0.0604,
      "step": 656
    },
    {
      "epoch": 0.9850074962518741,
      "grad_norm": 0.9195461431302344,
      "learning_rate": 5.117739965077382e-06,
      "loss": 0.0936,
      "step": 657
    },
    {
      "epoch": 0.9865067466266867,
      "grad_norm": 0.6162082174726452,
      "learning_rate": 5.105967829693155e-06,
      "loss": 0.0587,
      "step": 658
    },
    {
      "epoch": 0.9880059970014993,
      "grad_norm": 0.80315420718906,
      "learning_rate": 5.0941951066004906e-06,
      "loss": 0.0753,
      "step": 659
    },
    {
      "epoch": 0.9895052473763118,
      "grad_norm": 0.8313393197161202,
      "learning_rate": 5.082421861092116e-06,
      "loss": 0.0819,
      "step": 660
    },
    {
      "epoch": 0.9910044977511244,
      "grad_norm": 0.6281079463846668,
      "learning_rate": 5.0706481584636605e-06,
      "loss": 0.0724,
      "step": 661
    },
    {
      "epoch": 0.992503748125937,
      "grad_norm": 0.7039403329670302,
      "learning_rate": 5.0588740640132805e-06,
      "loss": 0.0623,
      "step": 662
    },
    {
      "epoch": 0.9940029985007496,
      "grad_norm": 0.7251278343671408,
      "learning_rate": 5.047099643041312e-06,
      "loss": 0.0654,
      "step": 663
    },
    {
      "epoch": 0.9955022488755623,
      "grad_norm": 0.6872163129446606,
      "learning_rate": 5.0353249608499e-06,
      "loss": 0.0748,
      "step": 664
    },
    {
      "epoch": 0.9970014992503748,
      "grad_norm": 0.9594106244230192,
      "learning_rate": 5.023550082742637e-06,
      "loss": 0.1157,
      "step": 665
    },
    {
      "epoch": 0.9985007496251874,
      "grad_norm": 0.7988296637561133,
      "learning_rate": 5.011775074024202e-06,
      "loss": 0.0704,
      "step": 666
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.4862993098481467,
      "learning_rate": 5e-06,
      "loss": 0.0362,
      "step": 667
    },
    {
      "epoch": 1.0014992503748126,
      "grad_norm": 0.4714614342777134,
      "learning_rate": 4.988224925975799e-06,
      "loss": 0.0381,
      "step": 668
    },
    {
      "epoch": 1.0029985007496252,
      "grad_norm": 0.4931722534365545,
      "learning_rate": 4.976449917257365e-06,
      "loss": 0.031,
      "step": 669
    },
    {
      "epoch": 1.0044977511244377,
      "grad_norm": 0.4514637246410976,
      "learning_rate": 4.964675039150102e-06,
      "loss": 0.0331,
      "step": 670
    },
    {
      "epoch": 1.0059970014992503,
      "grad_norm": 0.5447176883023673,
      "learning_rate": 4.952900356958689e-06,
      "loss": 0.0426,
      "step": 671
    },
    {
      "epoch": 1.0074962518740629,
      "grad_norm": 0.5292425180867121,
      "learning_rate": 4.941125935986721e-06,
      "loss": 0.0334,
      "step": 672
    },
    {
      "epoch": 1.0089955022488755,
      "grad_norm": 0.39912159365455835,
      "learning_rate": 4.929351841536342e-06,
      "loss": 0.0312,
      "step": 673
    },
    {
      "epoch": 1.0104947526236883,
      "grad_norm": 0.3816546075996076,
      "learning_rate": 4.917578138907884e-06,
      "loss": 0.0307,
      "step": 674
    },
    {
      "epoch": 1.0119940029985008,
      "grad_norm": 0.5070447865151556,
      "learning_rate": 4.90580489339951e-06,
      "loss": 0.034,
      "step": 675
    },
    {
      "epoch": 1.0134932533733134,
      "grad_norm": 0.42091080488215926,
      "learning_rate": 4.894032170306846e-06,
      "loss": 0.0251,
      "step": 676
    },
    {
      "epoch": 1.014992503748126,
      "grad_norm": 0.348488760124024,
      "learning_rate": 4.882260034922618e-06,
      "loss": 0.024,
      "step": 677
    },
    {
      "epoch": 1.0164917541229386,
      "grad_norm": 0.33417342426628344,
      "learning_rate": 4.870488552536296e-06,
      "loss": 0.0223,
      "step": 678
    },
    {
      "epoch": 1.0179910044977512,
      "grad_norm": 0.5921570010075866,
      "learning_rate": 4.858717788433725e-06,
      "loss": 0.0396,
      "step": 679
    },
    {
      "epoch": 1.0194902548725637,
      "grad_norm": 0.6852832771496845,
      "learning_rate": 4.846947807896771e-06,
      "loss": 0.0458,
      "step": 680
    },
    {
      "epoch": 1.0209895052473763,
      "grad_norm": 0.5056325889307726,
      "learning_rate": 4.83517867620295e-06,
      "loss": 0.0284,
      "step": 681
    },
    {
      "epoch": 1.0224887556221889,
      "grad_norm": 0.32273811112798284,
      "learning_rate": 4.823410458625072e-06,
      "loss": 0.0254,
      "step": 682
    },
    {
      "epoch": 1.0239880059970015,
      "grad_norm": 0.3904974767776841,
      "learning_rate": 4.811643220430877e-06,
      "loss": 0.0274,
      "step": 683
    },
    {
      "epoch": 1.025487256371814,
      "grad_norm": 0.5487428346315935,
      "learning_rate": 4.7998770268826726e-06,
      "loss": 0.0398,
      "step": 684
    },
    {
      "epoch": 1.0269865067466266,
      "grad_norm": 0.3544058395155773,
      "learning_rate": 4.788111943236973e-06,
      "loss": 0.024,
      "step": 685
    },
    {
      "epoch": 1.0284857571214392,
      "grad_norm": 0.4201746663487116,
      "learning_rate": 4.7763480347441395e-06,
      "loss": 0.0307,
      "step": 686
    },
    {
      "epoch": 1.0299850074962518,
      "grad_norm": 0.28816537858745955,
      "learning_rate": 4.7645853666480104e-06,
      "loss": 0.0246,
      "step": 687
    },
    {
      "epoch": 1.0314842578710646,
      "grad_norm": 0.501240227170363,
      "learning_rate": 4.752824004185548e-06,
      "loss": 0.0333,
      "step": 688
    },
    {
      "epoch": 1.0329835082458771,
      "grad_norm": 0.4340669538041209,
      "learning_rate": 4.7410640125864785e-06,
      "loss": 0.0244,
      "step": 689
    },
    {
      "epoch": 1.0344827586206897,
      "grad_norm": 0.5357606299718058,
      "learning_rate": 4.729305457072913e-06,
      "loss": 0.032,
      "step": 690
    },
    {
      "epoch": 1.0359820089955023,
      "grad_norm": 0.3363057675461335,
      "learning_rate": 4.717548402859008e-06,
      "loss": 0.0265,
      "step": 691
    },
    {
      "epoch": 1.0374812593703149,
      "grad_norm": 0.47475612091988084,
      "learning_rate": 4.7057929151505895e-06,
      "loss": 0.0405,
      "step": 692
    },
    {
      "epoch": 1.0389805097451275,
      "grad_norm": 0.44843440354610653,
      "learning_rate": 4.694039059144797e-06,
      "loss": 0.0244,
      "step": 693
    },
    {
      "epoch": 1.04047976011994,
      "grad_norm": 0.42425634859965683,
      "learning_rate": 4.6822869000297185e-06,
      "loss": 0.03,
      "step": 694
    },
    {
      "epoch": 1.0419790104947526,
      "grad_norm": 0.6724869523393214,
      "learning_rate": 4.670536502984033e-06,
      "loss": 0.0363,
      "step": 695
    },
    {
      "epoch": 1.0434782608695652,
      "grad_norm": 0.4729653096101201,
      "learning_rate": 4.6587879331766465e-06,
      "loss": 0.0271,
      "step": 696
    },
    {
      "epoch": 1.0449775112443778,
      "grad_norm": 0.41517583242757106,
      "learning_rate": 4.647041255766329e-06,
      "loss": 0.0246,
      "step": 697
    },
    {
      "epoch": 1.0464767616191903,
      "grad_norm": 0.37654407731317796,
      "learning_rate": 4.6352965359013576e-06,
      "loss": 0.0237,
      "step": 698
    },
    {
      "epoch": 1.047976011994003,
      "grad_norm": 0.5170805635742286,
      "learning_rate": 4.623553838719151e-06,
      "loss": 0.0332,
      "step": 699
    },
    {
      "epoch": 1.0494752623688155,
      "grad_norm": 0.47523582075770254,
      "learning_rate": 4.611813229345911e-06,
      "loss": 0.0289,
      "step": 700
    },
    {
      "epoch": 1.050974512743628,
      "grad_norm": 0.34733758771237794,
      "learning_rate": 4.6000747728962606e-06,
      "loss": 0.0242,
      "step": 701
    },
    {
      "epoch": 1.0524737631184409,
      "grad_norm": 0.46409220364966464,
      "learning_rate": 4.588338534472878e-06,
      "loss": 0.021,
      "step": 702
    },
    {
      "epoch": 1.0539730134932535,
      "grad_norm": 0.7867899447967596,
      "learning_rate": 4.576604579166147e-06,
      "loss": 0.0469,
      "step": 703
    },
    {
      "epoch": 1.055472263868066,
      "grad_norm": 0.6875792828129451,
      "learning_rate": 4.564872972053786e-06,
      "loss": 0.044,
      "step": 704
    },
    {
      "epoch": 1.0569715142428786,
      "grad_norm": 0.7817455527670667,
      "learning_rate": 4.553143778200486e-06,
      "loss": 0.0511,
      "step": 705
    },
    {
      "epoch": 1.0584707646176912,
      "grad_norm": 0.5254214194141112,
      "learning_rate": 4.541417062657561e-06,
      "loss": 0.0379,
      "step": 706
    },
    {
      "epoch": 1.0599700149925038,
      "grad_norm": 0.4694676782371513,
      "learning_rate": 4.529692890462574e-06,
      "loss": 0.026,
      "step": 707
    },
    {
      "epoch": 1.0614692653673163,
      "grad_norm": 0.39395612234869537,
      "learning_rate": 4.5179713266389866e-06,
      "loss": 0.0271,
      "step": 708
    },
    {
      "epoch": 1.062968515742129,
      "grad_norm": 0.43256011866873245,
      "learning_rate": 4.50625243619579e-06,
      "loss": 0.0248,
      "step": 709
    },
    {
      "epoch": 1.0644677661169415,
      "grad_norm": 0.3381222837887605,
      "learning_rate": 4.494536284127155e-06,
      "loss": 0.0159,
      "step": 710
    },
    {
      "epoch": 1.065967016491754,
      "grad_norm": 0.3997392398954939,
      "learning_rate": 4.4828229354120565e-06,
      "loss": 0.0174,
      "step": 711
    },
    {
      "epoch": 1.0674662668665666,
      "grad_norm": 0.6840911330958394,
      "learning_rate": 4.471112455013928e-06,
      "loss": 0.031,
      "step": 712
    },
    {
      "epoch": 1.0689655172413792,
      "grad_norm": 0.4513671397117882,
      "learning_rate": 4.459404907880293e-06,
      "loss": 0.0243,
      "step": 713
    },
    {
      "epoch": 1.0704647676161918,
      "grad_norm": 0.5267334601164341,
      "learning_rate": 4.447700358942407e-06,
      "loss": 0.035,
      "step": 714
    },
    {
      "epoch": 1.0719640179910046,
      "grad_norm": 0.3730551910856466,
      "learning_rate": 4.435998873114895e-06,
      "loss": 0.0209,
      "step": 715
    },
    {
      "epoch": 1.0734632683658172,
      "grad_norm": 0.5055141680134527,
      "learning_rate": 4.424300515295401e-06,
      "loss": 0.0278,
      "step": 716
    },
    {
      "epoch": 1.0749625187406298,
      "grad_norm": 0.40519730328435083,
      "learning_rate": 4.412605350364213e-06,
      "loss": 0.0194,
      "step": 717
    },
    {
      "epoch": 1.0764617691154423,
      "grad_norm": 0.3873528403478698,
      "learning_rate": 4.400913443183913e-06,
      "loss": 0.0211,
      "step": 718
    },
    {
      "epoch": 1.077961019490255,
      "grad_norm": 0.4973671859123383,
      "learning_rate": 4.389224858599015e-06,
      "loss": 0.0329,
      "step": 719
    },
    {
      "epoch": 1.0794602698650675,
      "grad_norm": 0.4296962805189119,
      "learning_rate": 4.377539661435608e-06,
      "loss": 0.0204,
      "step": 720
    },
    {
      "epoch": 1.08095952023988,
      "grad_norm": 0.5661049495636151,
      "learning_rate": 4.365857916500991e-06,
      "loss": 0.0343,
      "step": 721
    },
    {
      "epoch": 1.0824587706146926,
      "grad_norm": 0.48700921660626273,
      "learning_rate": 4.35417968858332e-06,
      "loss": 0.0244,
      "step": 722
    },
    {
      "epoch": 1.0839580209895052,
      "grad_norm": 0.5351448523985574,
      "learning_rate": 4.3425050424512405e-06,
      "loss": 0.0264,
      "step": 723
    },
    {
      "epoch": 1.0854572713643178,
      "grad_norm": 0.548520993441216,
      "learning_rate": 4.330834042853537e-06,
      "loss": 0.0279,
      "step": 724
    },
    {
      "epoch": 1.0869565217391304,
      "grad_norm": 0.959230579330483,
      "learning_rate": 4.319166754518768e-06,
      "loss": 0.0383,
      "step": 725
    },
    {
      "epoch": 1.088455772113943,
      "grad_norm": 0.5201352878198201,
      "learning_rate": 4.30750324215491e-06,
      "loss": 0.0224,
      "step": 726
    },
    {
      "epoch": 1.0899550224887555,
      "grad_norm": 0.5124799299801699,
      "learning_rate": 4.295843570448998e-06,
      "loss": 0.0272,
      "step": 727
    },
    {
      "epoch": 1.0914542728635683,
      "grad_norm": 0.5236084990504366,
      "learning_rate": 4.284187804066764e-06,
      "loss": 0.0273,
      "step": 728
    },
    {
      "epoch": 1.092953523238381,
      "grad_norm": 0.5668413045170417,
      "learning_rate": 4.272536007652281e-06,
      "loss": 0.0257,
      "step": 729
    },
    {
      "epoch": 1.0944527736131935,
      "grad_norm": 0.4023916028852343,
      "learning_rate": 4.260888245827608e-06,
      "loss": 0.0201,
      "step": 730
    },
    {
      "epoch": 1.095952023988006,
      "grad_norm": 0.46816636476142315,
      "learning_rate": 4.249244583192425e-06,
      "loss": 0.0296,
      "step": 731
    },
    {
      "epoch": 1.0974512743628186,
      "grad_norm": 0.5385870156439213,
      "learning_rate": 4.237605084323676e-06,
      "loss": 0.0232,
      "step": 732
    },
    {
      "epoch": 1.0989505247376312,
      "grad_norm": 0.6979503058378336,
      "learning_rate": 4.225969813775215e-06,
      "loss": 0.0302,
      "step": 733
    },
    {
      "epoch": 1.1004497751124438,
      "grad_norm": 0.448454227523974,
      "learning_rate": 4.214338836077444e-06,
      "loss": 0.0287,
      "step": 734
    },
    {
      "epoch": 1.1019490254872564,
      "grad_norm": 0.4909104912743045,
      "learning_rate": 4.202712215736955e-06,
      "loss": 0.0239,
      "step": 735
    },
    {
      "epoch": 1.103448275862069,
      "grad_norm": 0.4865918175667567,
      "learning_rate": 4.191090017236177e-06,
      "loss": 0.0228,
      "step": 736
    },
    {
      "epoch": 1.1049475262368815,
      "grad_norm": 0.39453414947437837,
      "learning_rate": 4.1794723050330125e-06,
      "loss": 0.0178,
      "step": 737
    },
    {
      "epoch": 1.106446776611694,
      "grad_norm": 0.4932744332368071,
      "learning_rate": 4.167859143560484e-06,
      "loss": 0.0238,
      "step": 738
    },
    {
      "epoch": 1.1079460269865067,
      "grad_norm": 0.5267737317406019,
      "learning_rate": 4.1562505972263735e-06,
      "loss": 0.0254,
      "step": 739
    },
    {
      "epoch": 1.1094452773613193,
      "grad_norm": 0.751513161496313,
      "learning_rate": 4.144646730412868e-06,
      "loss": 0.0351,
      "step": 740
    },
    {
      "epoch": 1.1109445277361318,
      "grad_norm": 0.5489928162363794,
      "learning_rate": 4.133047607476202e-06,
      "loss": 0.0315,
      "step": 741
    },
    {
      "epoch": 1.1124437781109444,
      "grad_norm": 0.4758057154366718,
      "learning_rate": 4.121453292746297e-06,
      "loss": 0.0203,
