{
  "best_metric": 1.3446228504180908,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.10527701013291223,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0005263850506645612,
      "grad_norm": 93.07259368896484,
      "learning_rate": 1e-05,
      "loss": 12.3105,
      "step": 1
    },
    {
      "epoch": 0.0005263850506645612,
      "eval_loss": 2.796645402908325,
      "eval_runtime": 309.2946,
      "eval_samples_per_second": 10.346,
      "eval_steps_per_second": 2.587,
      "step": 1
    },
    {
      "epoch": 0.0010527701013291224,
      "grad_norm": 131.77566528320312,
      "learning_rate": 2e-05,
      "loss": 9.2268,
      "step": 2
    },
    {
      "epoch": 0.0015791551519936833,
      "grad_norm": 71.25743865966797,
      "learning_rate": 3e-05,
      "loss": 8.0512,
      "step": 3
    },
    {
      "epoch": 0.0021055402026582447,
      "grad_norm": 46.8036003112793,
      "learning_rate": 4e-05,
      "loss": 7.2551,
      "step": 4
    },
    {
      "epoch": 0.0026319252533228057,
      "grad_norm": 36.05014419555664,
      "learning_rate": 5e-05,
      "loss": 6.3495,
      "step": 5
    },
    {
      "epoch": 0.0031583103039873666,
      "grad_norm": 26.287073135375977,
      "learning_rate": 6e-05,
      "loss": 5.9319,
      "step": 6
    },
    {
      "epoch": 0.003684695354651928,
      "grad_norm": 38.00836944580078,
      "learning_rate": 7e-05,
      "loss": 5.5911,
      "step": 7
    },
    {
      "epoch": 0.004211080405316489,
      "grad_norm": 23.294231414794922,
      "learning_rate": 8e-05,
      "loss": 5.2376,
      "step": 8
    },
    {
      "epoch": 0.00473746545598105,
      "grad_norm": 22.907855987548828,
      "learning_rate": 9e-05,
      "loss": 4.5457,
      "step": 9
    },
    {
      "epoch": 0.005263850506645611,
      "grad_norm": 27.766204833984375,
      "learning_rate": 0.0001,
      "loss": 6.4629,
      "step": 10
    },
    {
      "epoch": 0.005790235557310172,
      "grad_norm": 24.227497100830078,
      "learning_rate": 9.999316524962345e-05,
      "loss": 5.8682,
      "step": 11
    },
    {
      "epoch": 0.006316620607974733,
      "grad_norm": 20.358135223388672,
      "learning_rate": 9.997266286704631e-05,
      "loss": 5.0779,
      "step": 12
    },
    {
      "epoch": 0.006843005658639294,
      "grad_norm": 22.879426956176758,
      "learning_rate": 9.993849845741524e-05,
      "loss": 5.0858,
      "step": 13
    },
    {
      "epoch": 0.007369390709303856,
      "grad_norm": 22.2658634185791,
      "learning_rate": 9.989068136093873e-05,
      "loss": 5.305,
      "step": 14
    },
    {
      "epoch": 0.007895775759968417,
      "grad_norm": 29.477561950683594,
      "learning_rate": 9.98292246503335e-05,
      "loss": 4.8577,
      "step": 15
    },
    {
      "epoch": 0.008422160810632979,
      "grad_norm": 21.121549606323242,
      "learning_rate": 9.975414512725057e-05,
      "loss": 6.6094,
      "step": 16
    },
    {
      "epoch": 0.008948545861297539,
      "grad_norm": 21.887968063354492,
      "learning_rate": 9.966546331768191e-05,
      "loss": 6.1797,
      "step": 17
    },
    {
      "epoch": 0.0094749309119621,
      "grad_norm": 19.13234519958496,
      "learning_rate": 9.956320346634876e-05,
      "loss": 5.6551,
      "step": 18
    },
    {
      "epoch": 0.010001315962626661,
      "grad_norm": 18.616533279418945,
      "learning_rate": 9.944739353007344e-05,
      "loss": 5.9483,
      "step": 19
    },
    {
      "epoch": 0.010527701013291223,
      "grad_norm": 18.95231056213379,
      "learning_rate": 9.931806517013612e-05,
      "loss": 5.4522,
      "step": 20
    },
    {
      "epoch": 0.011054086063955783,
      "grad_norm": 18.209264755249023,
      "learning_rate": 9.917525374361912e-05,
      "loss": 6.1536,
      "step": 21
    },
    {
      "epoch": 0.011580471114620345,
      "grad_norm": 16.42949867248535,
      "learning_rate": 9.901899829374047e-05,
      "loss": 6.0107,
      "step": 22
    },
    {
      "epoch": 0.012106856165284906,
      "grad_norm": 15.05119514465332,
      "learning_rate": 9.884934153917997e-05,
      "loss": 6.5074,
      "step": 23
    },
    {
      "epoch": 0.012633241215949467,
      "grad_norm": 16.25713348388672,
      "learning_rate": 9.86663298624003e-05,
      "loss": 7.0152,
      "step": 24
    },
    {
      "epoch": 0.013159626266614028,
      "grad_norm": 13.90926742553711,
