{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 19.999910828404126,
  "global_step": 168200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.06,
      "learning_rate": 2.3305588585017838e-07,
      "loss": 1.5468,
      "step": 500
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.7086801426872774e-07,
      "loss": 1.1544,
      "step": 1000
    },
    {
      "epoch": 0.18,
      "learning_rate": 7.082045184304399e-07,
      "loss": 1.0372,
      "step": 1500
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.460166468489894e-07,
      "loss": 0.9823,
      "step": 2000
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.1838287752675386e-06,
      "loss": 0.9462,
      "step": 2500
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4216409036860881e-06,
      "loss": 0.9246,
      "step": 3000
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.6594530321046376e-06,
      "loss": 0.8938,
      "step": 3500
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.8972651605231867e-06,
      "loss": 0.8919,
      "step": 4000
    },
    {
      "epoch": 0.54,
      "learning_rate": 2.135077288941736e-06,
      "loss": 0.863,
      "step": 4500
    },
    {
      "epoch": 0.59,
      "learning_rate": 2.3728894173602857e-06,
      "loss": 0.8513,
      "step": 5000
    },
    {
      "epoch": 0.65,
      "learning_rate": 2.6107015457788348e-06,
      "loss": 0.8337,
      "step": 5500
    },
    {
      "epoch": 0.71,
      "learning_rate": 2.8485136741973847e-06,
      "loss": 0.8228,
      "step": 6000
    },
    {
      "epoch": 0.77,
      "learning_rate": 3.0863258026159338e-06,
      "loss": 0.787,
      "step": 6500
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.324137931034483e-06,
      "loss": 0.7696,
      "step": 7000
    },
    {
      "epoch": 0.89,
      "learning_rate": 3.5619500594530328e-06,
      "loss": 0.7353,
      "step": 7500
    },
    {
      "epoch": 0.95,
      "learning_rate": 3.799762187871582e-06,
      "loss": 0.7291,
      "step": 8000
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.037574316290131e-06,
      "loss": 0.7122,
      "step": 8500
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.27538644470868e-06,
      "loss": 0.7098,
      "step": 9000
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.5131985731272295e-06,
      "loss": 0.6884,
      "step": 9500
    },
    {
      "epoch": 1.19,
      "learning_rate": 4.751010701545779e-06,
      "loss": 0.678,
      "step": 10000
    },
    {
      "epoch": 1.25,
      "learning_rate": 4.9888228299643285e-06,
      "loss": 0.6688,
      "step": 10500
    },
    {
      "epoch": 1.31,
      "learning_rate": 5.2266349583828776e-06,
      "loss": 0.6706,
      "step": 11000
    },
    {
      "epoch": 1.37,
      "learning_rate": 5.4644470868014275e-06,
      "loss": 0.6705,
      "step": 11500
    },
    {
      "epoch": 1.43,
      "learning_rate": 5.7022592152199766e-06,
      "loss": 0.6593,
      "step": 12000
    },
    {
      "epoch": 1.49,
      "learning_rate": 5.940071343638526e-06,
      "loss": 0.669,
      "step": 12500
    },
    {
      "epoch": 1.55,
      "learning_rate": 6.1778834720570756e-06,
      "loss": 0.6495,
      "step": 13000
    },
    {
      "epoch": 1.61,
      "learning_rate": 6.415695600475625e-06,
      "loss": 0.6554,
      "step": 13500
    },
    {
      "epoch": 1.66,
      "learning_rate": 6.653507728894174e-06,
      "loss": 0.6467,
      "step": 14000
    },
    {
      "epoch": 1.72,
      "learning_rate": 6.891319857312724e-06,
      "loss": 0.6479,
      "step": 14500
    },
    {
      "epoch": 1.78,
      "learning_rate": 7.128180737217598e-06,
      "loss": 0.643,
      "step": 15000
    },
    {
      "epoch": 1.84,
      "learning_rate": 7.365992865636148e-06,
      "loss": 0.6369,
      "step": 15500
    },
    {
      "epoch": 1.9,
      "learning_rate": 7.603804994054697e-06,
      "loss": 0.6363,
      "step": 16000
    },
    {
      "epoch": 1.96,
      "learning_rate": 7.841617122473245e-06,
      "loss": 0.6272,
      "step": 16500
    },
    {
      "epoch": 2.02,
      "learning_rate": 8.079429250891796e-06,
      "loss": 0.6269,
      "step": 17000
    },
    {
      "epoch": 2.08,
      "learning_rate": 8.316765755053508e-06,
      "loss": 0.6116,
      "step": 17500
    },
    {
      "epoch": 2.14,
      "learning_rate": 8.554577883472059e-06,
      "loss": 0.6149,
      "step": 18000
    },
    {
      "epoch": 2.2,
      "learning_rate": 8.792390011890606e-06,
      "loss": 0.6143,
      "step": 18500
    },
    {
      "epoch": 2.26,
      "learning_rate": 9.030202140309157e-06,
      "loss": 0.6139,
      "step": 19000
    },
    {
      "epoch": 2.32,
      "learning_rate": 9.268014268727706e-06,
      "loss": 0.6048,
      "step": 19500
    },
    {
      "epoch": 2.38,
      "learning_rate": 9.505826397146255e-06,
      "loss": 0.6095,
      "step": 20000
    },
    {
      "epoch": 2.44,
      "learning_rate": 9.743638525564804e-06,
      "loss": 0.5988,
      "step": 20500
    },
    {
      "epoch": 2.5,
      "learning_rate": 9.981450653983355e-06,
      "loss": 0.595,
      "step": 21000
