{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 5000,
  "global_step": 36816,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 1.9972837896566713e-05,
      "loss": 1.0389,
      "step": 50
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.994567579313342e-05,
      "loss": 0.9783,
      "step": 100
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9918513689700133e-05,
      "loss": 0.8949,
      "step": 150
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9891351586266844e-05,
      "loss": 0.8832,
      "step": 200
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9864189482833552e-05,
      "loss": 0.8377,
      "step": 250
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.983702737940026e-05,
      "loss": 0.8297,
      "step": 300
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9809865275966972e-05,
      "loss": 0.7661,
      "step": 350
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.978270317253368e-05,
      "loss": 0.803,
      "step": 400
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9755541069100392e-05,
      "loss": 0.7971,
      "step": 450
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9728378965667103e-05,
      "loss": 0.7882,
      "step": 500
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9701216862233815e-05,
      "loss": 0.7406,
      "step": 550
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9674054758800523e-05,
      "loss": 0.7539,
      "step": 600
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9646892655367235e-05,
      "loss": 0.7562,
      "step": 650
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9619730551933943e-05,
      "loss": 0.7525,
      "step": 700
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9592568448500654e-05,
      "loss": 0.7378,
      "step": 750
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9565406345067362e-05,
      "loss": 0.738,
      "step": 800
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9538244241634074e-05,
      "loss": 0.727,
      "step": 850
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9511082138200782e-05,
      "loss": 0.7254,
      "step": 900
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9483920034767494e-05,
      "loss": 0.7433,
      "step": 950
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9456757931334205e-05,
      "loss": 0.7547,
      "step": 1000
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9429595827900913e-05,
      "loss": 0.7251,
      "step": 1050
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9402433724467625e-05,
      "loss": 0.7203,
      "step": 1100
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9375271621034336e-05,
      "loss": 0.704,
      "step": 1150
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9348109517601044e-05,
      "loss": 0.7271,
      "step": 1200
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9320947414167756e-05,
      "loss": 0.6955,
      "step": 1250
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9293785310734464e-05,
      "loss": 0.6565,
      "step": 1300
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9266623207301176e-05,
      "loss": 0.7266,
      "step": 1350
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9239461103867884e-05,
      "loss": 0.6809,
      "step": 1400
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9212299000434595e-05,
      "loss": 0.7345,
      "step": 1450
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9185136897001307e-05,
      "loss": 0.7052,
      "step": 1500
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9157974793568015e-05,
      "loss": 0.7097,
      "step": 1550
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9130812690134726e-05,
      "loss": 0.7224,
      "step": 1600
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9103650586701438e-05,
      "loss": 0.6998,
      "step": 1650
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9076488483268146e-05,
      "loss": 0.6799,
      "step": 1700
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9049326379834858e-05,
      "loss": 0.6957,
      "step": 1750
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9022164276401566e-05,
      "loss": 0.7111,
      "step": 1800
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.8995002172968274e-05,
      "loss": 0.7109,
      "step": 1850
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.8967840069534985e-05,
      "loss": 0.7088,
      "step": 1900
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.8940677966101697e-05,
      "loss": 0.6694,
      "step": 1950
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.8913515862668405e-05,
      "loss": 0.6598,
      "step": 2000
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8886353759235117e-05,
      "loss": 0.6834,
      "step": 2050
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8859191655801828e-05,
      "loss": 0.7251,
      "step": 2100
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8832029552368536e-05,
      "loss": 0.6978,
      "step": 2150
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8804867448935248e-05,
      "loss": 0.6761,
      "step": 2200
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8777705345501956e-05,
      "loss": 0.6638,
      "step": 2250
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8750543242068668e-05,
      "loss": 0.6987,
      "step": 2300
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8723381138635376e-05,
      "loss": 0.6922,
      "step": 2350