      "learning_rate": 9.847001329696653e-05,
      "loss": 5.5108,
      "step": 25
    },
    {
      "epoch": 0.013686011317278588,
      "grad_norm": 18.535930633544922,
      "learning_rate": 9.826044551386744e-05,
      "loss": 5.9702,
      "step": 26
    },
    {
      "epoch": 0.01421239636794315,
      "grad_norm": 15.170503616333008,
      "learning_rate": 9.803768380684242e-05,
      "loss": 5.659,
      "step": 27
    },
    {
      "epoch": 0.014738781418607712,
      "grad_norm": 17.371408462524414,
      "learning_rate": 9.780178907671789e-05,
      "loss": 5.9414,
      "step": 28
    },
    {
      "epoch": 0.015265166469272272,
      "grad_norm": 15.820178031921387,
      "learning_rate": 9.755282581475769e-05,
      "loss": 5.1419,
      "step": 29
    },
    {
      "epoch": 0.015791551519936834,
      "grad_norm": 14.776792526245117,
      "learning_rate": 9.729086208503174e-05,
      "loss": 5.402,
      "step": 30
    },
    {
      "epoch": 0.016317936570601394,
      "grad_norm": 16.17049217224121,
      "learning_rate": 9.701596950580806e-05,
      "loss": 6.1015,
      "step": 31
    },
    {
      "epoch": 0.016844321621265958,
      "grad_norm": 15.148641586303711,
      "learning_rate": 9.672822322997305e-05,
      "loss": 4.9903,
      "step": 32
    },
    {
      "epoch": 0.017370706671930518,
      "grad_norm": 15.8573637008667,
      "learning_rate": 9.642770192448536e-05,
      "loss": 6.3152,
      "step": 33
    },
    {
      "epoch": 0.017897091722595078,
      "grad_norm": 14.71132755279541,
      "learning_rate": 9.611448774886924e-05,
      "loss": 5.4481,
      "step": 34
    },
    {
      "epoch": 0.018423476773259638,
      "grad_norm": 14.21374797821045,
      "learning_rate": 9.578866633275288e-05,
      "loss": 6.3535,
      "step": 35
    },
    {
      "epoch": 0.0189498618239242,
      "grad_norm": 15.075162887573242,
      "learning_rate": 9.545032675245813e-05,
      "loss": 7.1606,
      "step": 36
    },
    {
      "epoch": 0.01947624687458876,
      "grad_norm": 14.579214096069336,
      "learning_rate": 9.509956150664796e-05,
      "loss": 7.253,
      "step": 37
    },
    {
      "epoch": 0.020002631925253322,
      "grad_norm": 13.596733093261719,
      "learning_rate": 9.473646649103818e-05,
      "loss": 6.1713,
      "step": 38
    },
    {
      "epoch": 0.020529016975917885,
      "grad_norm": 12.464555740356445,
      "learning_rate": 9.43611409721806e-05,
      "loss": 5.6624,
      "step": 39
    },
    {
      "epoch": 0.021055402026582445,
      "grad_norm": 15.085432052612305,
      "learning_rate": 9.397368756032445e-05,
      "loss": 6.6267,
      "step": 40
    },
    {
      "epoch": 0.021581787077247006,
      "grad_norm": 15.069063186645508,
      "learning_rate": 9.357421218136386e-05,
      "loss": 7.4622,
      "step": 41
    },
    {
      "epoch": 0.022108172127911566,
      "grad_norm": 13.953812599182129,
      "learning_rate": 9.316282404787871e-05,
      "loss": 6.3665,
      "step": 42
    },
    {
      "epoch": 0.02263455717857613,
      "grad_norm": 13.846559524536133,
      "learning_rate": 9.273963562927695e-05,
      "loss": 7.0589,
      "step": 43
    },
    {
      "epoch": 0.02316094222924069,
      "grad_norm": 14.49377155303955,
      "learning_rate": 9.230476262104677e-05,
      "loss": 7.26,
      "step": 44
    },
    {
      "epoch": 0.02368732727990525,
      "grad_norm": 12.730331420898438,
      "learning_rate": 9.185832391312644e-05,
      "loss": 7.1364,
      "step": 45
    },
    {
      "epoch": 0.024213712330569813,
      "grad_norm": 12.382709503173828,
      "learning_rate": 9.140044155740101e-05,
      "loss": 6.424,
      "step": 46
    },
    {
      "epoch": 0.024740097381234373,
      "grad_norm": 13.245918273925781,
      "learning_rate": 9.093124073433463e-05,
      "loss": 6.5557,
      "step": 47
    },
    {
      "epoch": 0.025266482431898933,
      "grad_norm": 16.19131851196289,
      "learning_rate": 9.045084971874738e-05,
      "loss": 6.636,
      "step": 48
    },
    {
      "epoch": 0.025792867482563497,
      "grad_norm": 13.08187484741211,
      "learning_rate": 8.995939984474624e-05,
      "loss": 7.1473,
      "step": 49
    },
    {
      "epoch": 0.026319252533228057,
      "grad_norm": 13.130578994750977,
      "learning_rate": 8.945702546981969e-05,
      "loss": 7.0864,
      "step": 50
    },
    {