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.0219262782401904e-05,
      "loss": 0.6005,
      "step": 21500
    },
    {
      "epoch": 2.62,
      "learning_rate": 1.0457074910820453e-05,
      "loss": 0.598,
      "step": 22000
    },
    {
      "epoch": 2.68,
      "learning_rate": 1.0694411414982165e-05,
      "loss": 0.5968,
      "step": 22500
    },
    {
      "epoch": 2.73,
      "learning_rate": 1.0932223543400716e-05,
      "loss": 0.5934,
      "step": 23000
    },
    {
      "epoch": 2.79,
      "learning_rate": 1.1170035671819263e-05,
      "loss": 0.5965,
      "step": 23500
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.1407847800237812e-05,
      "loss": 0.5946,
      "step": 24000
    },
    {
      "epoch": 2.91,
      "learning_rate": 1.1645659928656363e-05,
      "loss": 0.5904,
      "step": 24500
    },
    {
      "epoch": 2.97,
      "learning_rate": 1.1882996432818075e-05,
      "loss": 0.5917,
      "step": 25000
    },
    {
      "epoch": 3.03,
      "learning_rate": 1.2120808561236624e-05,
      "loss": 0.5802,
      "step": 25500
    },
    {
      "epoch": 3.09,
      "learning_rate": 1.2358620689655175e-05,
      "loss": 0.5753,
      "step": 26000
    },
    {
      "epoch": 3.15,
      "learning_rate": 1.2595957193816886e-05,
      "loss": 0.5687,
      "step": 26500
    },
    {
      "epoch": 3.21,
      "learning_rate": 1.2833769322235434e-05,
      "loss": 0.5755,
      "step": 27000
    },
    {
      "epoch": 3.27,
      "learning_rate": 1.3071581450653985e-05,
      "loss": 0.5718,
      "step": 27500
    },
    {
      "epoch": 3.33,
      "learning_rate": 1.3309393579072534e-05,
      "loss": 0.5759,
      "step": 28000
    },
    {
      "epoch": 3.39,
      "learning_rate": 1.3547205707491083e-05,
      "loss": 0.5748,
      "step": 28500
    },
    {
      "epoch": 3.45,
      "learning_rate": 1.3785017835909633e-05,
      "loss": 0.571,
      "step": 29000
    },
    {
      "epoch": 3.51,
      "learning_rate": 1.4022829964328183e-05,
      "loss": 0.5713,
      "step": 29500
    },
    {
      "epoch": 3.57,
      "learning_rate": 1.426064209274673e-05,
      "loss": 0.5688,
      "step": 30000
    },
    {
      "epoch": 3.63,
      "learning_rate": 1.4497978596908443e-05,
      "loss": 0.5744,
      "step": 30500
    },
    {
      "epoch": 3.69,
      "learning_rate": 1.4735790725326993e-05,
      "loss": 0.573,
      "step": 31000
    },
    {
      "epoch": 3.75,
      "learning_rate": 1.4973602853745542e-05,
      "loss": 0.5711,
      "step": 31500
    },
    {
      "epoch": 3.8,
      "learning_rate": 1.5211414982164092e-05,
      "loss": 0.5679,
      "step": 32000
    },
    {
      "epoch": 3.86,
      "learning_rate": 1.544922711058264e-05,
      "loss": 0.5702,
      "step": 32500
    },
    {
      "epoch": 3.92,
      "learning_rate": 1.568703923900119e-05,
      "loss": 0.57,
      "step": 33000
    },
    {
      "epoch": 3.98,
      "learning_rate": 1.592485136741974e-05,
      "loss": 0.5659,
      "step": 33500
    },
    {
      "epoch": 4.04,
      "learning_rate": 1.616266349583829e-05,
      "loss": 0.5553,
      "step": 34000
    },
    {
      "epoch": 4.1,
      "learning_rate": 1.6400475624256836e-05,
      "loss": 0.5508,
      "step": 34500
    },
    {
      "epoch": 4.16,
      "learning_rate": 1.663828775267539e-05,
      "loss": 0.5466,
      "step": 35000
    },
    {
      "epoch": 4.22,
      "learning_rate": 1.6876099881093938e-05,
      "loss": 0.5479,
      "step": 35500
    },
    {
      "epoch": 4.28,
      "learning_rate": 1.7113912009512487e-05,
      "loss": 0.5472,
      "step": 36000
    },
    {
      "epoch": 4.34,
      "learning_rate": 1.73512485136742e-05,
      "loss": 0.5462,
      "step": 36500
    },
    {
      "epoch": 4.4,
      "learning_rate": 1.758858501783591e-05,
      "loss": 0.5517,
      "step": 37000
    },
    {
      "epoch": 4.46,
      "learning_rate": 1.782639714625446e-05,
      "loss": 0.5448,
      "step": 37500
    },
    {
      "epoch": 4.52,
      "learning_rate": 1.8064209274673012e-05,
      "loss": 0.5473,
      "step": 38000
    },
    {
      "epoch": 4.58,
      "learning_rate": 1.8302021403091558e-05,
      "loss": 0.547,
      "step": 38500
    },
    {
      "epoch": 4.64,
      "learning_rate": 1.8539833531510107e-05,
      "loss": 0.5481,
      "step": 39000
    },
    {
      "epoch": 4.7,
      "learning_rate": 1.877764565992866e-05,
      "loss": 0.5534,
      "step": 39500
    },
    {
      "epoch": 4.76,
      "learning_rate": 1.901498216409037e-05,
      "loss": 0.5456,
      "step": 40000
    },
    {
      "epoch": 4.82,
      "learning_rate": 1.925279429250892e-05,
      "loss": 0.5543,
      "step": 40500
    },
    {
      "epoch": 4.88,
      "learning_rate": 1.949060642092747e-05,
      "loss": 0.5462,
      "step": 41000
    },
    {
      "epoch": 4.93,
      "learning_rate": 1.9728418549346018e-05,
      "loss": 0.5443,
      "step": 41500
    },
    {
      "epoch": 4.99,
      "learning_rate": 1.9966230677764567e-05,
      "loss": 0.5429,
      "step": 42000
    },
    {
      "epoch": 5.05,
      "learning_rate": 1.9977328577090766e-05,