      "step": 742
    },
    {
      "epoch": 1.1139430284857572,
      "grad_norm": 0.773634002035264,
      "learning_rate": 4.109863850526413e-06,
      "loss": 0.0317,
      "step": 743
    },
    {
      "epoch": 1.1154422788605698,
      "grad_norm": 0.4751072179957143,
      "learning_rate": 4.098279345092783e-06,
      "loss": 0.0221,
      "step": 744
    },
    {
      "epoch": 1.1169415292353824,
      "grad_norm": 0.4762481618789892,
      "learning_rate": 4.086699840694262e-06,
      "loss": 0.0192,
      "step": 745
    },
    {
      "epoch": 1.118440779610195,
      "grad_norm": 0.47118588296557146,
      "learning_rate": 4.075125401551963e-06,
      "loss": 0.0222,
      "step": 746
    },
    {
      "epoch": 1.1199400299850075,
      "grad_norm": 0.564207526077115,
      "learning_rate": 4.063556091858914e-06,
      "loss": 0.0303,
      "step": 747
    },
    {
      "epoch": 1.12143928035982,
      "grad_norm": 0.6206419634111711,
      "learning_rate": 4.051991975779691e-06,
      "loss": 0.0325,
      "step": 748
    },
    {
      "epoch": 1.1229385307346327,
      "grad_norm": 0.6250554766780263,
      "learning_rate": 4.040433117450066e-06,
      "loss": 0.0286,
      "step": 749
    },
    {
      "epoch": 1.1244377811094453,
      "grad_norm": 0.6557373598276082,
      "learning_rate": 4.0288795809766516e-06,
      "loss": 0.0312,
      "step": 750
    },
    {
      "epoch": 1.1259370314842578,
      "grad_norm": 0.5450022275255627,
      "learning_rate": 4.017331430436543e-06,
      "loss": 0.0314,
      "step": 751
    },
    {
      "epoch": 1.1274362818590704,
      "grad_norm": 0.7817794891776698,
      "learning_rate": 4.005788729876968e-06,
      "loss": 0.0291,
      "step": 752
    },
    {
      "epoch": 1.128935532233883,
      "grad_norm": 0.5024870821977194,
      "learning_rate": 3.994251543314925e-06,
      "loss": 0.0258,
      "step": 753
    },
    {
      "epoch": 1.1304347826086956,
      "grad_norm": 0.601430799185042,
      "learning_rate": 3.982719934736832e-06,
      "loss": 0.0307,
      "step": 754
    },
    {
      "epoch": 1.1319340329835081,
      "grad_norm": 0.5950591484456509,
      "learning_rate": 3.971193968098172e-06,
      "loss": 0.0353,
      "step": 755
    },
    {
      "epoch": 1.133433283358321,
      "grad_norm": 0.5501327053425047,
      "learning_rate": 3.959673707323135e-06,
      "loss": 0.0357,
      "step": 756
    },
    {
      "epoch": 1.1349325337331335,
      "grad_norm": 0.5309205312658972,
      "learning_rate": 3.948159216304269e-06,
      "loss": 0.0292,
      "step": 757
    },
    {
      "epoch": 1.136431784107946,
      "grad_norm": 0.5161788282077046,
      "learning_rate": 3.93665055890212e-06,
      "loss": 0.0212,
      "step": 758
    },
    {
      "epoch": 1.1379310344827587,
      "grad_norm": 0.646140052811195,
      "learning_rate": 3.92514779894488e-06,
      "loss": 0.0321,
      "step": 759
    },
    {
      "epoch": 1.1394302848575713,
      "grad_norm": 0.4297527131171912,
      "learning_rate": 3.9136510002280344e-06,
      "loss": 0.0232,
      "step": 760
    },
    {
      "epoch": 1.1409295352323838,
      "grad_norm": 0.41533619511706904,
      "learning_rate": 3.902160226514007e-06,
      "loss": 0.021,
      "step": 761
    },
    {
      "epoch": 1.1424287856071964,
      "grad_norm": 1.0244368271433324,
      "learning_rate": 3.8906755415318045e-06,
      "loss": 0.0374,
      "step": 762
    },
    {
      "epoch": 1.143928035982009,
      "grad_norm": 0.49863561175914095,
      "learning_rate": 3.8791970089766665e-06,
      "loss": 0.0271,
      "step": 763
    },
    {
      "epoch": 1.1454272863568216,
      "grad_norm": 0.6193638903494821,
      "learning_rate": 3.86772469250971e-06,
      "loss": 0.0284,
      "step": 764
    },
    {
      "epoch": 1.1469265367316341,
      "grad_norm": 0.4449285812300328,
      "learning_rate": 3.856258655757578e-06,
      "loss": 0.0163,
      "step": 765
    },
    {
      "epoch": 1.1484257871064467,
      "grad_norm": 0.6559867275908414,
      "learning_rate": 3.844798962312085e-06,
      "loss": 0.039,
      "step": 766
    },
    {
      "epoch": 1.1499250374812593,
      "grad_norm": 0.5193079352170892,
      "learning_rate": 3.833345675729863e-06,
      "loss": 0.0317,
      "step": 767
    },
    {
      "epoch": 1.1514242878560719,
      "grad_norm": 0.64184494564955,
      "learning_rate": 3.821898859532013e-06,
      "loss": 0.03,
      "step": 768
    },
    {
      "epoch": 1.1529235382308847,
      "grad_norm": 0.5528166901737965,
      "learning_rate": 3.8104585772037493e-06,
      "loss": 0.0208,
      "step": 769
    },
    {
      "epoch": 1.1544227886056972,
      "grad_norm": 0.5804094524857409,
      "learning_rate": 3.7990248921940485e-06,
      "loss": 0.0338,
      "step": 770
    },
    {
      "epoch": 1.1559220389805098,
      "grad_norm": 0.6649615548649997,
      "learning_rate": 3.787597867915303e-06,
      "loss": 0.0282,
      "step": 771
    },
    {
      "epoch": 1.1574212893553224,
      "grad_norm": 0.6705566090771073,
      "learning_rate": 3.7761775677429567e-06,
      "loss": 0.0278,
      "step": 772
    },
    {
      "epoch": 1.158920539730135,
      "grad_norm": 0.4733577730078053,
      "learning_rate": 3.7647640550151666e-06,
      "loss": 0.0238,
      "step": 773
    },
    {
      "epoch": 1.1604197901049476,
      "grad_norm": 0.5931187543928851,
      "learning_rate": 3.7533573930324395e-06,
      "loss": 0.0249,
      "step": 774
    },
    {
      "epoch": 1.1619190404797601,
      "grad_norm": 0.48236239293961397,
      "learning_rate": 3.7419576450572924e-06,
      "loss": 0.0264,
      "step": 775
    },
    {
      "epoch": 1.1634182908545727,
      "grad_norm": 0.5776663956333957,
      "learning_rate": 3.7305648743138966e-06,
      "loss": 0.026,
      "step": 776
    },
    {
      "epoch": 1.1649175412293853,
      "grad_norm": 0.5044807521783605,
      "learning_rate": 3.7191791439877236e-06,
      "loss": 0.0263,
      "step": 777
    },
    {
      "epoch": 1.1664167916041979,
      "grad_norm": 0.5593626580801786,
      "learning_rate": 3.7078005172252015e-06,
      "loss": 0.0182,
      "step": 778
    },
    {
      "epoch": 1.1679160419790104,
      "grad_norm": 0.5803367424590659,
      "learning_rate": 3.6964290571333583e-06,
      "loss": 0.0294,
      "step": 779
    },
    {
      "epoch": 1.169415292353823,
      "grad_norm": 0.5975429608944614,
      "learning_rate": 3.6850648267794776e-06,
      "loss": 0.0256,
      "step": 780
    },
    {
      "epoch": 1.1709145427286356,
      "grad_norm": 0.4436076190155131,
      "learning_rate": 3.673707889190744e-06,
      "loss": 0.0183,
      "step": 781
    },
    {
      "epoch": 1.1724137931034484,
      "grad_norm": 0.47485244493382683,
      "learning_rate": 3.662358307353897e-06,
      "loss": 0.0234,
      "step": 782
    },
    {
      "epoch": 1.1739130434782608,
      "grad_norm": 0.5353878598162267,
      "learning_rate": 3.6510161442148783e-06,
      "loss": 0.0202,
      "step": 783
    },
    {
      "epoch": 1.1754122938530736,
      "grad_norm": 0.5777562200179522,
      "learning_rate": 3.63968146267849e-06,
      "loss": 0.0318,
      "step": 784
    },
    {
      "epoch": 1.1769115442278861,
      "grad_norm": 0.6145817020129336,
      "learning_rate": 3.6283543256080334e-06,
      "loss": 0.0327,
      "step": 785
    },
    {
      "epoch": 1.1784107946026987,
      "grad_norm": 0.526537639823521,
      "learning_rate": 3.6170347958249728e-06,
      "loss": 0.0222,
      "step": 786
    },
    {
      "epoch": 1.1799100449775113,
      "grad_norm": 0.7044820964385141,
      "learning_rate": 3.605722936108579e-06,
      "loss": 0.0285,
      "step": 787
    },
    {
      "epoch": 1.1814092953523239,
      "grad_norm": 0.7214982178400132,
      "learning_rate": 3.5944188091955843e-06,
      "loss": 0.0292,
      "step": 788
    },
    {
      "epoch": 1.1829085457271364,
      "grad_norm": 0.4879432625261694,
      "learning_rate": 3.5831224777798346e-06,
      "loss": 0.022,
      "step": 789
    },
    {
      "epoch": 1.184407796101949,
      "grad_norm": 0.40171722657229975,
      "learning_rate": 3.5718340045119416e-06,
      "loss": 0.0159,
      "step": 790
    },
    {
      "epoch": 1.1859070464767616,
      "grad_norm": 0.5713374212602449,
      "learning_rate": 3.5605534519989327e-06,
      "loss": 0.0269,
      "step": 791
    },
    {
      "epoch": 1.1874062968515742,
      "grad_norm": 0.501233502776701,
      "learning_rate": 3.5492808828039083e-06,
      "loss": 0.0166,
      "step": 792
    },
    {
      "epoch": 1.1889055472263867,
      "grad_norm": 0.5159094352279,
      "learning_rate": 3.538016359445692e-06,
      "loss": 0.0252,
      "step": 793
    },
    {
      "epoch": 1.1904047976011993,
      "grad_norm": 0.572697071143813,
      "learning_rate": 3.5267599443984848e-06,
      "loss": 0.0288,
      "step": 794
    },
    {
      "epoch": 1.191904047976012,
      "grad_norm": 0.5776061885697024,
      "learning_rate": 3.5155117000915153e-06,
      "loss": 0.0382,
      "step": 795
    },
    {
      "epoch": 1.1934032983508245,
      "grad_norm": 0.5249283813271437,
      "learning_rate": 3.5042716889086998e-06,
      "loss": 0.0243,
      "step": 796
    },
    {
      "epoch": 1.1949025487256373,
      "grad_norm": 0.4649906582619793,
      "learning_rate": 3.493039973188289e-06,
      "loss": 0.0207,
      "step": 797
    },
    {
      "epoch": 1.1964017991004499,
      "grad_norm": 0.5227890044613182,
      "learning_rate": 3.481816615222533e-06,
      "loss": 0.0215,
      "step": 798
    },
    {
      "epoch": 1.1979010494752624,
      "grad_norm": 0.41799749894451804,
      "learning_rate": 3.470601677257323e-06,
      "loss": 0.024,
      "step": 799
    },
    {
      "epoch": 1.199400299850075,
      "grad_norm": 0.5576500283213088,
      "learning_rate": 3.459395221491853e-06,
      "loss": 0.0225,
      "step": 800
    },
    {
      "epoch": 1.199400299850075,
      "eval_loss": 0.10087519884109497,
      "eval_runtime": 2.193,
      "eval_samples_per_second": 24.624,
      "eval_steps_per_second": 6.384,
      "step": 800
    },
    {
      "epoch": 1.2008995502248876,
      "grad_norm": 0.4630604580846507,
      "learning_rate": 3.4481973100782756e-06,
      "loss": 0.0214,
      "step": 801
    },
    {
      "epoch": 1.2023988005997002,
      "grad_norm": 0.5031818688705914,
      "learning_rate": 3.4370080051213527e-06,
      "loss": 0.0227,
      "step": 802
    },
    {
      "epoch": 1.2038980509745127,
      "grad_norm": 0.4808222196273266,
      "learning_rate": 3.4258273686781156e-06,
      "loss": 0.0273,
      "step": 803
    },
    {
      "epoch": 1.2053973013493253,
      "grad_norm": 0.7504175000476707,
      "learning_rate": 3.4146554627575207e-06,
      "loss": 0.0312,
      "step": 804
    },
    {
      "epoch": 1.206896551724138,
      "grad_norm": 0.6130994742018914,
      "learning_rate": 3.403492349320101e-06,
      "loss": 0.0295,
      "step": 805
    },
    {
      "epoch": 1.2083958020989505,
      "grad_norm": 0.37591621419366317,
      "learning_rate": 3.392338090277628e-06,
      "loss": 0.0155,
      "step": 806
    },
    {
      "epoch": 1.209895052473763,
      "grad_norm": 0.5237789816414232,
      "learning_rate": 3.3811927474927644e-06,
      "loss": 0.0255,
      "step": 807
    },
    {
      "epoch": 1.2113943028485756,
      "grad_norm": 0.4671325545690538,
      "learning_rate": 3.3700563827787224e-06,
      "loss": 0.0213,
      "step": 808
    },
    {
      "epoch": 1.2128935532233882,
      "grad_norm": 0.44500262574673494,
      "learning_rate": 3.358929057898922e-06,
      "loss": 0.0181,
      "step": 809
    },
    {
      "epoch": 1.214392803598201,
      "grad_norm": 0.5518971820027787,
      "learning_rate": 3.3478108345666456e-06,
      "loss": 0.0278,
      "step": 810
    },
    {
      "epoch": 1.2158920539730136,
      "grad_norm": 0.5216849249717327,
      "learning_rate": 3.3367017744446995e-06,
      "loss": 0.0268,
      "step": 811
    },
    {
      "epoch": 1.2173913043478262,
      "grad_norm": 0.7067806859827608,
      "learning_rate": 3.3256019391450696e-06,
      "loss": 0.0386,
      "step": 812
    },
    {
      "epoch": 1.2188905547226387,
      "grad_norm": 0.47423826928666324,
      "learning_rate": 3.314511390228578e-06,
      "loss": 0.0246,
      "step": 813
    },
    {
      "epoch": 1.2203898050974513,
      "grad_norm": 0.6452965674867218,
      "learning_rate": 3.303430189204545e-06,
      "loss": 0.029,
      "step": 814
    },
    {
      "epoch": 1.221889055472264,
      "grad_norm": 0.5704433370205287,
      "learning_rate": 3.2923583975304474e-06,
      "loss": 0.0301,
      "step": 815
    },
    {
      "epoch": 1.2233883058470765,
      "grad_norm": 0.5793165854222945,
      "learning_rate": 3.2812960766115747e-06,
      "loss": 0.0252,
      "step": 816
    },
    {
      "epoch": 1.224887556221889,
      "grad_norm": 0.8142186154683048,
      "learning_rate": 3.270243287800693e-06,
      "loss": 0.0373,
      "step": 817
    },
    {
      "epoch": 1.2263868065967016,
      "grad_norm": 0.4198823348686633,
      "learning_rate": 3.2592000923976997e-06,
      "loss": 0.0158,
      "step": 818
    },
    {
      "epoch": 1.2278860569715142,
      "grad_norm": 0.6501561065725391,
      "learning_rate": 3.2481665516492876e-06,
      "loss": 0.0282,
      "step": 819
    },
    {
      "epoch": 1.2293853073463268,
      "grad_norm": 0.48607075316420323,
      "learning_rate": 3.2371427267486044e-06,
      "loss": 0.0228,
      "step": 820
    },
    {
      "epoch": 1.2308845577211394,
      "grad_norm": 0.5484972173495635,
      "learning_rate": 3.2261286788349127e-06,
      "loss": 0.022,
      "step": 821
    },
    {
      "epoch": 1.232383808095952,
      "grad_norm": 0.608681928583066,
      "learning_rate": 3.2151244689932505e-06,
      "loss": 0.026,
      "step": 822
    },
    {
      "epoch": 1.2338830584707647,
      "grad_norm": 0.5942532030608312,
      "learning_rate": 3.2041301582540903e-06,
      "loss": 0.0188,
      "step": 823
    },
    {
      "epoch": 1.235382308845577,
      "grad_norm": 0.677922556614439,
      "learning_rate": 3.1931458075930046e-06,
      "loss": 0.0296,
      "step": 824
    },
    {
      "epoch": 1.23688155922039,
      "grad_norm": 0.4994199849581584,
      "learning_rate": 3.182171477930332e-06,
      "loss": 0.0226,
      "step": 825
    },
    {
      "epoch": 1.2383808095952025,
      "grad_norm": 0.466744300553662,
      "learning_rate": 3.171207230130826e-06,
      "loss": 0.0228,
      "step": 826
    },
    {
      "epoch": 1.239880059970015,
      "grad_norm": 0.46939461730374626,