      "epoch": 0.026319252533228057,
      "eval_loss": 1.5200674533843994,
      "eval_runtime": 312.2573,
      "eval_samples_per_second": 10.248,
      "eval_steps_per_second": 2.562,
      "step": 50
    },
    {
      "epoch": 0.026845637583892617,
      "grad_norm": 19.267629623413086,
      "learning_rate": 8.894386393810563e-05,
      "loss": 9.1401,
      "step": 51
    },
    {
      "epoch": 0.027372022634557177,
      "grad_norm": 17.245580673217773,
      "learning_rate": 8.842005554284296e-05,
      "loss": 6.092,
      "step": 52
    },
    {
      "epoch": 0.02789840768522174,
      "grad_norm": 15.794323921203613,
      "learning_rate": 8.788574348801675e-05,
      "loss": 5.6327,
      "step": 53
    },
    {
      "epoch": 0.0284247927358863,
      "grad_norm": 14.931527137756348,
      "learning_rate": 8.73410738492077e-05,
      "loss": 6.219,
      "step": 54
    },
    {
      "epoch": 0.02895117778655086,
      "grad_norm": 14.541658401489258,
      "learning_rate": 8.678619553365659e-05,
      "loss": 6.4612,
      "step": 55
    },
    {
      "epoch": 0.029477562837215424,
      "grad_norm": 14.111733436584473,
      "learning_rate": 8.622126023955446e-05,
      "loss": 6.5942,
      "step": 56
    },
    {
      "epoch": 0.030003947887879984,
      "grad_norm": 13.476927757263184,
      "learning_rate": 8.564642241456986e-05,
      "loss": 5.099,
      "step": 57
    },
    {
      "epoch": 0.030530332938544544,
      "grad_norm": 15.090818405151367,
      "learning_rate": 8.506183921362443e-05,
      "loss": 4.8915,
      "step": 58
    },
    {
      "epoch": 0.031056717989209108,
      "grad_norm": 14.353198051452637,
      "learning_rate": 8.44676704559283e-05,
      "loss": 5.1166,
      "step": 59
    },
    {
      "epoch": 0.03158310303987367,
      "grad_norm": 16.560720443725586,
      "learning_rate": 8.386407858128706e-05,
      "loss": 4.8429,
      "step": 60
    },
    {
      "epoch": 0.03210948809053823,
      "grad_norm": 13.897335052490234,
      "learning_rate": 8.32512286056924e-05,
      "loss": 4.1389,
      "step": 61
    },
    {
      "epoch": 0.03263587314120279,
      "grad_norm": 16.702407836914062,
      "learning_rate": 8.262928807620843e-05,
      "loss": 4.9801,
      "step": 62
    },
    {
      "epoch": 0.03316225819186735,
      "grad_norm": 17.112253189086914,
      "learning_rate": 8.199842702516583e-05,
      "loss": 5.6549,
      "step": 63
    },
    {
      "epoch": 0.033688643242531915,
      "grad_norm": 15.21346378326416,
      "learning_rate": 8.135881792367686e-05,
      "loss": 5.5022,
      "step": 64
    },
    {
      "epoch": 0.034215028293196476,
      "grad_norm": 14.541852951049805,
      "learning_rate": 8.07106356344834e-05,
      "loss": 5.4754,
      "step": 65
    },
    {
      "epoch": 0.034741413343861036,
      "grad_norm": 13.415721893310547,
      "learning_rate": 8.005405736415126e-05,
      "loss": 4.9441,
      "step": 66
    },
    {
      "epoch": 0.035267798394525596,
      "grad_norm": 14.754863739013672,
      "learning_rate": 7.938926261462366e-05,
      "loss": 5.6391,
      "step": 67
    },
    {
      "epoch": 0.035794183445190156,
      "grad_norm": 12.466286659240723,
      "learning_rate": 7.871643313414718e-05,
      "loss": 4.4355,
      "step": 68
    },
    {
      "epoch": 0.036320568495854716,
      "grad_norm": 11.988637924194336,
      "learning_rate": 7.803575286758364e-05,
      "loss": 5.0228,
      "step": 69
    },
    {
      "epoch": 0.036846953546519276,
      "grad_norm": 12.856491088867188,
      "learning_rate": 7.734740790612136e-05,
      "loss": 4.4895,
      "step": 70
    },
    {
      "epoch": 0.03737333859718384,
      "grad_norm": 15.164680480957031,
      "learning_rate": 7.66515864363997e-05,
      "loss": 4.8249,
      "step": 71
    },
    {
      "epoch": 0.0378997236478484,
      "grad_norm": 13.548012733459473,
      "learning_rate": 7.594847868906076e-05,
      "loss": 5.7685,
      "step": 72
    },
    {
      "epoch": 0.03842610869851296,
      "grad_norm": 13.549013137817383,
      "learning_rate": 7.52382768867422e-05,
      "loss": 4.5342,
      "step": 73
    },
    {
      "epoch": 0.03895249374917752,
      "grad_norm": 14.29944896697998,
      "learning_rate": 7.452117519152542e-05,
      "loss": 5.9741,
      "step": 74
    },
    {
      "epoch": 0.03947887879984208,
      "grad_norm": 12.570438385009766,