      "loss": 0.5261,
      "step": 42500
    },
    {
      "epoch": 5.11,
      "learning_rate": 1.9950905007266485e-05,
      "loss": 0.521,
      "step": 43000
    },
    {
      "epoch": 5.17,
      "learning_rate": 1.99244814374422e-05,
      "loss": 0.5263,
      "step": 43500
    },
    {
      "epoch": 5.23,
      "learning_rate": 1.9898057867617915e-05,
      "loss": 0.5268,
      "step": 44000
    },
    {
      "epoch": 5.29,
      "learning_rate": 1.9871634297793634e-05,
      "loss": 0.5232,
      "step": 44500
    },
    {
      "epoch": 5.35,
      "learning_rate": 1.9845263575108998e-05,
      "loss": 0.5304,
      "step": 45000
    },
    {
      "epoch": 5.41,
      "learning_rate": 1.9818840005284717e-05,
      "loss": 0.52,
      "step": 45500
    },
    {
      "epoch": 5.47,
      "learning_rate": 1.9792416435460432e-05,
      "loss": 0.5266,
      "step": 46000
    },
    {
      "epoch": 5.53,
      "learning_rate": 1.976599286563615e-05,
      "loss": 0.5311,
      "step": 46500
    },
    {
      "epoch": 5.59,
      "learning_rate": 1.9739622142951514e-05,
      "loss": 0.526,
      "step": 47000
    },
    {
      "epoch": 5.65,
      "learning_rate": 1.971319857312723e-05,
      "loss": 0.5272,
      "step": 47500
    },
    {
      "epoch": 5.71,
      "learning_rate": 1.9686880697582245e-05,
      "loss": 0.5259,
      "step": 48000
    },
    {
      "epoch": 5.77,
      "learning_rate": 1.966045712775796e-05,
      "loss": 0.5324,
      "step": 48500
    },
    {
      "epoch": 5.83,
      "learning_rate": 1.963403355793368e-05,
      "loss": 0.532,
      "step": 49000
    },
    {
      "epoch": 5.89,
      "learning_rate": 1.9607609988109395e-05,
      "loss": 0.525,
      "step": 49500
    },
    {
      "epoch": 5.95,
      "learning_rate": 1.9581186418285113e-05,
      "loss": 0.528,
      "step": 50000
    },
    {
      "epoch": 6.0,
      "learning_rate": 1.955476284846083e-05,
      "loss": 0.5291,
      "step": 50500
    },
    {
      "epoch": 6.06,
      "learning_rate": 1.9528339278636547e-05,
      "loss": 0.5051,
      "step": 51000
    },
    {
      "epoch": 6.12,
      "learning_rate": 1.9501915708812262e-05,
      "loss": 0.497,
      "step": 51500
    },
    {
      "epoch": 6.18,
      "learning_rate": 1.9475492138987978e-05,
      "loss": 0.5067,
      "step": 52000
    },
    {
      "epoch": 6.24,
      "learning_rate": 1.9449068569163696e-05,
      "loss": 0.5062,
      "step": 52500
    },
    {
      "epoch": 6.3,
      "learning_rate": 1.942264499933941e-05,
      "loss": 0.5046,
      "step": 53000
    },
    {
      "epoch": 6.36,
      "learning_rate": 1.939622142951513e-05,
      "loss": 0.5094,
      "step": 53500
    },
    {
      "epoch": 6.42,
      "learning_rate": 1.9369797859690846e-05,
      "loss": 0.5061,
      "step": 54000
    },
    {
      "epoch": 6.48,
      "learning_rate": 1.9343374289866564e-05,
      "loss": 0.512,
      "step": 54500
    },
    {
      "epoch": 6.54,
      "learning_rate": 1.931695072004228e-05,
      "loss": 0.5048,
      "step": 55000
    },
    {
      "epoch": 6.6,
      "learning_rate": 1.9290579997357643e-05,
      "loss": 0.5083,
      "step": 55500
    },
    {
      "epoch": 6.66,
      "learning_rate": 1.9264156427533362e-05,
      "loss": 0.5021,
      "step": 56000
    },
    {
      "epoch": 6.72,
      "learning_rate": 1.9237732857709077e-05,
      "loss": 0.5067,
      "step": 56500
    },
    {
      "epoch": 6.78,
      "learning_rate": 1.9211309287884793e-05,
      "loss": 0.5107,
      "step": 57000
    },
    {
      "epoch": 6.84,
      "learning_rate": 1.918488571806051e-05,
      "loss": 0.5121,
      "step": 57500
    },
    {
      "epoch": 6.9,
      "learning_rate": 1.9158462148236226e-05,
      "loss": 0.5092,
      "step": 58000
    },
    {
      "epoch": 6.96,
      "learning_rate": 1.9132091425551594e-05,
      "loss": 0.5106,
      "step": 58500
    },
    {
      "epoch": 7.02,
      "learning_rate": 1.910566785572731e-05,
      "loss": 0.5081,
      "step": 59000
    },
    {
      "epoch": 7.07,
      "learning_rate": 1.9079244285903028e-05,
      "loss": 0.4826,
      "step": 59500
    },
    {
      "epoch": 7.13,
      "learning_rate": 1.905287356321839e-05,
      "loss": 0.4854,
      "step": 60000
    },
    {
      "epoch": 7.19,
      "learning_rate": 1.902644999339411e-05,
      "loss": 0.48,
      "step": 60500
    },
    {
      "epoch": 7.25,
      "learning_rate": 1.9000026423569825e-05,
      "loss": 0.4918,
      "step": 61000
    },
    {
      "epoch": 7.31,
      "learning_rate": 1.8973602853745544e-05,
      "loss": 0.4867,
      "step": 61500
    },
    {
      "epoch": 7.37,
      "learning_rate": 1.8947232131060908e-05,
      "loss": 0.4935,
      "step": 62000
    },
    {
      "epoch": 7.43,
      "learning_rate": 1.8920808561236626e-05,
      "loss": 0.4855,
      "step": 62500
    },
    {
      "epoch": 7.49,
      "learning_rate": 1.8894384991412342e-05,
      "loss": 0.4907,
      "step": 63000
    },
    {
      "epoch": 7.55,
      "learning_rate": 1.886796142158806e-05,
      "loss": 0.4913,
      "step": 63500
    },
    {
      "epoch": 7.61,