      "learning_rate": 3.1602531250033286e-06,
      "loss": 0.0203,
      "step": 827
    },
    {
      "epoch": 1.2413793103448276,
      "grad_norm": 0.5749000498919572,
      "learning_rate": 3.149309223300428e-06,
      "loss": 0.0214,
      "step": 828
    },
    {
      "epoch": 1.2428785607196402,
      "grad_norm": 0.5985901783601713,
      "learning_rate": 3.1383755857181253e-06,
      "loss": 0.0307,
      "step": 829
    },
    {
      "epoch": 1.2443778110944528,
      "grad_norm": 0.47387399319250706,
      "learning_rate": 3.1274522728954928e-06,
      "loss": 0.0214,
      "step": 830
    },
    {
      "epoch": 1.2458770614692654,
      "grad_norm": 0.49133236380867024,
      "learning_rate": 3.1165393454143423e-06,
      "loss": 0.0225,
      "step": 831
    },
    {
      "epoch": 1.247376311844078,
      "grad_norm": 0.7321160958754388,
      "learning_rate": 3.1056368637988876e-06,
      "loss": 0.0333,
      "step": 832
    },
    {
      "epoch": 1.2488755622188905,
      "grad_norm": 3.196419585593132,
      "learning_rate": 3.0947448885154085e-06,
      "loss": 0.0225,
      "step": 833
    },
    {
      "epoch": 1.250374812593703,
      "grad_norm": 0.6687748509396151,
      "learning_rate": 3.0838634799719157e-06,
      "loss": 0.0342,
      "step": 834
    },
    {
      "epoch": 1.2518740629685157,
      "grad_norm": 0.4970137247695218,
      "learning_rate": 3.072992698517815e-06,
      "loss": 0.025,
      "step": 835
    },
    {
      "epoch": 1.2533733133433285,
      "grad_norm": 0.4980484974537264,
      "learning_rate": 3.0621326044435738e-06,
      "loss": 0.0222,
      "step": 836
    },
    {
      "epoch": 1.2548725637181408,
      "grad_norm": 0.5461171179288082,
      "learning_rate": 3.0512832579803873e-06,
      "loss": 0.0335,
      "step": 837
    },
    {
      "epoch": 1.2563718140929536,
      "grad_norm": 0.5908014736589887,
      "learning_rate": 3.0404447192998398e-06,
      "loss": 0.0249,
      "step": 838
    },
    {
      "epoch": 1.2578710644677662,
      "grad_norm": 0.6495442494943365,
      "learning_rate": 3.029617048513579e-06,
      "loss": 0.0276,
      "step": 839
    },
    {
      "epoch": 1.2593703148425788,
      "grad_norm": 0.4244988784132044,
      "learning_rate": 3.0188003056729752e-06,
      "loss": 0.0214,
      "step": 840
    },
    {
      "epoch": 1.2608695652173914,
      "grad_norm": 0.6188489352193449,
      "learning_rate": 3.007994550768793e-06,
      "loss": 0.0274,
      "step": 841
    },
    {
      "epoch": 1.262368815592204,
      "grad_norm": 0.6887448659423938,
      "learning_rate": 2.9971998437308546e-06,
      "loss": 0.029,
      "step": 842
    },
    {
      "epoch": 1.2638680659670165,
      "grad_norm": 0.7273693907106565,
      "learning_rate": 2.9864162444277118e-06,
      "loss": 0.0408,
      "step": 843
    },
    {
      "epoch": 1.265367316341829,
      "grad_norm": 1.3017603505092656,
      "learning_rate": 2.97564381266631e-06,
      "loss": 0.0287,
      "step": 844
    },
    {
      "epoch": 1.2668665667166417,
      "grad_norm": 0.44171839591460266,
      "learning_rate": 2.964882608191659e-06,
      "loss": 0.018,
      "step": 845
    },
    {
      "epoch": 1.2683658170914542,
      "grad_norm": 0.5517856379328671,
      "learning_rate": 2.954132690686502e-06,
      "loss": 0.0267,
      "step": 846
    },
    {
      "epoch": 1.2698650674662668,
      "grad_norm": 0.5795634513244025,
      "learning_rate": 2.9433941197709813e-06,
      "loss": 0.0294,
      "step": 847
    },
    {
      "epoch": 1.2713643178410794,
      "grad_norm": 0.6762474078668322,
      "learning_rate": 2.9326669550023124e-06,
      "loss": 0.0299,
      "step": 848
    },
    {
      "epoch": 1.272863568215892,
      "grad_norm": 0.5175023571134142,
      "learning_rate": 2.921951255874449e-06,
      "loss": 0.0222,
      "step": 849
    },
    {
      "epoch": 1.2743628185907045,
      "grad_norm": 0.5733773290972851,
      "learning_rate": 2.9112470818177563e-06,
      "loss": 0.0247,
      "step": 850
    },
    {
      "epoch": 1.2758620689655173,
      "grad_norm": 0.534639933121592,
      "learning_rate": 2.9005544921986774e-06,
      "loss": 0.0296,
      "step": 851
    },
    {
      "epoch": 1.2773613193403297,
      "grad_norm": 0.5298152541366488,
      "learning_rate": 2.8898735463194128e-06,
      "loss": 0.025,
      "step": 852
    },
    {
      "epoch": 1.2788605697151425,
      "grad_norm": 0.5971858664423838,
      "learning_rate": 2.8792043034175817e-06,
      "loss": 0.0255,
      "step": 853
    },
    {
      "epoch": 1.280359820089955,
      "grad_norm": 0.4854114171022982,
      "learning_rate": 2.8685468226658974e-06,
      "loss": 0.0271,
      "step": 854
    },
    {
      "epoch": 1.2818590704647677,
      "grad_norm": 0.45918114104238916,
      "learning_rate": 2.85790116317184e-06,
      "loss": 0.0191,
      "step": 855
    },
    {
      "epoch": 1.2833583208395802,
      "grad_norm": 0.6309459954386983,
      "learning_rate": 2.8472673839773267e-06,
      "loss": 0.0334,
      "step": 856
    },
    {
      "epoch": 1.2848575712143928,
      "grad_norm": 0.521465842522808,
      "learning_rate": 2.8366455440583874e-06,
      "loss": 0.0249,
      "step": 857
    },
    {
      "epoch": 1.2863568215892054,
      "grad_norm": 0.7168803811851245,
      "learning_rate": 2.8260357023248323e-06,
      "loss": 0.0339,
      "step": 858
    },
    {
      "epoch": 1.287856071964018,
      "grad_norm": 0.5645479583393774,
      "learning_rate": 2.815437917619932e-06,
      "loss": 0.0293,
      "step": 859
    },
    {
      "epoch": 1.2893553223388305,
      "grad_norm": 0.7014181902083694,
      "learning_rate": 2.804852248720085e-06,
      "loss": 0.0384,
      "step": 860
    },
    {
      "epoch": 1.2908545727136431,
      "grad_norm": 0.48128628723304223,
      "learning_rate": 2.7942787543344957e-06,
      "loss": 0.0207,
      "step": 861
    },
    {
      "epoch": 1.2923538230884557,
      "grad_norm": 0.4801637872616389,
      "learning_rate": 2.783717493104846e-06,
      "loss": 0.017,
      "step": 862
    },
    {
      "epoch": 1.2938530734632683,
      "grad_norm": 0.5279005192080837,
      "learning_rate": 2.7731685236049745e-06,
      "loss": 0.0276,
      "step": 863
    },
    {
      "epoch": 1.295352323838081,
      "grad_norm": 0.5105732175930656,
      "learning_rate": 2.762631904340546e-06,
      "loss": 0.0219,
      "step": 864
    },
    {
      "epoch": 1.2968515742128934,
      "grad_norm": 0.697449115121136,
      "learning_rate": 2.7521076937487248e-06,
      "loss": 0.0274,
      "step": 865
    },
    {
      "epoch": 1.2983508245877062,
      "grad_norm": 0.44052223398312984,
      "learning_rate": 2.7415959501978674e-06,
      "loss": 0.0161,
      "step": 866
    },
    {
      "epoch": 1.2998500749625188,
      "grad_norm": 0.5692680696184986,
      "learning_rate": 2.731096731987177e-06,
      "loss": 0.0238,
      "step": 867
    },
    {
      "epoch": 1.3013493253373314,
      "grad_norm": 0.6663448729566707,
      "learning_rate": 2.7206100973463958e-06,
      "loss": 0.0252,
      "step": 868
    },
    {
      "epoch": 1.302848575712144,
      "grad_norm": 0.6459577220170198,
      "learning_rate": 2.71013610443547e-06,
      "loss": 0.0302,
      "step": 869
    },
    {
      "epoch": 1.3043478260869565,
      "grad_norm": 0.5350930974500245,
      "learning_rate": 2.6996748113442397e-06,
      "loss": 0.0232,
      "step": 870
    },
    {
      "epoch": 1.3058470764617691,
      "grad_norm": 0.472803271248211,
      "learning_rate": 2.689226276092107e-06,
      "loss": 0.0208,
      "step": 871
    },
    {
      "epoch": 1.3073463268365817,
      "grad_norm": 0.698273125508641,
      "learning_rate": 2.6787905566277185e-06,
      "loss": 0.0388,
      "step": 872
    },
    {
      "epoch": 1.3088455772113943,
      "grad_norm": 0.5574191872691284,
      "learning_rate": 2.6683677108286423e-06,
      "loss": 0.0295,
      "step": 873
    },
    {
      "epoch": 1.3103448275862069,
      "grad_norm": 0.6679581440201905,
      "learning_rate": 2.65795779650105e-06,
      "loss": 0.0265,
      "step": 874
    },
    {
      "epoch": 1.3118440779610194,
      "grad_norm": 0.6540754634060487,
      "learning_rate": 2.6475608713793923e-06,
      "loss": 0.0282,
      "step": 875
    },
    {
      "epoch": 1.313343328335832,
      "grad_norm": 0.5178226089461461,
      "learning_rate": 2.6371769931260806e-06,
      "loss": 0.0214,
      "step": 876
    },
    {
      "epoch": 1.3148425787106448,
      "grad_norm": 0.5553271736144924,
      "learning_rate": 2.6268062193311672e-06,
      "loss": 0.0252,
      "step": 877
    },
    {
      "epoch": 1.3163418290854572,
      "grad_norm": 0.5581283423489376,
      "learning_rate": 2.6164486075120245e-06,
      "loss": 0.0222,
      "step": 878
    },
    {
      "epoch": 1.31784107946027,
      "grad_norm": 0.7089351196949294,
      "learning_rate": 2.606104215113033e-06,
      "loss": 0.0295,
      "step": 879
    },
    {
      "epoch": 1.3193403298350825,
      "grad_norm": 0.5332844231147243,
      "learning_rate": 2.5957730995052477e-06,
      "loss": 0.0231,
      "step": 880
    },
    {
      "epoch": 1.3208395802098951,
      "grad_norm": 0.5485206627830297,
      "learning_rate": 2.585455317986095e-06,
      "loss": 0.0309,
      "step": 881
    },
    {
      "epoch": 1.3223388305847077,
      "grad_norm": 0.6125620850418239,
      "learning_rate": 2.5751509277790487e-06,
      "loss": 0.0254,
      "step": 882
    },
    {
      "epoch": 1.3238380809595203,
      "grad_norm": 0.4584192652419654,
      "learning_rate": 2.5648599860333122e-06,
      "loss": 0.0181,
      "step": 883
    },
    {
      "epoch": 1.3253373313343328,
      "grad_norm": 0.5620383637350448,
      "learning_rate": 2.554582549823502e-06,
      "loss": 0.021,
      "step": 884
    },
    {
      "epoch": 1.3268365817091454,
      "grad_norm": 0.9936021321006764,
      "learning_rate": 2.5443186761493327e-06,
      "loss": 0.0551,
      "step": 885
    },
    {
      "epoch": 1.328335832083958,
      "grad_norm": 0.635754759278019,
      "learning_rate": 2.5340684219352977e-06,
      "loss": 0.0258,
      "step": 886
    },
    {
      "epoch": 1.3298350824587706,
      "grad_norm": 0.6036440796682117,
      "learning_rate": 2.523831844030358e-06,
      "loss": 0.0239,
      "step": 887
    },
    {
      "epoch": 1.3313343328335832,
      "grad_norm": 0.5832425484606606,
      "learning_rate": 2.513608999207622e-06,
      "loss": 0.0265,
      "step": 888
    },
    {
      "epoch": 1.3328335832083957,
      "grad_norm": 0.7922416263435843,
      "learning_rate": 2.503399944164035e-06,
      "loss": 0.0274,
      "step": 889
    },
    {
      "epoch": 1.3343328335832085,
      "grad_norm": 0.5538676882712676,
      "learning_rate": 2.4932047355200613e-06,
      "loss": 0.0269,
      "step": 890
    },
    {
      "epoch": 1.3358320839580209,
      "grad_norm": 0.610250178656243,
      "learning_rate": 2.483023429819372e-06,
      "loss": 0.0262,
      "step": 891
    },
    {
      "epoch": 1.3373313343328337,
      "grad_norm": 0.7246597950598686,
      "learning_rate": 2.472856083528531e-06,
      "loss": 0.0314,
      "step": 892
    },
    {
      "epoch": 1.338830584707646,
      "grad_norm": 0.5292803505156931,
      "learning_rate": 2.4627027530366836e-06,
      "loss": 0.0191,
      "step": 893
    },
    {
      "epoch": 1.3403298350824588,
      "grad_norm": 0.4750401042055502,
      "learning_rate": 2.4525634946552405e-06,
      "loss": 0.0293,
      "step": 894
    },
    {
      "epoch": 1.3418290854572714,
      "grad_norm": 0.7322489161594081,
      "learning_rate": 2.442438364617567e-06,
      "loss": 0.0332,
      "step": 895
    },
    {
      "epoch": 1.343328335832084,
      "grad_norm": 0.6566001641558517,
      "learning_rate": 2.4323274190786703e-06,
      "loss": 0.0299,
      "step": 896
    },
    {
      "epoch": 1.3448275862068966,
      "grad_norm": 1.8173940568940206,
      "learning_rate": 2.422230714114891e-06,
      "loss": 0.0394,
      "step": 897
    },
    {
      "epoch": 1.3463268365817092,
      "grad_norm": 0.6450110459896183,
      "learning_rate": 2.4121483057235884e-06,
      "loss": 0.0351,
      "step": 898
    },
    {
      "epoch": 1.3478260869565217,
      "grad_norm": 0.580708424572825,
      "learning_rate": 2.4020802498228333e-06,
      "loss": 0.0213,
      "step": 899
    },
    {
      "epoch": 1.3493253373313343,
      "grad_norm": 0.6831337203990496,
      "learning_rate": 2.392026602251093e-06,
      "loss": 0.0244,
      "step": 900
    },
    {
      "epoch": 1.3508245877061469,
      "grad_norm": 0.7102430150131334,
      "learning_rate": 2.3819874187669266e-06,
      "loss": 0.0332,
      "step": 901
    },
    {
      "epoch": 1.3523238380809595,
      "grad_norm": 0.7169914044107734,
      "learning_rate": 2.371962755048675e-06,
      "loss": 0.0336,
      "step": 902
    },
    {
      "epoch": 1.353823088455772,
      "grad_norm": 0.609552058608014,
      "learning_rate": 2.36195266669415e-06,
      "loss": 0.0325,
      "step": 903
    },
    {
      "epoch": 1.3553223388305846,
      "grad_norm": 0.5173827298182067,
      "learning_rate": 2.351957209220326e-06,
      "loss": 0.0192,
      "step": 904
    },
    {
      "epoch": 1.3568215892053974,
      "grad_norm": 0.5541614369727061,
      "learning_rate": 2.341976438063035e-06,
      "loss": 0.0231,
      "step": 905
    },
    {
      "epoch": 1.3583208395802098,
      "grad_norm": 0.49479940200247235,
      "learning_rate": 2.332010408576653e-06,
      "loss": 0.0225,
      "step": 906
    },
    {
      "epoch": 1.3598200899550226,
      "grad_norm": 0.5566281280831088,
      "learning_rate": 2.3220591760338046e-06,
      "loss": 0.0213,
      "step": 907
    },
    {
      "epoch": 1.3613193403298351,
      "grad_norm": 0.5031335082102869,
      "learning_rate": 2.3121227956250435e-06,
      "loss": 0.0242,
      "step": 908
    },
    {
      "epoch": 1.3628185907046477,
      "grad_norm": 0.552271865317701,
      "learning_rate": 2.302201322458552e-06,
      "loss": 0.0223,
      "step": 909
    },
    {
      "epoch": 1.3643178410794603,
      "grad_norm": 0.4103822285378797,
      "learning_rate": 2.292294811559837e-06,
      "loss": 0.0163,
      "step": 910
    },
    {
      "epoch": 1.3658170914542729,
      "grad_norm": 0.5965232999068361,
      "learning_rate": 2.282403317871422e-06,
      "loss": 0.0345,
      "step": 911
    },
    {
      "epoch": 1.3673163418290855,
      "grad_norm": 0.539620798319084,
      "learning_rate": 2.2725268962525454e-06,
      "loss": 0.0306,
      "step": 912
    },
    {
      "epoch": 1.368815592203898,