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8696219035202087e-05,
      "loss": 0.6764,
      "step": 2400
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.86690569317688e-05,
      "loss": 0.6812,
      "step": 2450
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8641894828335507e-05,
      "loss": 0.6764,
      "step": 2500
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.861473272490222e-05,
      "loss": 0.6856,
      "step": 2550
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.858757062146893e-05,
      "loss": 0.7037,
      "step": 2600
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.8560408518035638e-05,
      "loss": 0.6843,
      "step": 2650
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.853324641460235e-05,
      "loss": 0.6761,
      "step": 2700
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.8506084311169058e-05,
      "loss": 0.6778,
      "step": 2750
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.847892220773577e-05,
      "loss": 0.6733,
      "step": 2800
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.8451760104302477e-05,
      "loss": 0.6711,
      "step": 2850
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.842459800086919e-05,
      "loss": 0.684,
      "step": 2900
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.8397435897435897e-05,
      "loss": 0.6783,
      "step": 2950
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.837027379400261e-05,
      "loss": 0.6788,
      "step": 3000
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.834311169056932e-05,
      "loss": 0.6497,
      "step": 3050
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.8315949587136028e-05,
      "loss": 0.6246,
      "step": 3100
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.828878748370274e-05,
      "loss": 0.6497,
      "step": 3150
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.826162538026945e-05,
      "loss": 0.663,
      "step": 3200
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.823446327683616e-05,
      "loss": 0.6542,
      "step": 3250
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.820730117340287e-05,
      "loss": 0.6734,
      "step": 3300
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.818013906996958e-05,
      "loss": 0.6548,
      "step": 3350
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.815297696653629e-05,
      "loss": 0.6362,
      "step": 3400
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.8125814863103e-05,
      "loss": 0.6544,
      "step": 3450
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.809865275966971e-05,
      "loss": 0.658,
      "step": 3500
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.8071490656236422e-05,
      "loss": 0.6443,
      "step": 3550
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.804432855280313e-05,
      "loss": 0.6707,
      "step": 3600
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.801716644936984e-05,
      "loss": 0.6438,
      "step": 3650
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.7990004345936553e-05,
      "loss": 0.6871,
      "step": 3700
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.796284224250326e-05,
      "loss": 0.6694,
      "step": 3750
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.793568013906997e-05,
      "loss": 0.6279,
      "step": 3800
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.790851803563668e-05,
      "loss": 0.634,
      "step": 3850
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.788135593220339e-05,
      "loss": 0.6535,
      "step": 3900
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.78541938287701e-05,
      "loss": 0.6388,
      "step": 3950
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.7827031725336812e-05,
      "loss": 0.6227,
      "step": 4000
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.7799869621903524e-05,
      "loss": 0.6741,
      "step": 4050
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.777270751847023e-05,
      "loss": 0.6566,
      "step": 4100
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.7745545415036943e-05,
      "loss": 0.6695,
      "step": 4150
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.771838331160365e-05,
      "loss": 0.6454,
      "step": 4200
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.7691221208170363e-05,
      "loss": 0.6531,
      "step": 4250
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.766405910473707e-05,
      "loss": 0.6287,
      "step": 4300
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.7636897001303783e-05,
      "loss": 0.6356,
      "step": 4350
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.760973489787049e-05,
      "loss": 0.6609,
      "step": 4400
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.7582572794437202e-05,
      "loss": 0.6502,
      "step": 4450
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.7555410691003914e-05,
      "loss": 0.6211,
      "step": 4500
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.7528248587570622e-05,
      "loss": 0.6432,
      "step": 4550
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.7501086484137333e-05,
      "loss": 0.6416,
      "step": 4600
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.7473924380704045e-05,
      "loss": 0.6211,
      "step": 4650
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.7446762277270753e-05,
      "loss": 0.6537,
      "step": 4700
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.7419600173837465e-05,