      "learning_rate": 1.8841537851763776e-05,
      "loss": 0.4914,
      "step": 64000
    },
    {
      "epoch": 7.67,
      "learning_rate": 1.881511428193949e-05,
      "loss": 0.4962,
      "step": 64500
    },
    {
      "epoch": 7.73,
      "learning_rate": 1.8788690712115206e-05,
      "loss": 0.4904,
      "step": 65000
    },
    {
      "epoch": 7.79,
      "learning_rate": 1.8762267142290925e-05,
      "loss": 0.4927,
      "step": 65500
    },
    {
      "epoch": 7.85,
      "learning_rate": 1.873584357246664e-05,
      "loss": 0.4934,
      "step": 66000
    },
    {
      "epoch": 7.91,
      "learning_rate": 1.8709472849782007e-05,
      "loss": 0.4926,
      "step": 66500
    },
    {
      "epoch": 7.97,
      "learning_rate": 1.8683049279957723e-05,
      "loss": 0.4965,
      "step": 67000
    },
    {
      "epoch": 8.03,
      "learning_rate": 1.865662571013344e-05,
      "loss": 0.4783,
      "step": 67500
    },
    {
      "epoch": 8.09,
      "learning_rate": 1.8630202140309157e-05,
      "loss": 0.4617,
      "step": 68000
    },
    {
      "epoch": 8.15,
      "learning_rate": 1.8603831417624524e-05,
      "loss": 0.4649,
      "step": 68500
    },
    {
      "epoch": 8.2,
      "learning_rate": 1.857740784780024e-05,
      "loss": 0.4696,
      "step": 69000
    },
    {
      "epoch": 8.26,
      "learning_rate": 1.8550984277975958e-05,
      "loss": 0.467,
      "step": 69500
    },
    {
      "epoch": 8.32,
      "learning_rate": 1.8524560708151673e-05,
      "loss": 0.4689,
      "step": 70000
    },
    {
      "epoch": 8.38,
      "learning_rate": 1.849813713832739e-05,
      "loss": 0.4686,
      "step": 70500
    },
    {
      "epoch": 8.44,
      "learning_rate": 1.8471766415642755e-05,
      "loss": 0.476,
      "step": 71000
    },
    {
      "epoch": 8.5,
      "learning_rate": 1.8445342845818474e-05,
      "loss": 0.4721,
      "step": 71500
    },
    {
      "epoch": 8.56,
      "learning_rate": 1.841891927599419e-05,
      "loss": 0.483,
      "step": 72000
    },
    {
      "epoch": 8.62,
      "learning_rate": 1.8392495706169905e-05,
      "loss": 0.4728,
      "step": 72500
    },
    {
      "epoch": 8.68,
      "learning_rate": 1.836607213634562e-05,
      "loss": 0.4742,
      "step": 73000
    },
    {
      "epoch": 8.74,
      "learning_rate": 1.833964856652134e-05,
      "loss": 0.4774,
      "step": 73500
    },
    {
      "epoch": 8.8,
      "learning_rate": 1.8313277843836702e-05,
      "loss": 0.4801,
      "step": 74000
    },
    {
      "epoch": 8.86,
      "learning_rate": 1.828690712115207e-05,
      "loss": 0.4777,
      "step": 74500
    },
    {
      "epoch": 8.92,
      "learning_rate": 1.8260483551327785e-05,
      "loss": 0.4803,
      "step": 75000
    },
    {
      "epoch": 8.98,
      "learning_rate": 1.8234059981503504e-05,
      "loss": 0.4783,
      "step": 75500
    },
    {
      "epoch": 9.04,
      "learning_rate": 1.820763641167922e-05,
      "loss": 0.4596,
      "step": 76000
    },
    {
      "epoch": 9.1,
      "learning_rate": 1.8181212841854937e-05,
      "loss": 0.4442,
      "step": 76500
    },
    {
      "epoch": 9.16,
      "learning_rate": 1.8154789272030653e-05,
      "loss": 0.46,
      "step": 77000
    },
    {
      "epoch": 9.22,
      "learning_rate": 1.812836570220637e-05,
      "loss": 0.4597,
      "step": 77500
    },
    {
      "epoch": 9.27,
      "learning_rate": 1.8101942132382087e-05,
      "loss": 0.4522,
      "step": 78000
    },
    {
      "epoch": 9.33,
      "learning_rate": 1.8075518562557802e-05,
      "loss": 0.4592,
      "step": 78500
    },
    {
      "epoch": 9.39,
      "learning_rate": 1.804914783987317e-05,
      "loss": 0.4617,
      "step": 79000
    },
    {
      "epoch": 9.45,
      "learning_rate": 1.8022724270048884e-05,
      "loss": 0.4601,
      "step": 79500
    },
    {
      "epoch": 9.51,
      "learning_rate": 1.7996300700224603e-05,
      "loss": 0.4606,
      "step": 80000
    },
    {
      "epoch": 9.57,
      "learning_rate": 1.7969877130400318e-05,
      "loss": 0.4616,
      "step": 80500
    },
    {
      "epoch": 9.63,
      "learning_rate": 1.7943506407715682e-05,
      "loss": 0.4566,
      "step": 81000
    },
    {
      "epoch": 9.69,
      "learning_rate": 1.79170828378914e-05,
      "loss": 0.4638,
      "step": 81500
    },
    {
      "epoch": 9.75,
      "learning_rate": 1.7890659268067116e-05,
      "loss": 0.467,
      "step": 82000
    },
    {
      "epoch": 9.81,
      "learning_rate": 1.7864235698242835e-05,
      "loss": 0.4662,
      "step": 82500
    },
    {
      "epoch": 9.87,
      "learning_rate": 1.783781212841855e-05,
      "loss": 0.4636,
      "step": 83000
    },
    {
      "epoch": 9.93,
      "learning_rate": 1.781138855859427e-05,
      "loss": 0.4641,
      "step": 83500
    },
    {
      "epoch": 9.99,
      "learning_rate": 1.7784964988769984e-05,
      "loss": 0.461,
      "step": 84000
    },
    {
      "epoch": 10.05,
      "learning_rate": 1.7758594266085348e-05,
      "loss": 0.4463,
      "step": 84500
    },
    {
      "epoch": 10.11,
      "learning_rate": 1.7732223543400715e-05,