      "grad_norm": 0.5776729186757286,
      "learning_rate": 2.262665601478852e-06,
      "loss": 0.0295,
      "step": 913
    },
    {
      "epoch": 1.3703148425787106,
      "grad_norm": 0.4735746461829636,
      "learning_rate": 2.252819488242093e-06,
      "loss": 0.0205,
      "step": 914
    },
    {
      "epoch": 1.3718140929535232,
      "grad_norm": 0.5387419841683003,
      "learning_rate": 2.24298861114982e-06,
      "loss": 0.0192,
      "step": 915
    },
    {
      "epoch": 1.3733133433283358,
      "grad_norm": 0.5570725732827941,
      "learning_rate": 2.2331730247250857e-06,
      "loss": 0.0247,
      "step": 916
    },
    {
      "epoch": 1.3748125937031483,
      "grad_norm": 0.5428129898348611,
      "learning_rate": 2.223372783406137e-06,
      "loss": 0.0218,
      "step": 917
    },
    {
      "epoch": 1.3763118440779611,
      "grad_norm": 0.5839426400418757,
      "learning_rate": 2.2135879415461152e-06,
      "loss": 0.0237,
      "step": 918
    },
    {
      "epoch": 1.3778110944527735,
      "grad_norm": 0.7981563342450242,
      "learning_rate": 2.203818553412757e-06,
      "loss": 0.0317,
      "step": 919
    },
    {
      "epoch": 1.3793103448275863,
      "grad_norm": 0.46495158131915615,
      "learning_rate": 2.1940646731880887e-06,
      "loss": 0.0191,
      "step": 920
    },
    {
      "epoch": 1.3808095952023989,
      "grad_norm": 0.4554114850421928,
      "learning_rate": 2.1843263549681287e-06,
      "loss": 0.0194,
      "step": 921
    },
    {
      "epoch": 1.3823088455772115,
      "grad_norm": 0.48123874846666054,
      "learning_rate": 2.174603652762588e-06,
      "loss": 0.0246,
      "step": 922
    },
    {
      "epoch": 1.383808095952024,
      "grad_norm": 0.7535785466464365,
      "learning_rate": 2.164896620494569e-06,
      "loss": 0.0364,
      "step": 923
    },
    {
      "epoch": 1.3853073463268366,
      "grad_norm": 0.608221475268455,
      "learning_rate": 2.1552053120002655e-06,
      "loss": 0.022,
      "step": 924
    },
    {
      "epoch": 1.3868065967016492,
      "grad_norm": 0.6641284560735451,
      "learning_rate": 2.145529781028668e-06,
      "loss": 0.0273,
      "step": 925
    },
    {
      "epoch": 1.3883058470764618,
      "grad_norm": 0.4485831854913931,
      "learning_rate": 2.1358700812412625e-06,
      "loss": 0.0178,
      "step": 926
    },
    {
      "epoch": 1.3898050974512743,
      "grad_norm": 0.4128885523896202,
      "learning_rate": 2.1262262662117327e-06,
      "loss": 0.0163,
      "step": 927
    },
    {
      "epoch": 1.391304347826087,
      "grad_norm": 0.5784815676253148,
      "learning_rate": 2.1165983894256647e-06,
      "loss": 0.0217,
      "step": 928
    },
    {
      "epoch": 1.3928035982008995,
      "grad_norm": 0.6937149569445334,
      "learning_rate": 2.1069865042802502e-06,
      "loss": 0.0315,
      "step": 929
    },
    {
      "epoch": 1.394302848575712,
      "grad_norm": 0.5718759069213786,
      "learning_rate": 2.0973906640839867e-06,
      "loss": 0.0222,
      "step": 930
    },
    {
      "epoch": 1.3958020989505249,
      "grad_norm": 0.41479598543828794,
      "learning_rate": 2.0878109220563884e-06,
      "loss": 0.0184,
      "step": 931
    },
    {
      "epoch": 1.3973013493253372,
      "grad_norm": 0.5285351295730194,
      "learning_rate": 2.078247331327685e-06,
      "loss": 0.0187,
      "step": 932
    },
    {
      "epoch": 1.39880059970015,
      "grad_norm": 0.37363233417156,
      "learning_rate": 2.0686999449385286e-06,
      "loss": 0.0149,
      "step": 933
    },
    {
      "epoch": 1.4002998500749624,
      "grad_norm": 0.4548691489164575,
      "learning_rate": 2.0591688158397054e-06,
      "loss": 0.0202,
      "step": 934
    },
    {
      "epoch": 1.4017991004497752,
      "grad_norm": 0.48621995299860105,
      "learning_rate": 2.0496539968918342e-06,
      "loss": 0.0179,
      "step": 935
    },
    {
      "epoch": 1.4032983508245878,
      "grad_norm": 0.49805634928683473,
      "learning_rate": 2.0401555408650714e-06,
      "loss": 0.0217,
      "step": 936
    },
    {
      "epoch": 1.4047976011994003,
      "grad_norm": 0.4778467392350359,
      "learning_rate": 2.030673500438828e-06,
      "loss": 0.0173,
      "step": 937
    },
    {
      "epoch": 1.406296851574213,
      "grad_norm": 0.4690644914338906,
      "learning_rate": 2.0212079282014725e-06,
      "loss": 0.0197,
      "step": 938
    },
    {
      "epoch": 1.4077961019490255,
      "grad_norm": 0.4670167950720762,
      "learning_rate": 2.0117588766500375e-06,
      "loss": 0.0236,
      "step": 939
    },
    {
      "epoch": 1.409295352323838,
      "grad_norm": 0.6347494663581341,
      "learning_rate": 2.002326398189931e-06,
      "loss": 0.0342,
      "step": 940
    },
    {
      "epoch": 1.4107946026986506,
      "grad_norm": 0.49576435271730346,
      "learning_rate": 1.9929105451346436e-06,
      "loss": 0.0176,
      "step": 941
    },
    {
      "epoch": 1.4122938530734632,
      "grad_norm": 0.42041946309225303,
      "learning_rate": 1.983511369705462e-06,
      "loss": 0.0151,
      "step": 942
    },
    {
      "epoch": 1.4137931034482758,
      "grad_norm": 0.4821486255891003,
      "learning_rate": 1.9741289240311757e-06,
      "loss": 0.017,
      "step": 943
    },
    {
      "epoch": 1.4152923538230884,
      "grad_norm": 0.48765785646365206,
      "learning_rate": 1.9647632601477877e-06,
      "loss": 0.0204,
      "step": 944
    },
    {
      "epoch": 1.416791604197901,
      "grad_norm": 0.7553378692143973,
      "learning_rate": 1.9554144299982314e-06,
      "loss": 0.0323,
      "step": 945
    },
    {
      "epoch": 1.4182908545727138,
      "grad_norm": 0.6792778207766065,
      "learning_rate": 1.9460824854320755e-06,
      "loss": 0.0274,
      "step": 946
    },
    {
      "epoch": 1.419790104947526,
      "grad_norm": 0.5471795049965752,
      "learning_rate": 1.9367674782052376e-06,
      "loss": 0.0265,
      "step": 947
    },
    {
      "epoch": 1.421289355322339,
      "grad_norm": 0.5256250142070851,
      "learning_rate": 1.9274694599797067e-06,
      "loss": 0.0222,
      "step": 948
    },
    {
      "epoch": 1.4227886056971515,
      "grad_norm": 0.4969042243145794,
      "learning_rate": 1.918188482323242e-06,
      "loss": 0.0215,
      "step": 949
    },
    {
      "epoch": 1.424287856071964,
      "grad_norm": 0.5733940246998651,
      "learning_rate": 1.9089245967090952e-06,
      "loss": 0.0257,
      "step": 950
    },
    {
      "epoch": 1.4257871064467766,
      "grad_norm": 0.6954892553724012,
      "learning_rate": 1.8996778545157263e-06,
      "loss": 0.0256,
      "step": 951
    },
    {
      "epoch": 1.4272863568215892,
      "grad_norm": 0.5787020858165131,
      "learning_rate": 1.8904483070265133e-06,
      "loss": 0.0273,
      "step": 952
    },
    {
      "epoch": 1.4287856071964018,
      "grad_norm": 0.760775309287129,
      "learning_rate": 1.8812360054294725e-06,
      "loss": 0.0334,
      "step": 953
    },
    {
      "epoch": 1.4302848575712144,
      "grad_norm": 0.6127241776136112,
      "learning_rate": 1.8720410008169727e-06,
      "loss": 0.0274,
      "step": 954
    },
    {
      "epoch": 1.431784107946027,
      "grad_norm": 0.7183340550591089,
      "learning_rate": 1.8628633441854515e-06,
      "loss": 0.0383,
      "step": 955
    },
    {
      "epoch": 1.4332833583208395,
      "grad_norm": 0.5281358769888785,
      "learning_rate": 1.8537030864351303e-06,
      "loss": 0.0277,
      "step": 956
    },
    {
      "epoch": 1.434782608695652,
      "grad_norm": 0.5499094909788287,
      "learning_rate": 1.8445602783697375e-06,
      "loss": 0.0241,
      "step": 957
    },
    {
      "epoch": 1.4362818590704647,
      "grad_norm": 0.6804591929888588,
      "learning_rate": 1.8354349706962243e-06,
      "loss": 0.0298,
      "step": 958
    },
    {
      "epoch": 1.4377811094452775,
      "grad_norm": 0.44602568369920315,
      "learning_rate": 1.8263272140244803e-06,
      "loss": 0.0185,
      "step": 959
    },
    {
      "epoch": 1.4392803598200898,
      "grad_norm": 0.531198404238344,
      "learning_rate": 1.8172370588670563e-06,
      "loss": 0.0188,
      "step": 960
    },
    {
      "epoch": 1.4407796101949026,
      "grad_norm": 0.5629237463836722,
      "learning_rate": 1.8081645556388866e-06,
      "loss": 0.0271,
      "step": 961
    },
    {
      "epoch": 1.4422788605697152,
      "grad_norm": 0.3848239364636344,
      "learning_rate": 1.7991097546570018e-06,
      "loss": 0.0135,
      "step": 962
    },
    {
      "epoch": 1.4437781109445278,
      "grad_norm": 0.6203258664825868,
      "learning_rate": 1.7900727061402556e-06,
      "loss": 0.03,
      "step": 963
    },
    {
      "epoch": 1.4452773613193404,
      "grad_norm": 0.5626625646639414,
      "learning_rate": 1.7810534602090445e-06,
      "loss": 0.0178,
      "step": 964
    },
    {
      "epoch": 1.446776611694153,
      "grad_norm": 0.7532633468070113,
      "learning_rate": 1.77205206688503e-06,
      "loss": 0.0384,
      "step": 965
    },
    {
      "epoch": 1.4482758620689655,
      "grad_norm": 0.5749661156165156,
      "learning_rate": 1.7630685760908623e-06,
      "loss": 0.0265,
      "step": 966
    },
    {
      "epoch": 1.449775112443778,
      "grad_norm": 0.5345770940519479,
      "learning_rate": 1.7541030376499002e-06,
      "loss": 0.0218,
      "step": 967
    },
    {
      "epoch": 1.4512743628185907,
      "grad_norm": 0.48992681097845453,
      "learning_rate": 1.745155501285939e-06,
      "loss": 0.0202,
      "step": 968
    },
    {
      "epoch": 1.4527736131934033,
      "grad_norm": 0.5453136735741831,
      "learning_rate": 1.736226016622931e-06,
      "loss": 0.0261,
      "step": 969
    },
    {
      "epoch": 1.4542728635682158,
      "grad_norm": 0.5788941276671012,
      "learning_rate": 1.727314633184714e-06,
      "loss": 0.0249,
      "step": 970
    },
    {
      "epoch": 1.4557721139430284,
      "grad_norm": 0.5047750406749484,
      "learning_rate": 1.718421400394732e-06,
      "loss": 0.0223,
      "step": 971
    },
    {
      "epoch": 1.4572713643178412,
      "grad_norm": 0.7427728759904221,
      "learning_rate": 1.7095463675757656e-06,
      "loss": 0.0339,
      "step": 972
    },
    {
      "epoch": 1.4587706146926536,
      "grad_norm": 0.6307733516726286,
      "learning_rate": 1.7006895839496557e-06,
      "loss": 0.0329,
      "step": 973
    },
    {
      "epoch": 1.4602698650674664,
      "grad_norm": 0.49563644215308855,
      "learning_rate": 1.6918510986370312e-06,
      "loss": 0.02,
      "step": 974
    },
    {
      "epoch": 1.461769115442279,
      "grad_norm": 0.5172302696706871,
      "learning_rate": 1.6830309606570372e-06,
      "loss": 0.0249,
      "step": 975
    },
    {
      "epoch": 1.4632683658170915,
      "grad_norm": 0.46584522653274624,
      "learning_rate": 1.674229218927062e-06,
      "loss": 0.0182,
      "step": 976
    },
    {
      "epoch": 1.464767616191904,
      "grad_norm": 0.7310596011460893,
      "learning_rate": 1.665445922262467e-06,
      "loss": 0.0321,
      "step": 977
    },
    {
      "epoch": 1.4662668665667167,
      "grad_norm": 0.5617265654083822,
      "learning_rate": 1.6566811193763149e-06,
      "loss": 0.0234,
      "step": 978
    },
    {
      "epoch": 1.4677661169415293,
      "grad_norm": 0.731986552368286,
      "learning_rate": 1.6479348588791e-06,
      "loss": 0.0337,
      "step": 979
    },
    {
      "epoch": 1.4692653673163418,
      "grad_norm": 0.6342866571687072,
      "learning_rate": 1.6392071892784789e-06,
      "loss": 0.0304,
      "step": 980
    },
    {
      "epoch": 1.4707646176911544,
      "grad_norm": 0.47916328790339985,
      "learning_rate": 1.6304981589790015e-06,
      "loss": 0.0196,
      "step": 981
    },
    {
      "epoch": 1.472263868065967,
      "grad_norm": 0.6174063736500165,
      "learning_rate": 1.6218078162818418e-06,
      "loss": 0.025,
      "step": 982
    },
    {
      "epoch": 1.4737631184407796,
      "grad_norm": 0.6416771584097057,
      "learning_rate": 1.6131362093845299e-06,
      "loss": 0.0301,
      "step": 983
    },
    {
      "epoch": 1.4752623688155921,
      "grad_norm": 0.7131922070919802,
      "learning_rate": 1.6044833863806864e-06,
      "loss": 0.0327,
      "step": 984
    },
    {
      "epoch": 1.4767616191904047,
      "grad_norm": 0.4737416673407811,
      "learning_rate": 1.5958493952597536e-06,
      "loss": 0.0218,
      "step": 985
    },
    {
      "epoch": 1.4782608695652173,
      "grad_norm": 1.1221270829550933,
      "learning_rate": 1.5872342839067305e-06,
      "loss": 0.0284,
      "step": 986
    },
    {
      "epoch": 1.47976011994003,
      "grad_norm": 0.6042823556309873,
      "learning_rate": 1.5786381001019052e-06,
      "loss": 0.0264,
      "step": 987
    },
    {
      "epoch": 1.4812593703148424,
      "grad_norm": 0.5256796027328223,
      "learning_rate": 1.5700608915205978e-06,
      "loss": 0.0265,
      "step": 988
    },
    {
      "epoch": 1.4827586206896552,
      "grad_norm": 0.5757131381783417,
      "learning_rate": 1.561502705732883e-06,
      "loss": 0.0284,
      "step": 989
    },
    {
      "epoch": 1.4842578710644678,
      "grad_norm": 0.46802076903446693,
      "learning_rate": 1.5529635902033358e-06,
      "loss": 0.0173,
      "step": 990
    },
    {
      "epoch": 1.4857571214392804,
      "grad_norm": 0.7374213443664737,
      "learning_rate": 1.5444435922907669e-06,
      "loss": 0.0332,
      "step": 991
    },
    {
      "epoch": 1.487256371814093,
      "grad_norm": 0.6539469090292306,
      "learning_rate": 1.5359427592479553e-06,
      "loss": 0.0295,
      "step": 992
    },
    {
      "epoch": 1.4887556221889056,
      "grad_norm": 0.4671846876587741,
      "learning_rate": 1.5274611382213922e-06,
      "loss": 0.0207,
      "step": 993
    },
    {
      "epoch": 1.4902548725637181,
      "grad_norm": 0.5656165276862621,
      "learning_rate": 1.5189987762510167e-06,
      "loss": 0.023,
      "step": 994
    },
    {
      "epoch": 1.4917541229385307,
      "grad_norm": 0.5000740482395006,
      "learning_rate": 1.510555720269955e-06,
      "loss": 0.0209,
      "step": 995
    },
    {
      "epoch": 1.4932533733133433,
      "grad_norm": 0.4763362065474865,
      "learning_rate": 1.5021320171042608e-06,
      "loss": 0.0207,
      "step": 996
    },
    {
      "epoch": 1.4947526236881559,
      "grad_norm": 0.8009149894514453,
      "learning_rate": 1.4937277134726542e-06,
      "loss": 0.0401,
      "step": 997
    },
    {
      "epoch": 1.4962518740629684,
      "grad_norm": 0.4529288397995999,
      "learning_rate": 1.4853428559862637e-06,