      "learning_rate": 7.379736965185368e-05,
      "loss": 5.0425,
      "step": 75
    },
    {
      "epoch": 0.040005263850506644,
      "grad_norm": 11.083076477050781,
      "learning_rate": 7.30670581489344e-05,
      "loss": 4.0083,
      "step": 76
    },
    {
      "epoch": 0.040531648901171204,
      "grad_norm": 13.603141784667969,
      "learning_rate": 7.233044034264034e-05,
      "loss": 6.5059,
      "step": 77
    },
    {
      "epoch": 0.04105803395183577,
      "grad_norm": 13.763607025146484,
      "learning_rate": 7.158771761692464e-05,
      "loss": 5.2911,
      "step": 78
    },
    {
      "epoch": 0.04158441900250033,
      "grad_norm": 13.06667423248291,
      "learning_rate": 7.083909302476453e-05,
      "loss": 4.4704,
      "step": 79
    },
    {
      "epoch": 0.04211080405316489,
      "grad_norm": 13.00338077545166,
      "learning_rate": 7.008477123264848e-05,
      "loss": 5.3127,
      "step": 80
    },
    {
      "epoch": 0.04263718910382945,
      "grad_norm": 13.851142883300781,
      "learning_rate": 6.932495846462261e-05,
      "loss": 6.4596,
      "step": 81
    },
    {
      "epoch": 0.04316357415449401,
      "grad_norm": 12.976868629455566,
      "learning_rate": 6.855986244591104e-05,
      "loss": 6.0087,
      "step": 82
    },
    {
      "epoch": 0.04368995920515857,
      "grad_norm": 12.991766929626465,
      "learning_rate": 6.778969234612584e-05,
      "loss": 5.1351,
      "step": 83
    },
    {
      "epoch": 0.04421634425582313,
      "grad_norm": 14.363101959228516,
      "learning_rate": 6.701465872208216e-05,
      "loss": 6.3058,
      "step": 84
    },
    {
      "epoch": 0.0447427293064877,
      "grad_norm": 14.271668434143066,
      "learning_rate": 6.623497346023418e-05,
      "loss": 5.8214,
      "step": 85
    },
    {
      "epoch": 0.04526911435715226,
      "grad_norm": 12.844442367553711,
      "learning_rate": 6.545084971874738e-05,
      "loss": 5.541,
      "step": 86
    },
    {
      "epoch": 0.04579549940781682,
      "grad_norm": 15.242436408996582,
      "learning_rate": 6.466250186922325e-05,
      "loss": 6.5904,
      "step": 87
    },
    {
      "epoch": 0.04632188445848138,
      "grad_norm": 13.513579368591309,
      "learning_rate": 6.387014543809223e-05,
      "loss": 6.675,
      "step": 88
    },
    {
      "epoch": 0.04684826950914594,
      "grad_norm": 13.96888256072998,
      "learning_rate": 6.307399704769099e-05,
      "loss": 6.8751,
      "step": 89
    },
    {
      "epoch": 0.0473746545598105,
      "grad_norm": 12.633319854736328,
      "learning_rate": 6.227427435703997e-05,
      "loss": 6.353,
      "step": 90
    },
    {
      "epoch": 0.047901039610475066,
      "grad_norm": 13.461629867553711,
      "learning_rate": 6.147119600233758e-05,
      "loss": 6.6228,
      "step": 91
    },
    {
      "epoch": 0.048427424661139626,
      "grad_norm": 12.572783470153809,
      "learning_rate": 6.066498153718735e-05,
      "loss": 6.7246,
      "step": 92
    },
    {
      "epoch": 0.048953809711804186,
      "grad_norm": 13.10534381866455,
      "learning_rate": 5.985585137257401e-05,
      "loss": 6.7975,
      "step": 93
    },
    {
      "epoch": 0.049480194762468746,
      "grad_norm": 11.880861282348633,
      "learning_rate": 5.90440267166055e-05,
      "loss": 6.4121,
      "step": 94
    },
    {
      "epoch": 0.050006579813133306,
      "grad_norm": 14.787824630737305,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 7.202,
      "step": 95
    },
    {
      "epoch": 0.050532964863797866,
      "grad_norm": 11.559120178222656,
      "learning_rate": 5.74131823855921e-05,
      "loss": 6.5163,
      "step": 96
    },
    {
      "epoch": 0.051059349914462426,
      "grad_norm": 11.575690269470215,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 6.5387,
      "step": 97
    },
    {
      "epoch": 0.05158573496512699,
      "grad_norm": 13.372803688049316,
      "learning_rate": 5.577423184847932e-05,
      "loss": 6.9658,
      "step": 98
    },
    {
      "epoch": 0.05211212001579155,
      "grad_norm": 11.88024616241455,
      "learning_rate": 5.495227651252315e-05,
      "loss": 6.3991,
      "step": 99
    },
    {
      "epoch": 0.052638505066456114,
      "grad_norm": 11.660654067993164,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 6.3039,
      "step": 100
    },
    {
      "epoch": 0.052638505066456114,