      "loss": 0.4379,
      "step": 85000
    },
    {
      "epoch": 10.17,
      "learning_rate": 1.770579997357643e-05,
      "loss": 0.4418,
      "step": 85500
    },
    {
      "epoch": 10.23,
      "learning_rate": 1.767937640375215e-05,
      "loss": 0.45,
      "step": 86000
    },
    {
      "epoch": 10.29,
      "learning_rate": 1.7652952833927864e-05,
      "loss": 0.4463,
      "step": 86500
    },
    {
      "epoch": 10.34,
      "learning_rate": 1.7626529264103583e-05,
      "loss": 0.4453,
      "step": 87000
    },
    {
      "epoch": 10.4,
      "learning_rate": 1.7600105694279298e-05,
      "loss": 0.4443,
      "step": 87500
    },
    {
      "epoch": 10.46,
      "learning_rate": 1.7573682124455017e-05,
      "loss": 0.4465,
      "step": 88000
    },
    {
      "epoch": 10.52,
      "learning_rate": 1.7547258554630732e-05,
      "loss": 0.4548,
      "step": 88500
    },
    {
      "epoch": 10.58,
      "learning_rate": 1.752083498480645e-05,
      "loss": 0.4485,
      "step": 89000
    },
    {
      "epoch": 10.64,
      "learning_rate": 1.7494411414982166e-05,
      "loss": 0.4437,
      "step": 89500
    },
    {
      "epoch": 10.7,
      "learning_rate": 1.7467987845157885e-05,
      "loss": 0.4508,
      "step": 90000
    },
    {
      "epoch": 10.76,
      "learning_rate": 1.74415642753336e-05,
      "loss": 0.4519,
      "step": 90500
    },
    {
      "epoch": 10.82,
      "learning_rate": 1.7415140705509315e-05,
      "loss": 0.4532,
      "step": 91000
    },
    {
      "epoch": 10.88,
      "learning_rate": 1.738876998282468e-05,
      "loss": 0.4536,
      "step": 91500
    },
    {
      "epoch": 10.94,
      "learning_rate": 1.7362399260140046e-05,
      "loss": 0.4571,
      "step": 92000
    },
    {
      "epoch": 11.0,
      "learning_rate": 1.733597569031576e-05,
      "loss": 0.4471,
      "step": 92500
    },
    {
      "epoch": 11.06,
      "learning_rate": 1.730955212049148e-05,
      "loss": 0.432,
      "step": 93000
    },
    {
      "epoch": 11.12,
      "learning_rate": 1.7283128550667195e-05,
      "loss": 0.4303,
      "step": 93500
    },
    {
      "epoch": 11.18,
      "learning_rate": 1.7256757827982563e-05,
      "loss": 0.4307,
      "step": 94000
    },
    {
      "epoch": 11.24,
      "learning_rate": 1.7230334258158278e-05,
      "loss": 0.4373,
      "step": 94500
    },
    {
      "epoch": 11.3,
      "learning_rate": 1.7203910688333996e-05,
      "loss": 0.4307,
      "step": 95000
    },
    {
      "epoch": 11.36,
      "learning_rate": 1.7177487118509712e-05,
      "loss": 0.4339,
      "step": 95500
    },
    {
      "epoch": 11.41,
      "learning_rate": 1.7151169242964728e-05,
      "loss": 0.4376,
      "step": 96000
    },
    {
      "epoch": 11.47,
      "learning_rate": 1.7124745673140443e-05,
      "loss": 0.4383,
      "step": 96500
    },
    {
      "epoch": 11.53,
      "learning_rate": 1.7098322103316158e-05,
      "loss": 0.4421,
      "step": 97000
    },
    {
      "epoch": 11.59,
      "learning_rate": 1.7071898533491877e-05,
      "loss": 0.4377,
      "step": 97500
    },
    {
      "epoch": 11.65,
      "learning_rate": 1.7045474963667592e-05,
      "loss": 0.4446,
      "step": 98000
    },
    {
      "epoch": 11.71,
      "learning_rate": 1.7019051393843307e-05,
      "loss": 0.4359,
      "step": 98500
    },
    {
      "epoch": 11.77,
      "learning_rate": 1.6992627824019026e-05,
      "loss": 0.4377,
      "step": 99000
    },
    {
      "epoch": 11.83,
      "learning_rate": 1.696620425419474e-05,
      "loss": 0.4412,
      "step": 99500
    },
    {
      "epoch": 11.89,
      "learning_rate": 1.693978068437046e-05,
      "loss": 0.4405,
      "step": 100000
    },
    {
      "epoch": 11.95,
      "learning_rate": 1.6913357114546175e-05,
      "loss": 0.4423,
      "step": 100500
    },
    {
      "epoch": 12.01,
      "learning_rate": 1.6886933544721894e-05,
      "loss": 0.4408,
      "step": 101000
    },
    {
      "epoch": 12.07,
      "learning_rate": 1.686050997489761e-05,
      "loss": 0.417,
      "step": 101500
    },
    {
      "epoch": 12.13,
      "learning_rate": 1.6834086405073328e-05,
      "loss": 0.4241,
      "step": 102000
    },
    {
      "epoch": 12.19,
      "learning_rate": 1.680771568238869e-05,
      "loss": 0.4251,
      "step": 102500
    },
    {
      "epoch": 12.25,
      "learning_rate": 1.678129211256441e-05,
      "loss": 0.4168,
      "step": 103000
    },
    {
      "epoch": 12.31,
      "learning_rate": 1.6754868542740125e-05,
      "loss": 0.421,
      "step": 103500
    },
    {
      "epoch": 12.37,
      "learning_rate": 1.6728444972915844e-05,
      "loss": 0.4247,
      "step": 104000
    },
    {
      "epoch": 12.43,
      "learning_rate": 1.6702074250231208e-05,
      "loss": 0.4226,
      "step": 104500
    },
    {
      "epoch": 12.49,
      "learning_rate": 1.6675650680406927e-05,
      "loss": 0.4232,
      "step": 105000
    },
    {
      "epoch": 12.54,
      "learning_rate": 1.6649227110582642e-05,
      "loss": 0.4287,
      "step": 105500
    },
    {
      "epoch": 12.6,
      "learning_rate": 1.6622803540758357e-05,
      "loss": 0.4273,