      "loss": 0.0186,
      "step": 998
    },
    {
      "epoch": 1.497751124437781,
      "grad_norm": 0.5754600633983106,
      "learning_rate": 1.4769774911483686e-06,
      "loss": 0.0324,
      "step": 999
    },
    {
      "epoch": 1.4992503748125938,
      "grad_norm": 0.6571122952403181,
      "learning_rate": 1.4686316653541377e-06,
      "loss": 0.023,
      "step": 1000
    },
    {
      "epoch": 1.4992503748125938,
      "eval_loss": 0.1016751378774643,
      "eval_runtime": 2.1838,
      "eval_samples_per_second": 24.728,
      "eval_steps_per_second": 6.411,
      "step": 1000
    },
    {
      "epoch": 1.5007496251874062,
      "grad_norm": 0.6695813342550073,
      "learning_rate": 1.4603054248903752e-06,
      "loss": 0.0368,
      "step": 1001
    },
    {
      "epoch": 1.502248875562219,
      "grad_norm": 0.49846843102140553,
      "learning_rate": 1.4519988159352665e-06,
      "loss": 0.0181,
      "step": 1002
    },
    {
      "epoch": 1.5037481259370313,
      "grad_norm": 0.8875237144950341,
      "learning_rate": 1.4437118845581138e-06,
      "loss": 0.0412,
      "step": 1003
    },
    {
      "epoch": 1.5052473763118441,
      "grad_norm": 0.6391910554411541,
      "learning_rate": 1.4354446767190873e-06,
      "loss": 0.0277,
      "step": 1004
    },
    {
      "epoch": 1.5067466266866567,
      "grad_norm": 0.7353944206484456,
      "learning_rate": 1.4271972382689685e-06,
      "loss": 0.0345,
      "step": 1005
    },
    {
      "epoch": 1.5082458770614693,
      "grad_norm": 0.6537036038774572,
      "learning_rate": 1.4189696149488956e-06,
      "loss": 0.0256,
      "step": 1006
    },
    {
      "epoch": 1.5097451274362819,
      "grad_norm": 0.6239098192932826,
      "learning_rate": 1.4107618523901101e-06,
      "loss": 0.0206,
      "step": 1007
    },
    {
      "epoch": 1.5112443778110944,
      "grad_norm": 0.45298373933576136,
      "learning_rate": 1.4025739961137043e-06,
      "loss": 0.0196,
      "step": 1008
    },
    {
      "epoch": 1.512743628185907,
      "grad_norm": 0.5004709636487935,
      "learning_rate": 1.394406091530367e-06,
      "loss": 0.0172,
      "step": 1009
    },
    {
      "epoch": 1.5142428785607196,
      "grad_norm": 0.5167361199123685,
      "learning_rate": 1.3862581839401346e-06,
      "loss": 0.0207,
      "step": 1010
    },
    {
      "epoch": 1.5157421289355324,
      "grad_norm": 0.6030090809091007,
      "learning_rate": 1.3781303185321377e-06,
      "loss": 0.0175,
      "step": 1011
    },
    {
      "epoch": 1.5172413793103448,
      "grad_norm": 0.5984433992602783,
      "learning_rate": 1.370022540384347e-06,
      "loss": 0.0225,
      "step": 1012
    },
    {
      "epoch": 1.5187406296851576,
      "grad_norm": 0.42400906485734946,
      "learning_rate": 1.3619348944633331e-06,
      "loss": 0.018,
      "step": 1013
    },
    {
      "epoch": 1.52023988005997,
      "grad_norm": 0.5996640005884758,
      "learning_rate": 1.3538674256240087e-06,
      "loss": 0.0262,
      "step": 1014
    },
    {
      "epoch": 1.5217391304347827,
      "grad_norm": 0.4643108602815457,
      "learning_rate": 1.3458201786093795e-06,
      "loss": 0.0167,
      "step": 1015
    },
    {
      "epoch": 1.523238380809595,
      "grad_norm": 0.6978759480256204,
      "learning_rate": 1.3377931980503055e-06,
      "loss": 0.0283,
      "step": 1016
    },
    {
      "epoch": 1.5247376311844079,
      "grad_norm": 0.41833365334980244,
      "learning_rate": 1.3297865284652417e-06,
      "loss": 0.0165,
      "step": 1017
    },
    {
      "epoch": 1.5262368815592204,
      "grad_norm": 0.5650411741047004,
      "learning_rate": 1.3218002142599973e-06,
      "loss": 0.0298,
      "step": 1018
    },
    {
      "epoch": 1.527736131934033,
      "grad_norm": 0.5045218607588701,
      "learning_rate": 1.3138342997274883e-06,
      "loss": 0.0241,
      "step": 1019
    },
    {
      "epoch": 1.5292353823088456,
      "grad_norm": 0.49823224767799534,
      "learning_rate": 1.3058888290474937e-06,
      "loss": 0.0234,
      "step": 1020
    },
    {
      "epoch": 1.5307346326836582,
      "grad_norm": 0.5354560074534213,
      "learning_rate": 1.2979638462864069e-06,
      "loss": 0.021,
      "step": 1021
    },
    {
      "epoch": 1.5322338830584707,
      "grad_norm": 0.47411679424520514,
      "learning_rate": 1.2900593953969947e-06,
      "loss": 0.0164,
      "step": 1022
    },
    {
      "epoch": 1.5337331334332833,
      "grad_norm": 0.46748107076007084,
      "learning_rate": 1.2821755202181503e-06,
      "loss": 0.016,
      "step": 1023
    },
    {
      "epoch": 1.535232383808096,
      "grad_norm": 0.4211649112854803,
      "learning_rate": 1.2743122644746536e-06,
      "loss": 0.0142,
      "step": 1024
    },
    {
      "epoch": 1.5367316341829085,
      "grad_norm": 0.6383326025032485,
      "learning_rate": 1.266469671776926e-06,
      "loss": 0.0328,
      "step": 1025
    },
    {
      "epoch": 1.5382308845577213,
      "grad_norm": 0.6514466360092246,
      "learning_rate": 1.2586477856207902e-06,
      "loss": 0.0279,
      "step": 1026
    },
    {
      "epoch": 1.5397301349325336,
      "grad_norm": 0.5129260829751254,
      "learning_rate": 1.2508466493872273e-06,
      "loss": 0.0233,
      "step": 1027
    },
    {
      "epoch": 1.5412293853073464,
      "grad_norm": 0.7760283966990239,
      "learning_rate": 1.2430663063421388e-06,
      "loss": 0.0405,
      "step": 1028
    },
    {
      "epoch": 1.5427286356821588,
      "grad_norm": 0.6938414003347936,
      "learning_rate": 1.2353067996361034e-06,
      "loss": 0.0332,
      "step": 1029
    },
    {
      "epoch": 1.5442278860569716,
      "grad_norm": 0.733298992887001,
      "learning_rate": 1.2275681723041406e-06,
      "loss": 0.0366,
      "step": 1030
    },
    {
      "epoch": 1.545727136431784,
      "grad_norm": 0.5286810410610624,
      "learning_rate": 1.2198504672654694e-06,
      "loss": 0.0209,
      "step": 1031
    },
    {
      "epoch": 1.5472263868065967,
      "grad_norm": 0.4105304572330887,
      "learning_rate": 1.212153727323273e-06,
      "loss": 0.0187,
      "step": 1032
    },
    {
      "epoch": 1.5487256371814093,
      "grad_norm": 0.5456198591036658,
      "learning_rate": 1.2044779951644586e-06,
      "loss": 0.0284,
      "step": 1033
    },
    {
      "epoch": 1.550224887556222,
      "grad_norm": 0.7539861878774872,
      "learning_rate": 1.1968233133594243e-06,
      "loss": 0.0346,
      "step": 1034
    },
    {
      "epoch": 1.5517241379310345,
      "grad_norm": 0.47029136480947925,
      "learning_rate": 1.1891897243618184e-06,
      "loss": 0.022,
      "step": 1035
    },
    {
      "epoch": 1.553223388305847,
      "grad_norm": 0.5135809664814628,
      "learning_rate": 1.1815772705083072e-06,
      "loss": 0.0187,
      "step": 1036
    },
    {
      "epoch": 1.5547226386806596,
      "grad_norm": 0.4580541600964491,
      "learning_rate": 1.17398599401834e-06,
      "loss": 0.0174,
      "step": 1037
    },
    {
      "epoch": 1.5562218890554722,
      "grad_norm": 0.6667746655359075,
      "learning_rate": 1.1664159369939137e-06,
      "loss": 0.0292,
      "step": 1038
    },
    {
      "epoch": 1.557721139430285,
      "grad_norm": 0.6712373239244093,
      "learning_rate": 1.1588671414193397e-06,
      "loss": 0.0314,
      "step": 1039
    },
    {
      "epoch": 1.5592203898050974,
      "grad_norm": 0.49335767248651224,
      "learning_rate": 1.1513396491610113e-06,
      "loss": 0.0242,
      "step": 1040
    },
    {
      "epoch": 1.5607196401799102,
      "grad_norm": 0.6159560382155416,
      "learning_rate": 1.1438335019671715e-06,
      "loss": 0.0246,
      "step": 1041
    },
    {
      "epoch": 1.5622188905547225,
      "grad_norm": 0.7928670993287587,
      "learning_rate": 1.1363487414676805e-06,
      "loss": 0.0333,
      "step": 1042
    },
    {
      "epoch": 1.5637181409295353,
      "grad_norm": 0.5466607572295086,
      "learning_rate": 1.128885409173789e-06,
      "loss": 0.024,
      "step": 1043
    },
    {
      "epoch": 1.5652173913043477,
      "grad_norm": 0.5188003848450965,
      "learning_rate": 1.1214435464779006e-06,
      "loss": 0.0226,
      "step": 1044
    },
    {
      "epoch": 1.5667166416791605,
      "grad_norm": 0.47136560838493624,
      "learning_rate": 1.1140231946533486e-06,
      "loss": 0.0187,
      "step": 1045
    },
    {
      "epoch": 1.568215892053973,
      "grad_norm": 0.5437108889091811,
      "learning_rate": 1.1066243948541638e-06,
      "loss": 0.0256,
      "step": 1046
    },
    {
      "epoch": 1.5697151424287856,
      "grad_norm": 0.434782205074127,
      "learning_rate": 1.0992471881148497e-06,
      "loss": 0.0152,
      "step": 1047
    },
    {
      "epoch": 1.5712143928035982,
      "grad_norm": 0.6890258963485254,
      "learning_rate": 1.091891615350147e-06,
      "loss": 0.0249,
      "step": 1048
    },
    {
      "epoch": 1.5727136431784108,
      "grad_norm": 0.7936469658858025,
      "learning_rate": 1.0845577173548172e-06,
      "loss": 0.0365,
      "step": 1049
    },
    {
      "epoch": 1.5742128935532234,
      "grad_norm": 0.3952859017065396,
      "learning_rate": 1.07724553480341e-06,
      "loss": 0.0149,
      "step": 1050
    },
    {
      "epoch": 1.575712143928036,
      "grad_norm": 0.635766560306531,
      "learning_rate": 1.0699551082500387e-06,
      "loss": 0.0213,
      "step": 1051
    },
    {
      "epoch": 1.5772113943028487,
      "grad_norm": 0.6766173196370536,
      "learning_rate": 1.0626864781281553e-06,
      "loss": 0.0238,
      "step": 1052
    },
    {
      "epoch": 1.578710644677661,
      "grad_norm": 0.48782393715749717,
      "learning_rate": 1.0554396847503272e-06,
      "loss": 0.0173,
      "step": 1053
    },
    {
      "epoch": 1.580209895052474,
      "grad_norm": 0.46888602166502774,
      "learning_rate": 1.0482147683080125e-06,
      "loss": 0.0161,
      "step": 1054
    },
    {
      "epoch": 1.5817091454272862,
      "grad_norm": 0.6101834267007092,
      "learning_rate": 1.0410117688713366e-06,
      "loss": 0.0266,
      "step": 1055
    },
    {
      "epoch": 1.583208395802099,
      "grad_norm": 0.5989382558689423,
      "learning_rate": 1.0338307263888748e-06,
      "loss": 0.0198,
      "step": 1056
    },
    {
      "epoch": 1.5847076461769114,
      "grad_norm": 0.5219929816940649,
      "learning_rate": 1.0266716806874227e-06,
      "loss": 0.0244,
      "step": 1057
    },
    {
      "epoch": 1.5862068965517242,
      "grad_norm": 0.4983267283810742,
      "learning_rate": 1.0195346714717813e-06,
      "loss": 0.0225,
      "step": 1058
    },
    {
      "epoch": 1.5877061469265368,
      "grad_norm": 0.6412123787823575,
      "learning_rate": 1.0124197383245344e-06,
      "loss": 0.0241,
      "step": 1059
    },
    {
      "epoch": 1.5892053973013494,
      "grad_norm": 0.4722790454204881,
      "learning_rate": 1.0053269207058298e-06,
      "loss": 0.0189,
      "step": 1060
    },
    {
      "epoch": 1.590704647676162,
      "grad_norm": 0.45291994717318196,
      "learning_rate": 9.982562579531607e-07,
      "loss": 0.0209,
      "step": 1061
    },
    {
      "epoch": 1.5922038980509745,
      "grad_norm": 0.43826180965430594,
      "learning_rate": 9.912077892811473e-07,
      "loss": 0.0144,
      "step": 1062
    },
    {
      "epoch": 1.593703148425787,
      "grad_norm": 0.5696212570529847,
      "learning_rate": 9.841815537813177e-07,
      "loss": 0.0239,
      "step": 1063
    },
    {
      "epoch": 1.5952023988005997,
      "grad_norm": 0.570142613704561,
      "learning_rate": 9.77177590421895e-07,
      "loss": 0.0214,
      "step": 1064
    },
    {
      "epoch": 1.5967016491754122,
      "grad_norm": 0.8979279987899056,
      "learning_rate": 9.70195938047576e-07,
      "loss": 0.0348,
      "step": 1065
    },
    {
      "epoch": 1.5982008995502248,
      "grad_norm": 0.5071607972445688,
      "learning_rate": 9.63236635379321e-07,
      "loss": 0.0169,
      "step": 1066
    },
    {
      "epoch": 1.5997001499250376,
      "grad_norm": 0.590332280287046,
      "learning_rate": 9.562997210141355e-07,
      "loss": 0.028,
      "step": 1067
    },
    {
      "epoch": 1.60119940029985,
      "grad_norm": 0.6082219732250134,
      "learning_rate": 9.49385233424856e-07,
      "loss": 0.03,
      "step": 1068
    },
    {
      "epoch": 1.6026986506746628,
      "grad_norm": 0.6764179666089184,
      "learning_rate": 9.424932109599372e-07,
      "loss": 0.0282,
      "step": 1069
    },
    {
      "epoch": 1.6041979010494751,
      "grad_norm": 0.6419749243063373,
      "learning_rate": 9.356236918432454e-07,
      "loss": 0.0237,
      "step": 1070
    },
    {
      "epoch": 1.605697151424288,
      "grad_norm": 0.6079565270634396,
      "learning_rate": 9.287767141738352e-07,
      "loss": 0.0217,
      "step": 1071
    },
    {
      "epoch": 1.6071964017991005,
      "grad_norm": 0.5266769435111732,
      "learning_rate": 9.21952315925746e-07,
      "loss": 0.0263,
      "step": 1072
    },
    {
      "epoch": 1.608695652173913,
      "grad_norm": 0.5488411017791768,
      "learning_rate": 9.151505349477901e-07,
      "loss": 0.0244,
      "step": 1073
    },
    {
      "epoch": 1.6101949025487257,
      "grad_norm": 0.7236518144784216,
      "learning_rate": 9.08371408963341e-07,
      "loss": 0.0379,
      "step": 1074
    },
    {
      "epoch": 1.6116941529235382,
      "grad_norm": 0.6840408613627699,
      "learning_rate": 9.016149755701259e-07,
      "loss": 0.033,
      "step": 1075
    },
    {
      "epoch": 1.6131934032983508,
      "grad_norm": 0.5667837980214311,
      "learning_rate": 8.948812722400157e-07,
      "loss": 0.022,
      "step": 1076
    },
    {
      "epoch": 1.6146926536731634,
      "grad_norm": 0.7741134751054364,
      "learning_rate": 8.881703363188199e-07,
      "loss": 0.043,
      "step": 1077
    },
    {
      "epoch": 1.616191904047976,
      "grad_norm": 0.39376383014129646,
      "learning_rate": 8.814822050260758e-07,
      "loss": 0.0162,
      "step": 1078
    },
    {
      "epoch": 1.6176911544227885,
      "grad_norm": 0.6858350404743934,
      "learning_rate": 8.748169154548448e-07,
      "loss": 0.0335,
      "step": 1079
    },
    {
      "epoch": 1.6191904047976013,
      "grad_norm": 0.5823431074807772,
      "learning_rate": 8.681745045715045e-07,
      "loss": 0.0299,
      "step": 1080
    },
    {
      "epoch": 1.6206896551724137,
      "grad_norm": 0.5128460654872932,
      "learning_rate": 8.615550092155478e-07,
      "loss": 0.0266,
      "step": 1081
    },
    {
      "epoch": 1.6221889055472265,
      "grad_norm": 0.5117643117342252,
      "learning_rate": 8.549584660993726e-07,