      "eval_loss": 1.4342728853225708,
      "eval_runtime": 310.4236,
      "eval_samples_per_second": 10.308,
      "eval_steps_per_second": 2.577,
      "step": 100
    },
    {
      "epoch": 0.053164890117120674,
      "grad_norm": 20.367204666137695,
      "learning_rate": 5.330452921628497e-05,
      "loss": 8.8982,
      "step": 101
    },
    {
      "epoch": 0.053691275167785234,
      "grad_norm": 18.706098556518555,
      "learning_rate": 5.247918773366112e-05,
      "loss": 6.6485,
      "step": 102
    },
    {
      "epoch": 0.054217660218449794,
      "grad_norm": 14.060456275939941,
      "learning_rate": 5.165316846586541e-05,
      "loss": 5.2166,
      "step": 103
    },
    {
      "epoch": 0.054744045269114354,
      "grad_norm": 17.20137596130371,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 6.642,
      "step": 104
    },
    {
      "epoch": 0.05527043031977892,
      "grad_norm": 11.964149475097656,
      "learning_rate": 5e-05,
      "loss": 5.254,
      "step": 105
    },
    {
      "epoch": 0.05579681537044348,
      "grad_norm": 11.72870922088623,
      "learning_rate": 4.917330276168208e-05,
      "loss": 5.2702,
      "step": 106
    },
    {
      "epoch": 0.05632320042110804,
      "grad_norm": 11.874157905578613,
      "learning_rate": 4.834683153413459e-05,
      "loss": 4.67,
      "step": 107
    },
    {
      "epoch": 0.0568495854717726,
      "grad_norm": 10.981062889099121,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 4.426,
      "step": 108
    },
    {
      "epoch": 0.05737597052243716,
      "grad_norm": 11.234898567199707,
      "learning_rate": 4.669547078371504e-05,
      "loss": 5.3235,
      "step": 109
    },
    {
      "epoch": 0.05790235557310172,
      "grad_norm": 12.923269271850586,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 5.076,
      "step": 110
    },
    {
      "epoch": 0.05842874062376629,
      "grad_norm": 12.604551315307617,
      "learning_rate": 4.504772348747687e-05,
      "loss": 4.8287,
      "step": 111
    },
    {
      "epoch": 0.05895512567443085,
      "grad_norm": 12.087059020996094,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 4.7475,
      "step": 112
    },
    {
      "epoch": 0.05948151072509541,
      "grad_norm": 11.601035118103027,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 4.3745,
      "step": 113
    },
    {
      "epoch": 0.06000789577575997,
      "grad_norm": 11.774445533752441,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 3.7109,
      "step": 114
    },
    {
      "epoch": 0.06053428082642453,
      "grad_norm": 11.968523979187012,
      "learning_rate": 4.17702704859633e-05,
      "loss": 4.1646,
      "step": 115
    },
    {
      "epoch": 0.06106066587708909,
      "grad_norm": 11.92697525024414,
      "learning_rate": 4.095597328339452e-05,
      "loss": 3.759,
      "step": 116
    },
    {
      "epoch": 0.06158705092775365,
      "grad_norm": 11.931136131286621,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 4.3903,
      "step": 117
    },
    {
      "epoch": 0.062113435978418216,
      "grad_norm": 13.77640151977539,
      "learning_rate": 3.933501846281267e-05,
      "loss": 5.2051,
      "step": 118
    },
    {
      "epoch": 0.06263982102908278,
      "grad_norm": 12.995139122009277,
      "learning_rate": 3.852880399766243e-05,
      "loss": 5.2394,
      "step": 119
    },
    {
      "epoch": 0.06316620607974734,
      "grad_norm": 14.54707145690918,
      "learning_rate": 3.772572564296005e-05,
      "loss": 5.6322,
      "step": 120
    },
    {
      "epoch": 0.0636925911304119,
      "grad_norm": 11.94774055480957,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 4.3082,
      "step": 121
    },
    {
      "epoch": 0.06421897618107646,
      "grad_norm": 11.902386665344238,
      "learning_rate": 3.612985456190778e-05,
      "loss": 4.8839,
      "step": 122
    },
    {
      "epoch": 0.06474536123174102,
      "grad_norm": 13.23066520690918,
      "learning_rate": 3.533749813077677e-05,
      "loss": 5.3997,
      "step": 123
    },
    {
      "epoch": 0.06527174628240558,
      "grad_norm": 11.977374076843262,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 5.5766,
      "step": 124
    },
    {
      "epoch": 0.06579813133307014,
      "grad_norm": 13.329026222229004,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 5.5163,
      "step": 125
    },
    {