      "step": 106000
    },
    {
      "epoch": 12.66,
      "learning_rate": 1.6596379970934076e-05,
      "loss": 0.4346,
      "step": 106500
    },
    {
      "epoch": 12.72,
      "learning_rate": 1.656995640110979e-05,
      "loss": 0.4287,
      "step": 107000
    },
    {
      "epoch": 12.78,
      "learning_rate": 1.6543532831285506e-05,
      "loss": 0.429,
      "step": 107500
    },
    {
      "epoch": 12.84,
      "learning_rate": 1.6517162108600874e-05,
      "loss": 0.4254,
      "step": 108000
    },
    {
      "epoch": 12.9,
      "learning_rate": 1.649073853877659e-05,
      "loss": 0.427,
      "step": 108500
    },
    {
      "epoch": 12.96,
      "learning_rate": 1.6464314968952307e-05,
      "loss": 0.4303,
      "step": 109000
    },
    {
      "epoch": 13.02,
      "learning_rate": 1.6437891399128023e-05,
      "loss": 0.4239,
      "step": 109500
    },
    {
      "epoch": 13.08,
      "learning_rate": 1.641146782930374e-05,
      "loss": 0.401,
      "step": 110000
    },
    {
      "epoch": 13.14,
      "learning_rate": 1.6385097106619105e-05,
      "loss": 0.4092,
      "step": 110500
    },
    {
      "epoch": 13.2,
      "learning_rate": 1.6358726383934472e-05,
      "loss": 0.4076,
      "step": 111000
    },
    {
      "epoch": 13.26,
      "learning_rate": 1.6332355661249836e-05,
      "loss": 0.4118,
      "step": 111500
    },
    {
      "epoch": 13.32,
      "learning_rate": 1.6305932091425555e-05,
      "loss": 0.4145,
      "step": 112000
    },
    {
      "epoch": 13.38,
      "learning_rate": 1.627950852160127e-05,
      "loss": 0.4132,
      "step": 112500
    },
    {
      "epoch": 13.44,
      "learning_rate": 1.625308495177699e-05,
      "loss": 0.416,
      "step": 113000
    },
    {
      "epoch": 13.5,
      "learning_rate": 1.6226661381952704e-05,
      "loss": 0.416,
      "step": 113500
    },
    {
      "epoch": 13.56,
      "learning_rate": 1.620023781212842e-05,
      "loss": 0.4126,
      "step": 114000
    },
    {
      "epoch": 13.61,
      "learning_rate": 1.6173814242304135e-05,
      "loss": 0.4179,
      "step": 114500
    },
    {
      "epoch": 13.67,
      "learning_rate": 1.6147443519619502e-05,
      "loss": 0.4242,
      "step": 115000
    },
    {
      "epoch": 13.73,
      "learning_rate": 1.6121019949795217e-05,
      "loss": 0.418,
      "step": 115500
    },
    {
      "epoch": 13.79,
      "learning_rate": 1.6094596379970936e-05,
      "loss": 0.4205,
      "step": 116000
    },
    {
      "epoch": 13.85,
      "learning_rate": 1.606817281014665e-05,
      "loss": 0.422,
      "step": 116500
    },
    {
      "epoch": 13.91,
      "learning_rate": 1.604174924032237e-05,
      "loss": 0.4257,
      "step": 117000
    },
    {
      "epoch": 13.97,
      "learning_rate": 1.6015325670498085e-05,
      "loss": 0.4146,
      "step": 117500
    },
    {
      "epoch": 14.03,
      "learning_rate": 1.5988902100673804e-05,
      "loss": 0.4144,
      "step": 118000
    },
    {
      "epoch": 14.09,
      "learning_rate": 1.596247853084952e-05,
      "loss": 0.4009,
      "step": 118500
    },
    {
      "epoch": 14.15,
      "learning_rate": 1.5936054961025234e-05,
      "loss": 0.4037,
      "step": 119000
    },
    {
      "epoch": 14.21,
      "learning_rate": 1.59096842383406e-05,
      "loss": 0.4011,
      "step": 119500
    },
    {
      "epoch": 14.27,
      "learning_rate": 1.5883260668516317e-05,
      "loss": 0.4045,
      "step": 120000
    },
    {
      "epoch": 14.33,
      "learning_rate": 1.5856837098692035e-05,
      "loss": 0.4127,
      "step": 120500
    },
    {
      "epoch": 14.39,
      "learning_rate": 1.583041352886775e-05,
      "loss": 0.4051,
      "step": 121000
    },
    {
      "epoch": 14.45,
      "learning_rate": 1.580398995904347e-05,
      "loss": 0.4064,
      "step": 121500
    },
    {
      "epoch": 14.51,
      "learning_rate": 1.5777566389219184e-05,
      "loss": 0.4046,
      "step": 122000
    },
    {
      "epoch": 14.57,
      "learning_rate": 1.5751142819394903e-05,
      "loss": 0.4041,
      "step": 122500
    },
    {
      "epoch": 14.63,
      "learning_rate": 1.572471924957062e-05,
      "loss": 0.4116,
      "step": 123000
    },
    {
      "epoch": 14.68,
      "learning_rate": 1.5698295679746337e-05,
      "loss": 0.4061,
      "step": 123500
    },
    {
      "epoch": 14.74,
      "learning_rate": 1.5671872109922052e-05,
      "loss": 0.4119,
      "step": 124000
    },
    {
      "epoch": 14.8,
      "learning_rate": 1.564544854009777e-05,
      "loss": 0.4101,
      "step": 124500
    },
    {
      "epoch": 14.86,
      "learning_rate": 1.5619024970273483e-05,
      "loss": 0.412,
      "step": 125000
    },
    {
      "epoch": 14.92,
      "learning_rate": 1.55926014004492e-05,
      "loss": 0.4089,
      "step": 125500
    },
    {
      "epoch": 14.98,
      "learning_rate": 1.5566230677764565e-05,
      "loss": 0.4119,
      "step": 126000
    },
    {
      "epoch": 15.04,
      "learning_rate": 1.5539807107940284e-05,
      "loss": 0.3987,
      "step": 126500
    },
    {
      "epoch": 15.1,
      "learning_rate": 1.5513383538116e-05,