      "loss": 0.0221,
      "step": 1082
    },
    {
      "epoch": 1.6236881559220389,
      "grad_norm": 0.6372351435625767,
      "learning_rate": 8.483849118080828e-07,
      "loss": 0.0253,
      "step": 1083
    },
    {
      "epoch": 1.6251874062968517,
      "grad_norm": 0.5926828078773826,
      "learning_rate": 8.418343827992842e-07,
      "loss": 0.0238,
      "step": 1084
    },
    {
      "epoch": 1.626686656671664,
      "grad_norm": 0.8383252566451661,
      "learning_rate": 8.353069154028814e-07,
      "loss": 0.0402,
      "step": 1085
    },
    {
      "epoch": 1.6281859070464768,
      "grad_norm": 0.6315572568666615,
      "learning_rate": 8.28802545820877e-07,
      "loss": 0.0371,
      "step": 1086
    },
    {
      "epoch": 1.6296851574212894,
      "grad_norm": 0.6816024172232844,
      "learning_rate": 8.223213101271709e-07,
      "loss": 0.0255,
      "step": 1087
    },
    {
      "epoch": 1.631184407796102,
      "grad_norm": 0.5762412489110237,
      "learning_rate": 8.158632442673603e-07,
      "loss": 0.0282,
      "step": 1088
    },
    {
      "epoch": 1.6326836581709145,
      "grad_norm": 0.5959387183813116,
      "learning_rate": 8.094283840585398e-07,
      "loss": 0.0279,
      "step": 1089
    },
    {
      "epoch": 1.6341829085457271,
      "grad_norm": 0.6125157767131403,
      "learning_rate": 8.03016765189103e-07,
      "loss": 0.0265,
      "step": 1090
    },
    {
      "epoch": 1.6356821589205397,
      "grad_norm": 0.540434351436337,
      "learning_rate": 7.966284232185451e-07,
      "loss": 0.0209,
      "step": 1091
    },
    {
      "epoch": 1.6371814092953523,
      "grad_norm": 0.5823102323403485,
      "learning_rate": 7.902633935772647e-07,
      "loss": 0.026,
      "step": 1092
    },
    {
      "epoch": 1.638680659670165,
      "grad_norm": 0.7602071035318266,
      "learning_rate": 7.839217115663683e-07,
      "loss": 0.0418,
      "step": 1093
    },
    {
      "epoch": 1.6401799100449774,
      "grad_norm": 0.4906041129596735,
      "learning_rate": 7.776034123574738e-07,
      "loss": 0.028,
      "step": 1094
    },
    {
      "epoch": 1.6416791604197902,
      "grad_norm": 0.5260916485948183,
      "learning_rate": 7.713085309925156e-07,
      "loss": 0.0205,
      "step": 1095
    },
    {
      "epoch": 1.6431784107946026,
      "grad_norm": 0.6954092026586055,
      "learning_rate": 7.650371023835495e-07,
      "loss": 0.0297,
      "step": 1096
    },
    {
      "epoch": 1.6446776611694154,
      "grad_norm": 0.5829643188372207,
      "learning_rate": 7.587891613125631e-07,
      "loss": 0.0253,
      "step": 1097
    },
    {
      "epoch": 1.6461769115442277,
      "grad_norm": 0.518386860381134,
      "learning_rate": 7.525647424312766e-07,
      "loss": 0.0238,
      "step": 1098
    },
    {
      "epoch": 1.6476761619190405,
      "grad_norm": 0.5501244998959788,
      "learning_rate": 7.46363880260954e-07,
      "loss": 0.0246,
      "step": 1099
    },
    {
      "epoch": 1.6491754122938531,
      "grad_norm": 0.5800040963232465,
      "learning_rate": 7.401866091922133e-07,
      "loss": 0.0236,
      "step": 1100
    },
    {
      "epoch": 1.6506746626686657,
      "grad_norm": 0.7515558330109627,
      "learning_rate": 7.340329634848309e-07,
      "loss": 0.0373,
      "step": 1101
    },
    {
      "epoch": 1.6521739130434783,
      "grad_norm": 0.7000031037706768,
      "learning_rate": 7.279029772675572e-07,
      "loss": 0.0284,
      "step": 1102
    },
    {
      "epoch": 1.6536731634182908,
      "grad_norm": 0.5524552048351863,
      "learning_rate": 7.217966845379243e-07,
      "loss": 0.0234,
      "step": 1103
    },
    {
      "epoch": 1.6551724137931034,
      "grad_norm": 0.4057591363218374,
      "learning_rate": 7.157141191620548e-07,
      "loss": 0.0126,
      "step": 1104
    },
    {
      "epoch": 1.656671664167916,
      "grad_norm": 0.43385669820434114,
      "learning_rate": 7.096553148744806e-07,
      "loss": 0.0152,
      "step": 1105
    },
    {
      "epoch": 1.6581709145427288,
      "grad_norm": 0.5613659518653189,
      "learning_rate": 7.036203052779506e-07,
      "loss": 0.024,
      "step": 1106
    },
    {
      "epoch": 1.6596701649175412,
      "grad_norm": 0.4660136519667188,
      "learning_rate": 6.97609123843247e-07,
      "loss": 0.0164,
      "step": 1107
    },
    {
      "epoch": 1.661169415292354,
      "grad_norm": 0.7502578473847489,
      "learning_rate": 6.916218039089961e-07,
      "loss": 0.0257,
      "step": 1108
    },
    {
      "epoch": 1.6626686656671663,
      "grad_norm": 0.604165388467696,
      "learning_rate": 6.856583786814891e-07,
      "loss": 0.0252,
      "step": 1109
    },
    {
      "epoch": 1.6641679160419791,
      "grad_norm": 0.6380763082516621,
      "learning_rate": 6.797188812344907e-07,
      "loss": 0.0207,
      "step": 1110
    },
    {
      "epoch": 1.6656671664167915,
      "grad_norm": 0.6485768231430634,
      "learning_rate": 6.738033445090653e-07,
      "loss": 0.0325,
      "step": 1111
    },
    {
      "epoch": 1.6671664167916043,
      "grad_norm": 0.6136423723423484,
      "learning_rate": 6.67911801313384e-07,
      "loss": 0.0259,
      "step": 1112
    },
    {
      "epoch": 1.6686656671664168,
      "grad_norm": 0.5970162286096742,
      "learning_rate": 6.620442843225483e-07,
      "loss": 0.0287,
      "step": 1113
    },
    {
      "epoch": 1.6701649175412294,
      "grad_norm": 0.4552599207459205,
      "learning_rate": 6.562008260784092e-07,
      "loss": 0.0163,
      "step": 1114
    },
    {
      "epoch": 1.671664167916042,
      "grad_norm": 0.4523480798019963,
      "learning_rate": 6.503814589893836e-07,
      "loss": 0.0141,
      "step": 1115
    },
    {
      "epoch": 1.6731634182908546,
      "grad_norm": 0.5525051639383304,
      "learning_rate": 6.445862153302784e-07,
      "loss": 0.024,
      "step": 1116
    },
    {
      "epoch": 1.6746626686656672,
      "grad_norm": 0.5927208302302158,
      "learning_rate": 6.388151272421078e-07,
      "loss": 0.0254,
      "step": 1117
    },
    {
      "epoch": 1.6761619190404797,
      "grad_norm": 0.40097239873608587,
      "learning_rate": 6.330682267319177e-07,
      "loss": 0.014,
      "step": 1118
    },
    {
      "epoch": 1.6776611694152923,
      "grad_norm": 0.6414516801325132,
      "learning_rate": 6.273455456726074e-07,
      "loss": 0.0241,
      "step": 1119
    },
    {
      "epoch": 1.6791604197901049,
      "grad_norm": 0.5330902523835881,
      "learning_rate": 6.216471158027515e-07,
      "loss": 0.0279,
      "step": 1120
    },
    {
      "epoch": 1.6806596701649177,
      "grad_norm": 0.420065449725033,
      "learning_rate": 6.159729687264254e-07,
      "loss": 0.0127,
      "step": 1121
    },
    {
      "epoch": 1.68215892053973,
      "grad_norm": 0.8490695243130283,
      "learning_rate": 6.103231359130308e-07,
      "loss": 0.0259,
      "step": 1122
    },
    {
      "epoch": 1.6836581709145428,
      "grad_norm": 0.656779730596281,
      "learning_rate": 6.046976486971201e-07,
      "loss": 0.0277,
      "step": 1123
    },
    {
      "epoch": 1.6851574212893552,
      "grad_norm": 0.7273342007096357,
      "learning_rate": 5.990965382782177e-07,
      "loss": 0.0277,
      "step": 1124
    },
    {
      "epoch": 1.686656671664168,
      "grad_norm": 0.6408475874003648,
      "learning_rate": 5.935198357206595e-07,
      "loss": 0.0268,
      "step": 1125
    },
    {
      "epoch": 1.6881559220389803,
      "grad_norm": 0.5894499256200443,
      "learning_rate": 5.879675719534078e-07,
      "loss": 0.0268,
      "step": 1126
    },
    {
      "epoch": 1.6896551724137931,
      "grad_norm": 0.5941827331951004,
      "learning_rate": 5.824397777698859e-07,
      "loss": 0.0258,
      "step": 1127
    },
    {
      "epoch": 1.6911544227886057,
      "grad_norm": 0.5856578408857893,
      "learning_rate": 5.769364838278063e-07,
      "loss": 0.0211,
      "step": 1128
    },
    {
      "epoch": 1.6926536731634183,
      "grad_norm": 0.5899393594258006,
      "learning_rate": 5.714577206490018e-07,
      "loss": 0.0253,
      "step": 1129
    },
    {
      "epoch": 1.6941529235382309,
      "grad_norm": 0.4590048370790555,
      "learning_rate": 5.660035186192531e-07,
      "loss": 0.0183,
      "step": 1130
    },
    {
      "epoch": 1.6956521739130435,
      "grad_norm": 0.5983970095854699,
      "learning_rate": 5.60573907988124e-07,
      "loss": 0.0266,
      "step": 1131
    },
    {
      "epoch": 1.697151424287856,
      "grad_norm": 0.8457173790907914,
      "learning_rate": 5.551689188687909e-07,
      "loss": 0.0354,
      "step": 1132
    },
    {
      "epoch": 1.6986506746626686,
      "grad_norm": 0.46781559923295196,
      "learning_rate": 5.497885812378772e-07,
      "loss": 0.0225,
      "step": 1133
    },
    {
      "epoch": 1.7001499250374814,
      "grad_norm": 0.7482541322105855,
      "learning_rate": 5.444329249352859e-07,
      "loss": 0.0381,
      "step": 1134
    },
    {
      "epoch": 1.7016491754122938,
      "grad_norm": 0.725483149408877,
      "learning_rate": 5.391019796640362e-07,
      "loss": 0.0247,
      "step": 1135
    },
    {
      "epoch": 1.7031484257871066,
      "grad_norm": 0.7345000796422639,
      "learning_rate": 5.337957749900958e-07,
      "loss": 0.0271,
      "step": 1136
    },
    {
      "epoch": 1.704647676161919,
      "grad_norm": 0.41901449000663044,
      "learning_rate": 5.285143403422188e-07,
      "loss": 0.0155,
      "step": 1137
    },
    {
      "epoch": 1.7061469265367317,
      "grad_norm": 0.44415921179139134,
      "learning_rate": 5.23257705011786e-07,
      "loss": 0.0175,
      "step": 1138
    },
    {
      "epoch": 1.707646176911544,
      "grad_norm": 0.5614715259357606,
      "learning_rate": 5.18025898152631e-07,
      "loss": 0.0196,
      "step": 1139
    },
    {
      "epoch": 1.7091454272863569,
      "grad_norm": 0.4492947242398021,
      "learning_rate": 5.128189487808927e-07,
      "loss": 0.0111,
      "step": 1140
    },
    {
      "epoch": 1.7106446776611695,
      "grad_norm": 0.6971793252447149,
      "learning_rate": 5.076368857748454e-07,
      "loss": 0.0304,
      "step": 1141
    },
    {
      "epoch": 1.712143928035982,
      "grad_norm": 0.7677839065259273,
      "learning_rate": 5.024797378747414e-07,
      "loss": 0.03,
      "step": 1142
    },
    {
      "epoch": 1.7136431784107946,
      "grad_norm": 0.5760725096633648,
      "learning_rate": 4.973475336826506e-07,
      "loss": 0.0205,
      "step": 1143
    },
    {
      "epoch": 1.7151424287856072,
      "grad_norm": 0.49927254427148077,
      "learning_rate": 4.922403016623034e-07,
      "loss": 0.0247,
      "step": 1144
    },
    {
      "epoch": 1.7166416791604198,
      "grad_norm": 0.5472580079655377,
      "learning_rate": 4.871580701389316e-07,
      "loss": 0.0226,
      "step": 1145
    },
    {
      "epoch": 1.7181409295352323,
      "grad_norm": 0.6437732250206993,
      "learning_rate": 4.821008672991118e-07,
      "loss": 0.033,
      "step": 1146
    },
    {
      "epoch": 1.7196401799100451,
      "grad_norm": 0.5417881610907915,
      "learning_rate": 4.770687211906089e-07,
      "loss": 0.0238,
      "step": 1147
    },
    {
      "epoch": 1.7211394302848575,
      "grad_norm": 0.6172471339425802,
      "learning_rate": 4.720616597222205e-07,
      "loss": 0.0292,
      "step": 1148
    },
    {
      "epoch": 1.7226386806596703,
      "grad_norm": 0.44440843075530234,
      "learning_rate": 4.6707971066362324e-07,
      "loss": 0.0163,
      "step": 1149
    },
    {
      "epoch": 1.7241379310344827,
      "grad_norm": 0.6731531132436804,
      "learning_rate": 4.6212290164521554e-07,
      "loss": 0.0264,
      "step": 1150
    },
    {
      "epoch": 1.7256371814092955,
      "grad_norm": 0.41215708960893366,
      "learning_rate": 4.5719126015796757e-07,
      "loss": 0.015,
      "step": 1151
    },
    {
      "epoch": 1.7271364317841078,
      "grad_norm": 0.6347209856474826,
      "learning_rate": 4.522848135532698e-07,
      "loss": 0.0233,
      "step": 1152
    },
    {
      "epoch": 1.7286356821589206,
      "grad_norm": 0.5738521943745555,
      "learning_rate": 4.474035890427769e-07,
      "loss": 0.0245,
      "step": 1153
    },
    {
      "epoch": 1.7301349325337332,
      "grad_norm": 0.6816287269367703,
      "learning_rate": 4.4254761369825984e-07,
      "loss": 0.0216,
      "step": 1154
    },
    {
      "epoch": 1.7316341829085458,
      "grad_norm": 0.44195789202002556,
      "learning_rate": 4.377169144514554e-07,
      "loss": 0.0142,
      "step": 1155
    },
    {
      "epoch": 1.7331334332833583,
      "grad_norm": 0.602138007682664,
      "learning_rate": 4.329115180939164e-07,
      "loss": 0.0246,
      "step": 1156
    },
    {
      "epoch": 1.734632683658171,
      "grad_norm": 0.5628633692607186,
      "learning_rate": 4.281314512768625e-07,
      "loss": 0.0232,
      "step": 1157
    },
    {
      "epoch": 1.7361319340329835,
      "grad_norm": 0.7400669393864304,
      "learning_rate": 4.2337674051103504e-07,
      "loss": 0.0326,
      "step": 1158
    },
    {
      "epoch": 1.737631184407796,
      "grad_norm": 0.6681427912103419,
      "learning_rate": 4.186474121665468e-07,
      "loss": 0.0306,
      "step": 1159
    },
    {
      "epoch": 1.7391304347826086,
      "grad_norm": 0.3735223568673427,
      "learning_rate": 4.139434924727359e-07,
      "loss": 0.0158,
      "step": 1160
    },
    {
      "epoch": 1.7406296851574212,
      "grad_norm": 0.45570172329357683,
      "learning_rate": 4.092650075180232e-07,
      "loss": 0.0178,
      "step": 1161
    },
    {
      "epoch": 1.742128935532234,
      "grad_norm": 0.4601004031855763,
      "learning_rate": 4.046119832497658e-07,
      "loss": 0.0189,
      "step": 1162
    },
    {
      "epoch": 1.7436281859070464,
      "grad_norm": 0.5512892492529917,
      "learning_rate": 3.9998444547411255e-07,
      "loss": 0.0246,
      "step": 1163
    },
    {
      "epoch": 1.7451274362818592,
      "grad_norm": 0.5568635326520949,
      "learning_rate": 3.9538241985586144e-07,
      "loss": 0.0256,
      "step": 1164
    },
    {
      "epoch": 1.7466266866566715,
      "grad_norm": 0.6698831186444958,
      "learning_rate": 3.908059319183194e-07,
      "loss": 0.033,
      "step": 1165
    },
    {
      "epoch": 1.7481259370314843,
      "grad_norm": 0.457640379216394,
      "learning_rate": 3.8625500704315645e-07,
      "loss": 0.0197,
      "step": 1166
    },
    {
      "epoch": 1.7496251874062967,
      "grad_norm": 0.6520546474983886,
      "learning_rate": 3.8172967047026834e-07,
      "loss": 0.0285,
      "step": 1167
    },
    {
      "epoch": 1.7511244377811095,