      "epoch": 0.0663245163837347,
      "grad_norm": 11.015464782714844,
      "learning_rate": 3.298534127791785e-05,
      "loss": 4.8713,
      "step": 126
    },
    {
      "epoch": 0.06685090143439926,
      "grad_norm": 11.989096641540527,
      "learning_rate": 3.221030765387417e-05,
      "loss": 5.1716,
      "step": 127
    },
    {
      "epoch": 0.06737728648506383,
      "grad_norm": 11.40512466430664,
      "learning_rate": 3.144013755408895e-05,
      "loss": 4.6448,
      "step": 128
    },
    {
      "epoch": 0.06790367153572839,
      "grad_norm": 13.799883842468262,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 4.7036,
      "step": 129
    },
    {
      "epoch": 0.06843005658639295,
      "grad_norm": 11.917877197265625,
      "learning_rate": 2.991522876735154e-05,
      "loss": 5.3958,
      "step": 130
    },
    {
      "epoch": 0.06895644163705751,
      "grad_norm": 11.720928192138672,
      "learning_rate": 2.916090697523549e-05,
      "loss": 5.4023,
      "step": 131
    },
    {
      "epoch": 0.06948282668772207,
      "grad_norm": 12.156689643859863,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 5.8364,
      "step": 132
    },
    {
      "epoch": 0.07000921173838663,
      "grad_norm": 12.81257438659668,
      "learning_rate": 2.766955965735968e-05,
      "loss": 5.9446,
      "step": 133
    },
    {
      "epoch": 0.07053559678905119,
      "grad_norm": 13.053471565246582,
      "learning_rate": 2.693294185106562e-05,
      "loss": 6.1621,
      "step": 134
    },
    {
      "epoch": 0.07106198183971575,
      "grad_norm": 11.917896270751953,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 6.2142,
      "step": 135
    },
    {
      "epoch": 0.07158836689038031,
      "grad_norm": 13.821593284606934,
      "learning_rate": 2.547882480847461e-05,
      "loss": 6.2556,
      "step": 136
    },
    {
      "epoch": 0.07211475194104487,
      "grad_norm": 12.433488845825195,
      "learning_rate": 2.476172311325783e-05,
      "loss": 6.4588,
      "step": 137
    },
    {
      "epoch": 0.07264113699170943,
      "grad_norm": 12.372336387634277,
      "learning_rate": 2.405152131093926e-05,
      "loss": 6.2338,
      "step": 138
    },
    {
      "epoch": 0.07316752204237399,
      "grad_norm": 10.557406425476074,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 5.8078,
      "step": 139
    },
    {
      "epoch": 0.07369390709303855,
      "grad_norm": 11.982460975646973,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 6.768,
      "step": 140
    },
    {
      "epoch": 0.07422029214370311,
      "grad_norm": 11.688639640808105,
      "learning_rate": 2.196424713241637e-05,
      "loss": 5.6643,
      "step": 141
    },
    {
      "epoch": 0.07474667719436769,
      "grad_norm": 12.019471168518066,
      "learning_rate": 2.128356686585282e-05,
      "loss": 6.3878,
      "step": 142
    },
    {
      "epoch": 0.07527306224503225,
      "grad_norm": 11.111176490783691,
      "learning_rate": 2.061073738537635e-05,
      "loss": 6.4321,
      "step": 143
    },
    {
      "epoch": 0.0757994472956968,
      "grad_norm": 10.541357040405273,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 6.4127,
      "step": 144
    },
    {
      "epoch": 0.07632583234636137,
      "grad_norm": 12.148856163024902,
      "learning_rate": 1.928936436551661e-05,
      "loss": 6.2329,
      "step": 145
    },
    {
      "epoch": 0.07685221739702593,
      "grad_norm": 12.134868621826172,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 6.2659,
      "step": 146
    },
    {
      "epoch": 0.07737860244769049,
      "grad_norm": 11.201704025268555,
      "learning_rate": 1.800157297483417e-05,
      "loss": 5.8522,
      "step": 147
    },
    {
      "epoch": 0.07790498749835505,
      "grad_norm": 11.124740600585938,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 6.5159,
      "step": 148
    },
    {
      "epoch": 0.0784313725490196,
      "grad_norm": 11.063993453979492,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 6.2262,
      "step": 149
    },
    {
      "epoch": 0.07895775759968417,
      "grad_norm": 14.342657089233398,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 6.7652,
      "step": 150
    },
    {
      "epoch": 0.07895775759968417,
      "eval_loss": 1.3660376071929932,
      "eval_runtime": 310.4891,
      "eval_samples_per_second": 10.306,