      "loss": 0.3884,
      "step": 127000
    },
    {
      "epoch": 15.16,
      "learning_rate": 1.5486959968291718e-05,
      "loss": 0.3889,
      "step": 127500
    },
    {
      "epoch": 15.22,
      "learning_rate": 1.5460589245607082e-05,
      "loss": 0.3947,
      "step": 128000
    },
    {
      "epoch": 15.28,
      "learning_rate": 1.543421852292245e-05,
      "loss": 0.3915,
      "step": 128500
    },
    {
      "epoch": 15.34,
      "learning_rate": 1.5407794953098164e-05,
      "loss": 0.3933,
      "step": 129000
    },
    {
      "epoch": 15.4,
      "learning_rate": 1.5381371383273883e-05,
      "loss": 0.3884,
      "step": 129500
    },
    {
      "epoch": 15.46,
      "learning_rate": 1.5354947813449598e-05,
      "loss": 0.3966,
      "step": 130000
    },
    {
      "epoch": 15.52,
      "learning_rate": 1.5328524243625317e-05,
      "loss": 0.4029,
      "step": 130500
    },
    {
      "epoch": 15.58,
      "learning_rate": 1.5302100673801032e-05,
      "loss": 0.394,
      "step": 131000
    },
    {
      "epoch": 15.64,
      "learning_rate": 1.527567710397675e-05,
      "loss": 0.3999,
      "step": 131500
    },
    {
      "epoch": 15.7,
      "learning_rate": 1.5249253534152466e-05,
      "loss": 0.4,
      "step": 132000
    },
    {
      "epoch": 15.75,
      "learning_rate": 1.522288281146783e-05,
      "loss": 0.4006,
      "step": 132500
    },
    {
      "epoch": 15.81,
      "learning_rate": 1.5196459241643547e-05,
      "loss": 0.4007,
      "step": 133000
    },
    {
      "epoch": 15.87,
      "learning_rate": 1.5170035671819264e-05,
      "loss": 0.4058,
      "step": 133500
    },
    {
      "epoch": 15.93,
      "learning_rate": 1.514361210199498e-05,
      "loss": 0.4032,
      "step": 134000
    },
    {
      "epoch": 15.99,
      "learning_rate": 1.5117188532170698e-05,
      "loss": 0.4027,
      "step": 134500
    },
    {
      "epoch": 16.05,
      "learning_rate": 1.5090817809486063e-05,
      "loss": 0.3827,
      "step": 135000
    },
    {
      "epoch": 16.11,
      "learning_rate": 1.5064394239661778e-05,
      "loss": 0.3889,
      "step": 135500
    },
    {
      "epoch": 16.17,
      "learning_rate": 1.5037970669837495e-05,
      "loss": 0.3876,
      "step": 136000
    },
    {
      "epoch": 16.23,
      "learning_rate": 1.5011547100013212e-05,
      "loss": 0.3825,
      "step": 136500
    },
    {
      "epoch": 16.29,
      "learning_rate": 1.498512353018893e-05,
      "loss": 0.3872,
      "step": 137000
    },
    {
      "epoch": 16.35,
      "learning_rate": 1.4958699960364646e-05,
      "loss": 0.3895,
      "step": 137500
    },
    {
      "epoch": 16.41,
      "learning_rate": 1.4932276390540363e-05,
      "loss": 0.3884,
      "step": 138000
    },
    {
      "epoch": 16.47,
      "learning_rate": 1.490585282071608e-05,
      "loss": 0.3863,
      "step": 138500
    },
    {
      "epoch": 16.53,
      "learning_rate": 1.4879482098031446e-05,
      "loss": 0.3914,
      "step": 139000
    },
    {
      "epoch": 16.59,
      "learning_rate": 1.4853111375346811e-05,
      "loss": 0.3982,
      "step": 139500
    },
    {
      "epoch": 16.65,
      "learning_rate": 1.4826687805522528e-05,
      "loss": 0.3949,
      "step": 140000
    },
    {
      "epoch": 16.71,
      "learning_rate": 1.4800264235698245e-05,
      "loss": 0.3924,
      "step": 140500
    },
    {
      "epoch": 16.77,
      "learning_rate": 1.4773893513013609e-05,
      "loss": 0.3933,
      "step": 141000
    },
    {
      "epoch": 16.83,
      "learning_rate": 1.4747469943189326e-05,
      "loss": 0.394,
      "step": 141500
    },
    {
      "epoch": 16.88,
      "learning_rate": 1.4721046373365043e-05,
      "loss": 0.3907,
      "step": 142000
    },
    {
      "epoch": 16.94,
      "learning_rate": 1.4694622803540758e-05,
      "loss": 0.3955,
      "step": 142500
    },
    {
      "epoch": 17.0,
      "learning_rate": 1.4668199233716475e-05,
      "loss": 0.3903,
      "step": 143000
    },
    {
      "epoch": 17.06,
      "learning_rate": 1.4641775663892192e-05,
      "loss": 0.3735,
      "step": 143500
    },
    {
      "epoch": 17.12,
      "learning_rate": 1.4615352094067909e-05,
      "loss": 0.379,
      "step": 144000
    },
    {
      "epoch": 17.18,
      "learning_rate": 1.4589034218522923e-05,
      "loss": 0.3799,
      "step": 144500
    },
    {
      "epoch": 17.24,
      "learning_rate": 1.456261064869864e-05,
      "loss": 0.3772,
      "step": 145000
    },
    {
      "epoch": 17.3,
      "learning_rate": 1.4536187078874357e-05,
      "loss": 0.3806,
      "step": 145500
    },
    {
      "epoch": 17.36,
      "learning_rate": 1.4509763509050074e-05,
      "loss": 0.3775,
      "step": 146000
    },
    {
      "epoch": 17.42,
      "learning_rate": 1.4483339939225791e-05,
      "loss": 0.3799,
      "step": 146500
    },
    {
      "epoch": 17.48,
      "learning_rate": 1.4456916369401508e-05,
      "loss": 0.3835,
      "step": 147000
    },
    {
      "epoch": 17.54,
      "learning_rate": 1.4430492799577225e-05,
      "loss": 0.3806,
      "step": 147500
    },
    {
      "epoch": 17.6,