      "grad_norm": 0.5171778163390587,
      "learning_rate": 3.7722994729763427e-07,
      "loss": 0.0243,
      "step": 1168
    },
    {
      "epoch": 1.752623688155922,
      "grad_norm": 0.617124456912294,
      "learning_rate": 3.7275586248118114e-07,
      "loss": 0.031,
      "step": 1169
    },
    {
      "epoch": 1.7541229385307346,
      "grad_norm": 0.5076278498725939,
      "learning_rate": 3.683074408346404e-07,
      "loss": 0.0209,
      "step": 1170
    },
    {
      "epoch": 1.7556221889055472,
      "grad_norm": 0.5362181566697122,
      "learning_rate": 3.6388470702941436e-07,
      "loss": 0.0267,
      "step": 1171
    },
    {
      "epoch": 1.7571214392803598,
      "grad_norm": 0.7564618922381177,
      "learning_rate": 3.594876855944385e-07,
      "loss": 0.0327,
      "step": 1172
    },
    {
      "epoch": 1.7586206896551724,
      "grad_norm": 0.6265537380929622,
      "learning_rate": 3.5511640091604293e-07,
      "loss": 0.0249,
      "step": 1173
    },
    {
      "epoch": 1.760119940029985,
      "grad_norm": 0.48140826714372775,
      "learning_rate": 3.50770877237821e-07,
      "loss": 0.019,
      "step": 1174
    },
    {
      "epoch": 1.7616191904047978,
      "grad_norm": 0.5839096983927832,
      "learning_rate": 3.4645113866049187e-07,
      "loss": 0.0229,
      "step": 1175
    },
    {
      "epoch": 1.76311844077961,
      "grad_norm": 0.456556900828423,
      "learning_rate": 3.42157209141768e-07,
      "loss": 0.0211,
      "step": 1176
    },
    {
      "epoch": 1.764617691154423,
      "grad_norm": 0.6233052971826757,
      "learning_rate": 3.3788911249622194e-07,
      "loss": 0.0265,
      "step": 1177
    },
    {
      "epoch": 1.7661169415292353,
      "grad_norm": 0.6475981607195184,
      "learning_rate": 3.336468723951558e-07,
      "loss": 0.0228,
      "step": 1178
    },
    {
      "epoch": 1.767616191904048,
      "grad_norm": 0.5406230357145534,
      "learning_rate": 3.294305123664665e-07,
      "loss": 0.0213,
      "step": 1179
    },
    {
      "epoch": 1.7691154422788604,
      "grad_norm": 0.5788289679631263,
      "learning_rate": 3.2524005579452014e-07,
      "loss": 0.0256,
      "step": 1180
    },
    {
      "epoch": 1.7706146926536732,
      "grad_norm": 0.5375606754370195,
      "learning_rate": 3.2107552592001657e-07,
      "loss": 0.0229,
      "step": 1181
    },
    {
      "epoch": 1.7721139430284858,
      "grad_norm": 0.4581981815192777,
      "learning_rate": 3.169369458398652e-07,
      "loss": 0.0164,
      "step": 1182
    },
    {
      "epoch": 1.7736131934032984,
      "grad_norm": 0.5393683390135412,
      "learning_rate": 3.128243385070562e-07,
      "loss": 0.0251,
      "step": 1183
    },
    {
      "epoch": 1.775112443778111,
      "grad_norm": 0.5287668231285577,
      "learning_rate": 3.087377267305297e-07,
      "loss": 0.0272,
      "step": 1184
    },
    {
      "epoch": 1.7766116941529235,
      "grad_norm": 0.4900736216451649,
      "learning_rate": 3.0467713317505363e-07,
      "loss": 0.0199,
      "step": 1185
    },
    {
      "epoch": 1.778110944527736,
      "grad_norm": 0.4702350707276514,
      "learning_rate": 3.006425803610963e-07,
      "loss": 0.0247,
      "step": 1186
    },
    {
      "epoch": 1.7796101949025487,
      "grad_norm": 0.6808918730987898,
      "learning_rate": 2.9663409066470025e-07,
      "loss": 0.024,
      "step": 1187
    },
    {
      "epoch": 1.7811094452773615,
      "grad_norm": 0.5637035235892154,
      "learning_rate": 2.9265168631736005e-07,
      "loss": 0.0246,
      "step": 1188
    },
    {
      "epoch": 1.7826086956521738,
      "grad_norm": 0.6083320161053476,
      "learning_rate": 2.88695389405898e-07,
      "loss": 0.0212,
      "step": 1189
    },
    {
      "epoch": 1.7841079460269866,
      "grad_norm": 0.5952035828803542,
      "learning_rate": 2.8476522187234177e-07,
      "loss": 0.0233,
      "step": 1190
    },
    {
      "epoch": 1.785607196401799,
      "grad_norm": 0.5540342707573074,
      "learning_rate": 2.808612055138038e-07,
      "loss": 0.0296,
      "step": 1191
    },
    {
      "epoch": 1.7871064467766118,
      "grad_norm": 0.4929856338527852,
      "learning_rate": 2.76983361982357e-07,
      "loss": 0.0199,
      "step": 1192
    },
    {
      "epoch": 1.7886056971514241,
      "grad_norm": 0.4144356066565379,
      "learning_rate": 2.731317127849209e-07,
      "loss": 0.0178,
      "step": 1193
    },
    {
      "epoch": 1.790104947526237,
      "grad_norm": 0.6141306831727124,
      "learning_rate": 2.693062792831358e-07,
      "loss": 0.0265,
      "step": 1194
    },
    {
      "epoch": 1.7916041979010495,
      "grad_norm": 0.6742382532326417,
      "learning_rate": 2.655070826932471e-07,
      "loss": 0.0282,
      "step": 1195
    },
    {
      "epoch": 1.793103448275862,
      "grad_norm": 0.5177365180595115,
      "learning_rate": 2.617341440859883e-07,
      "loss": 0.0202,
      "step": 1196
    },
    {
      "epoch": 1.7946026986506747,
      "grad_norm": 0.41345007525946736,
      "learning_rate": 2.5798748438646326e-07,
      "loss": 0.0193,
      "step": 1197
    },
    {
      "epoch": 1.7961019490254873,
      "grad_norm": 0.45640102124614507,
      "learning_rate": 2.5426712437403134e-07,
      "loss": 0.023,
      "step": 1198
    },
    {
      "epoch": 1.7976011994002998,
      "grad_norm": 0.34635056851302554,
      "learning_rate": 2.5057308468218913e-07,
      "loss": 0.0137,
      "step": 1199
    },
    {
      "epoch": 1.7991004497751124,
      "grad_norm": 0.4828044655999779,
      "learning_rate": 2.4690538579845933e-07,
      "loss": 0.0248,
      "step": 1200
    },
    {
      "epoch": 1.7991004497751124,
      "eval_loss": 0.10092770308256149,
      "eval_runtime": 2.1944,
      "eval_samples_per_second": 24.608,
      "eval_steps_per_second": 6.38,
      "step": 1200
    },
    {
      "epoch": 1.800599700149925,
      "grad_norm": 0.5875386887288677,
      "learning_rate": 2.432640480642756e-07,
      "loss": 0.0243,
      "step": 1201
    },
    {
      "epoch": 1.8020989505247376,
      "grad_norm": 0.5267758107842172,
      "learning_rate": 2.396490916748706e-07,
      "loss": 0.0236,
      "step": 1202
    },
    {
      "epoch": 1.8035982008995504,
      "grad_norm": 0.747393692390445,
      "learning_rate": 2.360605366791624e-07,
      "loss": 0.0367,
      "step": 1203
    },
    {
      "epoch": 1.8050974512743627,
      "grad_norm": 0.49378031842703524,
      "learning_rate": 2.32498402979644e-07,
      "loss": 0.0213,
      "step": 1204
    },
    {
      "epoch": 1.8065967016491755,
      "grad_norm": 0.5651273416130995,
      "learning_rate": 2.2896271033227392e-07,
      "loss": 0.0287,
      "step": 1205
    },
    {
      "epoch": 1.8080959520239879,
      "grad_norm": 0.8778286066687495,
      "learning_rate": 2.2545347834636632e-07,
      "loss": 0.0386,
      "step": 1206
    },
    {
      "epoch": 1.8095952023988007,
      "grad_norm": 0.4537622377112817,
      "learning_rate": 2.219707264844806e-07,
      "loss": 0.0218,
      "step": 1207
    },
    {
      "epoch": 1.811094452773613,
      "grad_norm": 0.6687814275982513,
      "learning_rate": 2.1851447406231573e-07,
      "loss": 0.028,
      "step": 1208
    },
    {
      "epoch": 1.8125937031484258,
      "grad_norm": 0.569153043304943,
      "learning_rate": 2.1508474024860171e-07,
      "loss": 0.0326,
      "step": 1209
    },
    {
      "epoch": 1.8140929535232384,
      "grad_norm": 0.45427566207888304,
      "learning_rate": 2.1168154406499275e-07,
      "loss": 0.0189,
      "step": 1210
    },
    {
      "epoch": 1.815592203898051,
      "grad_norm": 0.6470597113990347,
      "learning_rate": 2.0830490438596418e-07,
      "loss": 0.0283,
      "step": 1211
    },
    {
      "epoch": 1.8170914542728636,
      "grad_norm": 0.4290432812223703,
      "learning_rate": 2.0495483993870578e-07,
      "loss": 0.0208,
      "step": 1212
    },
    {
      "epoch": 1.8185907046476761,
      "grad_norm": 0.4507478986965756,
      "learning_rate": 2.0163136930301696e-07,
      "loss": 0.0195,
      "step": 1213
    },
    {
      "epoch": 1.8200899550224887,
      "grad_norm": 0.7457745508596998,
      "learning_rate": 1.9833451091120727e-07,
      "loss": 0.0333,
      "step": 1214
    },
    {
      "epoch": 1.8215892053973013,
      "grad_norm": 0.8252570815927348,
      "learning_rate": 1.9506428304799095e-07,
      "loss": 0.0345,
      "step": 1215
    },
    {
      "epoch": 1.823088455772114,
      "grad_norm": 0.5369295796730613,
      "learning_rate": 1.9182070385038555e-07,
      "loss": 0.024,
      "step": 1216
    },
    {
      "epoch": 1.8245877061469264,
      "grad_norm": 0.4662755792800672,
      "learning_rate": 1.886037913076144e-07,
      "loss": 0.0171,
      "step": 1217
    },
    {
      "epoch": 1.8260869565217392,
      "grad_norm": 0.7510299250201603,
      "learning_rate": 1.8541356326100436e-07,
      "loss": 0.0314,
      "step": 1218
    },
    {
      "epoch": 1.8275862068965516,
      "grad_norm": 0.5947149502069593,
      "learning_rate": 1.8225003740388546e-07,
      "loss": 0.0297,
      "step": 1219
    },
    {
      "epoch": 1.8290854572713644,
      "grad_norm": 0.4483790114223667,
      "learning_rate": 1.791132312814975e-07,
      "loss": 0.0178,
      "step": 1220
    },
    {
      "epoch": 1.8305847076461768,
      "grad_norm": 0.5808853370991969,
      "learning_rate": 1.760031622908881e-07,
      "loss": 0.0211,
      "step": 1221
    },
    {
      "epoch": 1.8320839580209896,
      "grad_norm": 0.481349833973364,
      "learning_rate": 1.729198476808186e-07,
      "loss": 0.0207,
      "step": 1222
    },
    {
      "epoch": 1.8335832083958021,
      "grad_norm": 0.509601303472534,
      "learning_rate": 1.6986330455166733e-07,
      "loss": 0.0223,
      "step": 1223
    },
    {
      "epoch": 1.8350824587706147,
      "grad_norm": 0.7107287368595905,
      "learning_rate": 1.6683354985533583e-07,
      "loss": 0.027,
      "step": 1224
    },
    {
      "epoch": 1.8365817091454273,
      "grad_norm": 0.5437327100505285,
      "learning_rate": 1.6383060039515343e-07,
      "loss": 0.0214,
      "step": 1225
    },
    {
      "epoch": 1.8380809595202399,
      "grad_norm": 0.6869256450059652,
      "learning_rate": 1.6085447282578548e-07,
      "loss": 0.0316,
      "step": 1226
    },
    {
      "epoch": 1.8395802098950524,
      "grad_norm": 0.5261727150573893,
      "learning_rate": 1.579051836531409e-07,
      "loss": 0.0216,
      "step": 1227
    },
    {
      "epoch": 1.841079460269865,
      "grad_norm": 0.7516210032192877,
      "learning_rate": 1.5498274923427925e-07,
      "loss": 0.0308,
      "step": 1228
    },
    {
      "epoch": 1.8425787106446778,
      "grad_norm": 0.6730946284648598,
      "learning_rate": 1.5208718577732096e-07,
      "loss": 0.0331,
      "step": 1229
    },
    {
      "epoch": 1.8440779610194902,
      "grad_norm": 0.4501401402749414,
      "learning_rate": 1.4921850934135785e-07,
      "loss": 0.0211,
      "step": 1230
    },
    {
      "epoch": 1.845577211394303,
      "grad_norm": 0.8081430940337837,
      "learning_rate": 1.463767358363627e-07,
      "loss": 0.0422,
      "step": 1231
    },
    {
      "epoch": 1.8470764617691153,
      "grad_norm": 0.5971551858197677,
      "learning_rate": 1.4356188102310266e-07,
      "loss": 0.0298,
      "step": 1232
    },
    {
      "epoch": 1.8485757121439281,
      "grad_norm": 0.673894890482827,
      "learning_rate": 1.4077396051305093e-07,
      "loss": 0.0282,
      "step": 1233
    },
    {
      "epoch": 1.8500749625187405,
      "grad_norm": 0.4927990523562101,
      "learning_rate": 1.3801298976830025e-07,
      "loss": 0.0223,
      "step": 1234
    },
    {
      "epoch": 1.8515742128935533,
      "grad_norm": 0.45076858158968264,
      "learning_rate": 1.3527898410147677e-07,
      "loss": 0.0199,
      "step": 1235
    },
    {
      "epoch": 1.8530734632683659,
      "grad_norm": 0.6046457364834953,
      "learning_rate": 1.325719586756563e-07,
      "loss": 0.0278,
      "step": 1236
    },
    {
      "epoch": 1.8545727136431784,
      "grad_norm": 0.5580999464240284,
      "learning_rate": 1.2989192850427933e-07,
      "loss": 0.0291,
      "step": 1237
    },
    {
      "epoch": 1.856071964017991,
      "grad_norm": 0.6176143048110353,
      "learning_rate": 1.2723890845106723e-07,
      "loss": 0.0226,
      "step": 1238
    },
    {
      "epoch": 1.8575712143928036,
      "grad_norm": 0.4871289413787136,
      "learning_rate": 1.2461291322994118e-07,
      "loss": 0.0279,
      "step": 1239
    },
    {
      "epoch": 1.8590704647676162,
      "grad_norm": 0.5434547461934179,
      "learning_rate": 1.2201395740493948e-07,
      "loss": 0.023,
      "step": 1240
    },
    {
      "epoch": 1.8605697151424287,
      "grad_norm": 0.7417943492972645,
      "learning_rate": 1.1944205539013708e-07,
      "loss": 0.0307,
      "step": 1241
    },
    {
      "epoch": 1.8620689655172413,
      "grad_norm": 0.35040518071938626,
      "learning_rate": 1.1689722144956672e-07,
      "loss": 0.0149,
      "step": 1242
    },
    {
      "epoch": 1.863568215892054,
      "grad_norm": 0.5625424800321914,
      "learning_rate": 1.1437946969713731e-07,
      "loss": 0.0306,
      "step": 1243
    },
    {
      "epoch": 1.8650674662668667,
      "grad_norm": 0.6414350091989585,
      "learning_rate": 1.1188881409655849e-07,
      "loss": 0.0274,
      "step": 1244
    },
    {
      "epoch": 1.866566716641679,
      "grad_norm": 0.4822600980746867,
      "learning_rate": 1.0942526846126122e-07,
      "loss": 0.0217,
      "step": 1245
    },
    {
      "epoch": 1.8680659670164919,
      "grad_norm": 0.5265865201081962,
      "learning_rate": 1.0698884645432117e-07,
      "loss": 0.0182,
      "step": 1246
    },
    {
      "epoch": 1.8695652173913042,
      "grad_norm": 0.37757292161976186,
      "learning_rate": 1.0457956158838545e-07,
      "loss": 0.0163,
      "step": 1247
    },
    {
      "epoch": 1.871064467766117,
      "grad_norm": 0.5539671486005535,
      "learning_rate": 1.0219742722559433e-07,
      "loss": 0.0215,
      "step": 1248
    },
    {
      "epoch": 1.8725637181409296,
      "grad_norm": 0.5565667215916354,
      "learning_rate": 9.984245657750857e-08,
      "loss": 0.0297,
      "step": 1249
    },
    {
      "epoch": 1.8740629685157422,
      "grad_norm": 1.0039129503650757,
      "learning_rate": 9.751466270503718e-08,
      "loss": 0.0357,
      "step": 1250
    },
    {
      "epoch": 1.8755622188905547,
      "grad_norm": 0.6587567610012585,
      "learning_rate": 9.521405851836252e-08,
      "loss": 0.0281,
      "step": 1251
    },
    {
      "epoch": 1.8770614692653673,