      "eval_steps_per_second": 2.577,
      "step": 150
    },
    {
      "epoch": 0.07948414265034873,
      "grad_norm": 16.622774124145508,
      "learning_rate": 1.553232954407171e-05,
      "loss": 6.6075,
      "step": 151
    },
    {
      "epoch": 0.08001052770101329,
      "grad_norm": 14.846519470214844,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 5.52,
      "step": 152
    },
    {
      "epoch": 0.08053691275167785,
      "grad_norm": 12.70969009399414,
      "learning_rate": 1.435357758543015e-05,
      "loss": 4.9792,
      "step": 153
    },
    {
      "epoch": 0.08106329780234241,
      "grad_norm": 13.970699310302734,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 5.0739,
      "step": 154
    },
    {
      "epoch": 0.08158968285300698,
      "grad_norm": 12.604689598083496,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 4.1007,
      "step": 155
    },
    {
      "epoch": 0.08211606790367154,
      "grad_norm": 11.775655746459961,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 4.2018,
      "step": 156
    },
    {
      "epoch": 0.0826424529543361,
      "grad_norm": 11.937134742736816,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 4.6492,
      "step": 157
    },
    {
      "epoch": 0.08316883800500066,
      "grad_norm": 13.603174209594727,
      "learning_rate": 1.157994445715706e-05,
      "loss": 4.7931,
      "step": 158
    },
    {
      "epoch": 0.08369522305566522,
      "grad_norm": 13.220638275146484,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 4.8519,
      "step": 159
    },
    {
      "epoch": 0.08422160810632978,
      "grad_norm": 10.09309196472168,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 3.647,
      "step": 160
    },
    {
      "epoch": 0.08474799315699434,
      "grad_norm": 12.403929710388184,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 4.4756,
      "step": 161
    },
    {
      "epoch": 0.0852743782076589,
      "grad_norm": 13.083516120910645,
      "learning_rate": 9.549150281252633e-06,
      "loss": 5.1003,
      "step": 162
    },
    {
      "epoch": 0.08580076325832346,
      "grad_norm": 12.505382537841797,
      "learning_rate": 9.068759265665384e-06,
      "loss": 4.7213,
      "step": 163
    },
    {
      "epoch": 0.08632714830898802,
      "grad_norm": 13.353013038635254,
      "learning_rate": 8.599558442598998e-06,
      "loss": 5.4313,
      "step": 164
    },
    {
      "epoch": 0.08685353335965258,
      "grad_norm": 11.2360258102417,
      "learning_rate": 8.141676086873572e-06,
      "loss": 4.4819,
      "step": 165
    },
    {
      "epoch": 0.08737991841031714,
      "grad_norm": 11.619230270385742,
      "learning_rate": 7.695237378953223e-06,
      "loss": 4.7151,
      "step": 166
    },
    {
      "epoch": 0.0879063034609817,
      "grad_norm": 12.413002014160156,
      "learning_rate": 7.260364370723044e-06,
      "loss": 4.2748,
      "step": 167
    },
    {
      "epoch": 0.08843268851164626,
      "grad_norm": 12.870382308959961,
      "learning_rate": 6.837175952121306e-06,
      "loss": 4.6202,
      "step": 168
    },
    {
      "epoch": 0.08895907356231084,
      "grad_norm": 12.119048118591309,
      "learning_rate": 6.425787818636131e-06,
      "loss": 5.0414,
      "step": 169
    },
    {
      "epoch": 0.0894854586129754,
      "grad_norm": 13.448798179626465,
      "learning_rate": 6.026312439675552e-06,
      "loss": 4.5453,
      "step": 170
    },
    {
      "epoch": 0.09001184366363996,
      "grad_norm": 10.608859062194824,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 4.1188,
      "step": 171
    },
    {
      "epoch": 0.09053822871430452,
      "grad_norm": 11.093206405639648,
      "learning_rate": 5.263533508961827e-06,
      "loss": 4.611,
      "step": 172
    },
    {
      "epoch": 0.09106461376496908,
      "grad_norm": 9.835367202758789,
      "learning_rate": 4.900438493352055e-06,
      "loss": 3.3329,
      "step": 173
    },
    {
      "epoch": 0.09159099881563364,
      "grad_norm": 10.847572326660156,
      "learning_rate": 4.549673247541875e-06,
      "loss": 4.6726,
      "step": 174
    },
    {
      "epoch": 0.0921173838662982,
      "grad_norm": 11.087268829345703,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 4.0335,
      "step": 175
    },
    {
      "epoch": 0.09264376891696276,
      "grad_norm": 12.334137916564941,
      "learning_rate": 3.885512251130763e-06,