      "learning_rate": 1.4404069229752942e-05,
      "loss": 0.3825,
      "step": 148000
    },
    {
      "epoch": 17.66,
      "learning_rate": 1.4377645659928659e-05,
      "loss": 0.3873,
      "step": 148500
    },
    {
      "epoch": 17.72,
      "learning_rate": 1.4351222090104376e-05,
      "loss": 0.3889,
      "step": 149000
    },
    {
      "epoch": 17.78,
      "learning_rate": 1.4324798520280093e-05,
      "loss": 0.3855,
      "step": 149500
    },
    {
      "epoch": 17.84,
      "learning_rate": 1.4298374950455806e-05,
      "loss": 0.3866,
      "step": 150000
    },
    {
      "epoch": 17.9,
      "learning_rate": 1.4271951380631523e-05,
      "loss": 0.385,
      "step": 150500
    },
    {
      "epoch": 17.95,
      "learning_rate": 1.424552781080724e-05,
      "loss": 0.391,
      "step": 151000
    },
    {
      "epoch": 18.01,
      "learning_rate": 1.4219104240982957e-05,
      "loss": 0.379,
      "step": 151500
    },
    {
      "epoch": 18.07,
      "learning_rate": 1.4192680671158674e-05,
      "loss": 0.369,
      "step": 152000
    },
    {
      "epoch": 18.13,
      "learning_rate": 1.416630994847404e-05,
      "loss": 0.3725,
      "step": 152500
    },
    {
      "epoch": 18.19,
      "learning_rate": 1.4139886378649757e-05,
      "loss": 0.3731,
      "step": 153000
    },
    {
      "epoch": 18.25,
      "learning_rate": 1.4113462808825474e-05,
      "loss": 0.3704,
      "step": 153500
    },
    {
      "epoch": 18.31,
      "learning_rate": 1.408703923900119e-05,
      "loss": 0.3696,
      "step": 154000
    },
    {
      "epoch": 18.37,
      "learning_rate": 1.4060668516316556e-05,
      "loss": 0.3682,
      "step": 154500
    },
    {
      "epoch": 18.43,
      "learning_rate": 1.4034244946492273e-05,
      "loss": 0.3772,
      "step": 155000
    },
    {
      "epoch": 18.49,
      "learning_rate": 1.400782137666799e-05,
      "loss": 0.3794,
      "step": 155500
    },
    {
      "epoch": 18.55,
      "learning_rate": 1.3981397806843705e-05,
      "loss": 0.3799,
      "step": 156000
    },
    {
      "epoch": 18.61,
      "learning_rate": 1.3954974237019422e-05,
      "loss": 0.3695,
      "step": 156500
    },
    {
      "epoch": 18.67,
      "learning_rate": 1.392855066719514e-05,
      "loss": 0.3768,
      "step": 157000
    },
    {
      "epoch": 18.73,
      "learning_rate": 1.3902179944510505e-05,
      "loss": 0.381,
      "step": 157500
    },
    {
      "epoch": 18.79,
      "learning_rate": 1.3875756374686222e-05,
      "loss": 0.375,
      "step": 158000
    },
    {
      "epoch": 18.85,
      "learning_rate": 1.3849332804861937e-05,
      "loss": 0.377,
      "step": 158500
    },
    {
      "epoch": 18.91,
      "learning_rate": 1.3822909235037654e-05,
      "loss": 0.3853,
      "step": 159000
    },
    {
      "epoch": 18.97,
      "learning_rate": 1.3796485665213371e-05,
      "loss": 0.3778,
      "step": 159500
    },
    {
      "epoch": 19.02,
      "learning_rate": 1.3770062095389088e-05,
      "loss": 0.3688,
      "step": 160000
    },
    {
      "epoch": 19.08,
      "learning_rate": 1.3743691372704453e-05,
      "loss": 0.3653,
      "step": 160500
    },
    {
      "epoch": 19.14,
      "learning_rate": 1.371726780288017e-05,
      "loss": 0.355,
      "step": 161000
    },
    {
      "epoch": 19.2,
      "learning_rate": 1.3690844233055887e-05,
      "loss": 0.3644,
      "step": 161500
    },
    {
      "epoch": 19.26,
      "learning_rate": 1.3664420663231603e-05,
      "loss": 0.3673,
      "step": 162000
    },
    {
      "epoch": 19.32,
      "learning_rate": 1.363799709340732e-05,
      "loss": 0.3691,
      "step": 162500
    },
    {
      "epoch": 19.38,
      "learning_rate": 1.3611626370722685e-05,
      "loss": 0.3544,
      "step": 163000
    },
    {
      "epoch": 19.44,
      "learning_rate": 1.3585202800898402e-05,
      "loss": 0.368,
      "step": 163500
    },
    {
      "epoch": 19.5,
      "learning_rate": 1.3558779231074119e-05,
      "loss": 0.3691,
      "step": 164000
    },
    {
      "epoch": 19.56,
      "learning_rate": 1.3532355661249836e-05,
      "loss": 0.3655,
      "step": 164500
    },
    {
      "epoch": 19.62,
      "learning_rate": 1.3505932091425553e-05,
      "loss": 0.3692,
      "step": 165000
    },
    {
      "epoch": 19.68,
      "learning_rate": 1.3479561368740918e-05,
      "loss": 0.3782,
      "step": 165500
    },
    {
      "epoch": 19.74,
      "learning_rate": 1.3453137798916635e-05,
      "loss": 0.3708,
      "step": 166000
    },
    {
      "epoch": 19.8,
      "learning_rate": 1.3426714229092352e-05,
      "loss": 0.3673,
      "step": 166500
    },
    {
      "epoch": 19.86,
      "learning_rate": 1.3400343506407716e-05,
      "loss": 0.3678,
      "step": 167000
    },
    {
      "epoch": 19.92,
      "learning_rate": 1.3373919936583433e-05,
      "loss": 0.3721,
      "step": 167500
    },
    {
      "epoch": 19.98,
      "learning_rate": 1.334749636675915e-05,
      "loss": 0.3741,
      "step": 168000
    }
  ],
  "max_steps": 420500,
  "num_train_epochs": 50,
  "total_flos": 7.76486352033908e+19,
  "trial_name": null,
  "trial_params": null
}