      "grad_norm": 0.6059668143091147,
      "learning_rate": 9.294065677687202e-08,
      "loss": 0.0262,
      "step": 1252
    },
    {
      "epoch": 1.87856071964018,
      "grad_norm": 0.5746681473308829,
      "learning_rate": 9.069447008908383e-08,
      "loss": 0.0326,
      "step": 1253
    },
    {
      "epoch": 1.8800599700149925,
      "grad_norm": 0.45915163038825874,
      "learning_rate": 8.847551091257956e-08,
      "loss": 0.0163,
      "step": 1254
    },
    {
      "epoch": 1.881559220389805,
      "grad_norm": 0.5440883980180552,
      "learning_rate": 8.62837915539344e-08,
      "loss": 0.0267,
      "step": 1255
    },
    {
      "epoch": 1.8830584707646176,
      "grad_norm": 0.4377138125820403,
      "learning_rate": 8.411932416864832e-08,
      "loss": 0.0163,
      "step": 1256
    },
    {
      "epoch": 1.8845577211394304,
      "grad_norm": 0.5557313493069,
      "learning_rate": 8.198212076107881e-08,
      "loss": 0.0266,
      "step": 1257
    },
    {
      "epoch": 1.8860569715142428,
      "grad_norm": 0.3639946450874956,
      "learning_rate": 7.987219318437489e-08,
      "loss": 0.0117,
      "step": 1258
    },
    {
      "epoch": 1.8875562218890556,
      "grad_norm": 0.5975114625015109,
      "learning_rate": 7.778955314041103e-08,
      "loss": 0.0181,
      "step": 1259
    },
    {
      "epoch": 1.889055472263868,
      "grad_norm": 0.4571948051817939,
      "learning_rate": 7.573421217972222e-08,
      "loss": 0.0167,
      "step": 1260
    },
    {
      "epoch": 1.8905547226386807,
      "grad_norm": 0.879798632412098,
      "learning_rate": 7.370618170144062e-08,
      "loss": 0.0481,
      "step": 1261
    },
    {
      "epoch": 1.892053973013493,
      "grad_norm": 0.6855241385848212,
      "learning_rate": 7.170547295323016e-08,
      "loss": 0.0415,
      "step": 1262
    },
    {
      "epoch": 1.893553223388306,
      "grad_norm": 0.5529688263640988,
      "learning_rate": 6.973209703122652e-08,
      "loss": 0.0246,
      "step": 1263
    },
    {
      "epoch": 1.8950524737631185,
      "grad_norm": 0.6133751221959671,
      "learning_rate": 6.778606487997496e-08,
      "loss": 0.0286,
      "step": 1264
    },
    {
      "epoch": 1.896551724137931,
      "grad_norm": 0.6955761132658265,
      "learning_rate": 6.58673872923693e-08,
      "loss": 0.0281,
      "step": 1265
    },
    {
      "epoch": 1.8980509745127436,
      "grad_norm": 0.5915247352838723,
      "learning_rate": 6.397607490959134e-08,
      "loss": 0.0248,
      "step": 1266
    },
    {
      "epoch": 1.8995502248875562,
      "grad_norm": 0.6981310897882675,
      "learning_rate": 6.211213822105378e-08,
      "loss": 0.0355,
      "step": 1267
    },
    {
      "epoch": 1.9010494752623688,
      "grad_norm": 0.6014536127769257,
      "learning_rate": 6.027558756434015e-08,
      "loss": 0.0228,
      "step": 1268
    },
    {
      "epoch": 1.9025487256371814,
      "grad_norm": 0.4570278806308648,
      "learning_rate": 5.846643312514888e-08,
      "loss": 0.0214,
      "step": 1269
    },
    {
      "epoch": 1.9040479760119942,
      "grad_norm": 0.593646007879774,
      "learning_rate": 5.668468493723489e-08,
      "loss": 0.0156,
      "step": 1270
    },
    {
      "epoch": 1.9055472263868065,
      "grad_norm": 0.45766334488378513,
      "learning_rate": 5.4930352882357486e-08,
      "loss": 0.0154,
      "step": 1271
    },
    {
      "epoch": 1.9070464767616193,
      "grad_norm": 0.6222327480918961,
      "learning_rate": 5.3203446690220374e-08,
      "loss": 0.0296,
      "step": 1272
    },
    {
      "epoch": 1.9085457271364317,
      "grad_norm": 0.6770925211911946,
      "learning_rate": 5.1503975938422824e-08,
      "loss": 0.0292,
      "step": 1273
    },
    {
      "epoch": 1.9100449775112445,
      "grad_norm": 0.5774997914414777,
      "learning_rate": 4.983195005240415e-08,
      "loss": 0.0192,
      "step": 1274
    },
    {
      "epoch": 1.9115442278860568,
      "grad_norm": 0.5532315152254872,
      "learning_rate": 4.8187378305390994e-08,
      "loss": 0.0243,
      "step": 1275
    },
    {
      "epoch": 1.9130434782608696,
      "grad_norm": 0.3926998497308617,
      "learning_rate": 4.657026981834623e-08,
      "loss": 0.016,
      "step": 1276
    },
    {
      "epoch": 1.9145427286356822,
      "grad_norm": 0.6453618827342124,
      "learning_rate": 4.498063355991955e-08,
      "loss": 0.0303,
      "step": 1277
    },
    {
      "epoch": 1.9160419790104948,
      "grad_norm": 0.5925324209676621,
      "learning_rate": 4.341847834639645e-08,
      "loss": 0.0243,
      "step": 1278
    },
    {
      "epoch": 1.9175412293853074,
      "grad_norm": 0.46725443455980314,
      "learning_rate": 4.188381284164933e-08,
      "loss": 0.0144,
      "step": 1279
    },
    {
      "epoch": 1.91904047976012,
      "grad_norm": 0.5680014941493454,
      "learning_rate": 4.0376645557090864e-08,
      "loss": 0.0326,
      "step": 1280
    },
    {
      "epoch": 1.9205397301349325,
      "grad_norm": 0.7217504125320175,
      "learning_rate": 3.889698485162463e-08,
      "loss": 0.0303,
      "step": 1281
    },
    {
      "epoch": 1.922038980509745,
      "grad_norm": 0.656126581778829,
      "learning_rate": 3.744483893160067e-08,
      "loss": 0.0306,
      "step": 1282
    },
    {
      "epoch": 1.9235382308845579,
      "grad_norm": 0.6042953421656458,
      "learning_rate": 3.602021585076942e-08,
      "loss": 0.0271,
      "step": 1283
    },
    {
      "epoch": 1.9250374812593702,
      "grad_norm": 0.7486704805438954,
      "learning_rate": 3.462312351023567e-08,
      "loss": 0.0324,
      "step": 1284
    },
    {
      "epoch": 1.926536731634183,
      "grad_norm": 0.7174026644193179,
      "learning_rate": 3.325356965841686e-08,
      "loss": 0.0223,
      "step": 1285
    },
    {
      "epoch": 1.9280359820089954,
      "grad_norm": 0.5148143130677811,
      "learning_rate": 3.191156189099931e-08,
      "loss": 0.0204,
      "step": 1286
    },
    {
      "epoch": 1.9295352323838082,
      "grad_norm": 0.47499908573706423,
      "learning_rate": 3.0597107650894855e-08,
      "loss": 0.0232,
      "step": 1287
    },
    {
      "epoch": 1.9310344827586206,
      "grad_norm": 0.44251785872104077,
      "learning_rate": 2.9310214228202016e-08,
      "loss": 0.0172,
      "step": 1288
    },
    {
      "epoch": 1.9325337331334334,
      "grad_norm": 0.526820321070341,
      "learning_rate": 2.8050888760163265e-08,
      "loss": 0.0219,
      "step": 1289
    },
    {
      "epoch": 1.934032983508246,
      "grad_norm": 0.6341385668170912,
      "learning_rate": 2.6819138231126695e-08,
      "loss": 0.0287,
      "step": 1290
    },
    {
      "epoch": 1.9355322338830585,
      "grad_norm": 0.5163753189407518,
      "learning_rate": 2.5614969472506634e-08,
      "loss": 0.0186,
      "step": 1291
    },
    {
      "epoch": 1.937031484257871,
      "grad_norm": 0.5478723279883146,
      "learning_rate": 2.4438389162746434e-08,
      "loss": 0.0203,
      "step": 1292
    },
    {
      "epoch": 1.9385307346326837,
      "grad_norm": 0.48924388067417557,
      "learning_rate": 2.3289403827281287e-08,
      "loss": 0.019,
      "step": 1293
    },
    {
      "epoch": 1.9400299850074962,
      "grad_norm": 0.5566860541554749,
      "learning_rate": 2.2168019838501032e-08,
      "loss": 0.0295,
      "step": 1294
    },
    {
      "epoch": 1.9415292353823088,
      "grad_norm": 0.44878555067344117,
      "learning_rate": 2.1074243415716288e-08,
      "loss": 0.0225,
      "step": 1295
    },
    {
      "epoch": 1.9430284857571214,
      "grad_norm": 0.47245058810205054,
      "learning_rate": 2.0008080625124048e-08,
      "loss": 0.0226,
      "step": 1296
    },
    {
      "epoch": 1.944527736131934,
      "grad_norm": 0.46341878595305913,
      "learning_rate": 1.896953737977103e-08,
      "loss": 0.0225,
      "step": 1297
    },
    {
      "epoch": 1.9460269865067468,
      "grad_norm": 0.4489267987813382,
      "learning_rate": 1.7958619439524817e-08,
      "loss": 0.0219,
      "step": 1298
    },
    {
      "epoch": 1.9475262368815591,
      "grad_norm": 0.5456008670199369,
      "learning_rate": 1.6975332411040547e-08,
      "loss": 0.0201,
      "step": 1299
    },
    {
      "epoch": 1.949025487256372,
      "grad_norm": 0.6636964664780095,
      "learning_rate": 1.601968174772761e-08,
      "loss": 0.0233,
      "step": 1300
    },
    {
      "epoch": 1.9505247376311843,
      "grad_norm": 0.5187115529757086,
      "learning_rate": 1.5091672749723564e-08,
      "loss": 0.0226,
      "step": 1301
    },
    {
      "epoch": 1.952023988005997,
      "grad_norm": 0.5623180070853436,
      "learning_rate": 1.4191310563860806e-08,
      "loss": 0.0201,
      "step": 1302
    },
    {
      "epoch": 1.9535232383808094,
      "grad_norm": 0.4527229220106935,
      "learning_rate": 1.331860018363995e-08,
      "loss": 0.016,
      "step": 1303
    },
    {
      "epoch": 1.9550224887556222,
      "grad_norm": 0.5052426213409227,
      "learning_rate": 1.2473546449203178e-08,
      "loss": 0.0298,
      "step": 1304
    },
    {
      "epoch": 1.9565217391304348,
      "grad_norm": 0.4807150509325621,
      "learning_rate": 1.1656154047303691e-08,
      "loss": 0.0221,
      "step": 1305
    },
    {
      "epoch": 1.9580209895052474,
      "grad_norm": 0.47404208126804454,
      "learning_rate": 1.0866427511285194e-08,
      "loss": 0.0172,
      "step": 1306
    },
    {
      "epoch": 1.95952023988006,
      "grad_norm": 0.46974276716267965,
      "learning_rate": 1.0104371221050236e-08,
      "loss": 0.0182,
      "step": 1307
    },
    {
      "epoch": 1.9610194902548725,
      "grad_norm": 0.47635524078339525,
      "learning_rate": 9.369989403041347e-09,
      "loss": 0.0226,
      "step": 1308
    },
    {
      "epoch": 1.9625187406296851,
      "grad_norm": 0.4662219390583888,
      "learning_rate": 8.663286130216608e-09,
      "loss": 0.0197,
      "step": 1309
    },
    {
      "epoch": 1.9640179910044977,
      "grad_norm": 0.7148804489508339,
      "learning_rate": 7.984265322023011e-09,
      "loss": 0.0242,
      "step": 1310
    },
    {
      "epoch": 1.9655172413793105,
      "grad_norm": 0.7481675915137074,
      "learning_rate": 7.332930744380906e-09,
      "loss": 0.0444,
      "step": 1311
    },
    {
      "epoch": 1.9670164917541229,
      "grad_norm": 0.567371402181547,
      "learning_rate": 6.709286009657368e-09,
      "loss": 0.0217,
      "step": 1312
    },
    {
      "epoch": 1.9685157421289357,
      "grad_norm": 0.7029861862724672,
      "learning_rate": 6.1133345766511975e-09,
      "loss": 0.0338,
      "step": 1313
    },
    {
      "epoch": 1.970014992503748,
      "grad_norm": 0.5392234049493475,
      "learning_rate": 5.5450797505690605e-09,
      "loss": 0.0266,
      "step": 1314
    },
    {
      "epoch": 1.9715142428785608,
      "grad_norm": 0.5700325752794452,
      "learning_rate": 5.004524683011048e-09,
      "loss": 0.0205,
      "step": 1315
    },
    {
      "epoch": 1.9730134932533732,
      "grad_norm": 0.5484329006112951,
      "learning_rate": 4.491672371950695e-09,
      "loss": 0.0218,
      "step": 1316
    },
    {
      "epoch": 1.974512743628186,
      "grad_norm": 0.5351881162114712,
      "learning_rate": 4.0065256617199954e-09,
      "loss": 0.0195,
      "step": 1317
    },
    {
      "epoch": 1.9760119940029985,
      "grad_norm": 0.6214124385909919,
      "learning_rate": 3.5490872429910784e-09,
      "loss": 0.0261,
      "step": 1318
    },
    {
      "epoch": 1.9775112443778111,
      "grad_norm": 0.533516834711269,
      "learning_rate": 3.119359652765108e-09,
      "loss": 0.0254,
      "step": 1319
    },
    {
      "epoch": 1.9790104947526237,
      "grad_norm": 0.857602743485462,
      "learning_rate": 2.7173452743550767e-09,
      "loss": 0.0345,
      "step": 1320
    },
    {
      "epoch": 1.9805097451274363,
      "grad_norm": 0.5671615457022771,
      "learning_rate": 2.343046337374144e-09,
      "loss": 0.0243,
      "step": 1321
    },
    {
      "epoch": 1.9820089955022488,
      "grad_norm": 0.6476175170333314,
      "learning_rate": 1.9964649177223184e-09,
      "loss": 0.0281,
      "step": 1322
    },
    {
      "epoch": 1.9835082458770614,
      "grad_norm": 0.4760511651560054,
      "learning_rate": 1.6776029375759062e-09,
      "loss": 0.0262,
      "step": 1323
    },
    {
      "epoch": 1.9850074962518742,
      "grad_norm": 0.8388940877739349,
      "learning_rate": 1.386462165375857e-09,
      "loss": 0.0374,
      "step": 1324
    },
    {
      "epoch": 1.9865067466266866,
      "grad_norm": 0.4466744774968853,
      "learning_rate": 1.1230442158188804e-09,
      "loss": 0.0232,
      "step": 1325
    },
    {
      "epoch": 1.9880059970014994,
      "grad_norm": 0.647695564951376,
      "learning_rate": 8.873505498474544e-10,
      "loss": 0.0247,
      "step": 1326
    },
    {
      "epoch": 1.9895052473763117,
      "grad_norm": 0.649834542691094,
      "learning_rate": 6.793824746437194e-10,
      "loss": 0.0276,
      "step": 1327
    },
    {
      "epoch": 1.9910044977511245,
      "grad_norm": 0.4298889587309842,
      "learning_rate": 4.991411436189308e-10,
      "loss": 0.0166,
      "step": 1328
    },
    {
      "epoch": 1.992503748125937,
      "grad_norm": 0.541111977463262,
      "learning_rate": 3.466275564101285e-10,
      "loss": 0.0242,
      "step": 1329
    },
    {
      "epoch": 1.9940029985007497,
      "grad_norm": 0.6240717111987047,
      "learning_rate": 2.2184255887403028e-10,
      "loss": 0.0284,
      "step": 1330
    },
    {
      "epoch": 1.9955022488755623,
      "grad_norm": 0.5776241722343967,
      "learning_rate": 1.2478684308037115e-10,
      "loss": 0.0244,
      "step": 1331
    },
    {
      "epoch": 1.9970014992503748,
      "grad_norm": 0.6548426663228895,
      "learning_rate": 5.5460947310237435e-11,
      "loss": 0.0295,
      "step": 1332
    },
    {
      "epoch": 1.9985007496251874,
      "grad_norm": 0.4798280662387754,
      "learning_rate": 1.3865256052181252e-11,
      "loss": 0.0167,
      "step": 1333
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.41990943220901145,
      "learning_rate": 0.0,
      "loss": 0.0173,
      "step": 1334
    },
    {
      "epoch": 2.0,
      "step": 1334,
      "total_flos": 5978663976960.0,
      "train_loss": 0.053979536121211265,
      "train_runtime": 1013.256,
      "train_samples_per_second": 10.519,
      "train_steps_per_second": 1.317
    }
  ],
  "logging_steps": 1,
  "max_steps": 1334,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 5000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 5978663976960.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}