      "loss": 4.7912,
      "step": 176
    },
    {
      "epoch": 0.09317015396762732,
      "grad_norm": 13.143092155456543,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 5.2709,
      "step": 177
    },
    {
      "epoch": 0.09369653901829188,
      "grad_norm": 12.453194618225098,
      "learning_rate": 3.271776770026963e-06,
      "loss": 5.4586,
      "step": 178
    },
    {
      "epoch": 0.09422292406895644,
      "grad_norm": 12.238334655761719,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 3.9605,
      "step": 179
    },
    {
      "epoch": 0.094749309119621,
      "grad_norm": 12.200905799865723,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 5.4847,
      "step": 180
    },
    {
      "epoch": 0.09527569417028556,
      "grad_norm": 12.460379600524902,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 4.9445,
      "step": 181
    },
    {
      "epoch": 0.09580207922095013,
      "grad_norm": 12.011752128601074,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 5.0377,
      "step": 182
    },
    {
      "epoch": 0.09632846427161469,
      "grad_norm": 12.021615028381348,
      "learning_rate": 1.962316193157593e-06,
      "loss": 5.4708,
      "step": 183
    },
    {
      "epoch": 0.09685484932227925,
      "grad_norm": 12.846208572387695,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 6.1394,
      "step": 184
    },
    {
      "epoch": 0.09738123437294381,
      "grad_norm": 12.173249244689941,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 6.0229,
      "step": 185
    },
    {
      "epoch": 0.09790761942360837,
      "grad_norm": 11.38800048828125,
      "learning_rate": 1.333670137599713e-06,
      "loss": 6.2028,
      "step": 186
    },
    {
      "epoch": 0.09843400447427293,
      "grad_norm": 12.05564022064209,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 6.3845,
      "step": 187
    },
    {
      "epoch": 0.09896038952493749,
      "grad_norm": 12.936962127685547,
      "learning_rate": 9.810017062595322e-07,
      "loss": 6.1623,
      "step": 188
    },
    {
      "epoch": 0.09948677457560205,
      "grad_norm": 12.60428524017334,
      "learning_rate": 8.247462563808817e-07,
      "loss": 6.1015,
      "step": 189
    },
    {
      "epoch": 0.10001315962626661,
      "grad_norm": 11.307320594787598,
      "learning_rate": 6.819348298638839e-07,
      "loss": 5.8205,
      "step": 190
    },
    {
      "epoch": 0.10053954467693117,
      "grad_norm": 12.578415870666504,
      "learning_rate": 5.526064699265753e-07,
      "loss": 6.678,
      "step": 191
    },
    {
      "epoch": 0.10106592972759573,
      "grad_norm": 10.613789558410645,
      "learning_rate": 4.367965336512403e-07,
      "loss": 5.7437,
      "step": 192
    },
    {
      "epoch": 0.10159231477826029,
      "grad_norm": 11.73043441772461,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 6.6499,
      "step": 193
    },
    {
      "epoch": 0.10211869982892485,
      "grad_norm": 10.590353012084961,
      "learning_rate": 2.458548727494292e-07,
      "loss": 6.2539,
      "step": 194
    },
    {
      "epoch": 0.10264508487958941,
      "grad_norm": 12.387429237365723,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 6.4081,
      "step": 195
    },
    {
      "epoch": 0.10317146993025399,
      "grad_norm": 11.481491088867188,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 5.98,
      "step": 196
    },
    {
      "epoch": 0.10369785498091855,
      "grad_norm": 13.055595397949219,
      "learning_rate": 6.150154258476315e-08,
      "loss": 5.9356,
      "step": 197
    },
    {
      "epoch": 0.1042242400315831,
      "grad_norm": 12.20886516571045,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 5.8957,
      "step": 198
    },
    {
      "epoch": 0.10475062508224767,
      "grad_norm": 12.601096153259277,
      "learning_rate": 6.834750376549792e-09,
      "loss": 6.1263,
      "step": 199
    },
    {
      "epoch": 0.10527701013291223,
      "grad_norm": 13.229594230651855,
      "learning_rate": 0.0,
      "loss": 6.3569,
      "step": 200
    },
    {
      "epoch": 0.10527701013291223,
      "eval_loss": 1.3446228504180908,
      "eval_runtime": 311.7876,
      "eval_samples_per_second": 10.263,
      "eval_steps_per_second": 2.566,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2.897770968199987e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}