      "loss": 0.6001,
      "step": 4750
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.7392438070404173e-05,
      "loss": 0.6057,
      "step": 4800
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.7365275966970884e-05,
      "loss": 0.6716,
      "step": 4850
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.7338113863537592e-05,
      "loss": 0.6447,
      "step": 4900
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.7310951760104304e-05,
      "loss": 0.6371,
      "step": 4950
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.7283789656671015e-05,
      "loss": 0.6369,
      "step": 5000
    },
    {
      "epoch": 0.41,
      "eval_accuracy": 0.7400916963830871,
      "eval_loss": 0.6398988962173462,
      "eval_runtime": 16.7037,
      "eval_samples_per_second": 587.595,
      "eval_steps_per_second": 36.758,
      "step": 5000
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.7256627553237724e-05,
      "loss": 0.6108,
      "step": 5050
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.7229465449804435e-05,
      "loss": 0.6118,
      "step": 5100
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.7202303346371147e-05,
      "loss": 0.6306,
      "step": 5150
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.7175141242937855e-05,
      "loss": 0.6089,
      "step": 5200
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.7147979139504566e-05,
      "loss": 0.6283,
      "step": 5250
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.7120817036071274e-05,
      "loss": 0.6383,
      "step": 5300
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.7093654932637983e-05,
      "loss": 0.6179,
      "step": 5350
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.7066492829204694e-05,
      "loss": 0.6357,
      "step": 5400
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.7039330725771406e-05,
      "loss": 0.634,
      "step": 5450
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.7012168622338114e-05,
      "loss": 0.5852,
      "step": 5500
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.6985006518904825e-05,
      "loss": 0.6249,
      "step": 5550
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.6957844415471537e-05,
      "loss": 0.6043,
      "step": 5600
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.6930682312038245e-05,
      "loss": 0.6171,
      "step": 5650
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.6903520208604957e-05,
      "loss": 0.6358,
      "step": 5700
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.6876358105171665e-05,
      "loss": 0.6129,
      "step": 5750
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.6849196001738376e-05,
      "loss": 0.6069,
      "step": 5800
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.6822033898305084e-05,
      "loss": 0.6553,
      "step": 5850
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.6794871794871796e-05,
      "loss": 0.6372,
      "step": 5900
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.6767709691438507e-05,
      "loss": 0.6348,
      "step": 5950
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.6740547588005215e-05,
      "loss": 0.596,
      "step": 6000
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.6713385484571927e-05,
      "loss": 0.6192,
      "step": 6050
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.668622338113864e-05,
      "loss": 0.6435,
      "step": 6100
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.6659061277705347e-05,
      "loss": 0.6135,
      "step": 6150
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.6631899174272058e-05,
      "loss": 0.6476,
      "step": 6200
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.6604737070838766e-05,
      "loss": 0.646,
      "step": 6250
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.6577574967405478e-05,
      "loss": 0.6104,
      "step": 6300
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.6550412863972186e-05,
      "loss": 0.6151,
      "step": 6350
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.6523250760538898e-05,
      "loss": 0.6314,
      "step": 6400
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.6496088657105606e-05,
      "loss": 0.6256,
      "step": 6450
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.6468926553672317e-05,
      "loss": 0.6284,
      "step": 6500
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.644176445023903e-05,
      "loss": 0.6134,
      "step": 6550
    },
    {
      "epoch": 0.54,
      "learning_rate": 1.6414602346805737e-05,
      "loss": 0.6203,
      "step": 6600
    },
    {
      "epoch": 0.54,
      "learning_rate": 1.638744024337245e-05,
      "loss": 0.5652,
      "step": 6650
    },
    {
      "epoch": 0.55,
      "learning_rate": 1.636027813993916e-05,
      "loss": 0.6353,
      "step": 6700
    },
    {
      "epoch": 0.55,
      "learning_rate": 1.6333116036505868e-05,
      "loss": 0.6013,
      "step": 6750
    },
    {
      "epoch": 0.55,
      "learning_rate": 1.630595393307258e-05,
      "loss": 0.6112,
      "step": 6800
    },
    {
      "epoch": 0.56,
      "learning_rate": 1.6278791829639288e-05,
      "loss": 0.6161,
      "step": 6850
    },
    {
      "epoch": 0.56,
      "learning_rate": 1.6251629726206e-05,
      "loss": 0.6173,
      "step": 6900
    },
    {
      "epoch": 0.57,
      "learning_rate": 1.6224467622772707e-05,
      "loss": 0.6203,
      "step": 6950
    },
    {
      "epoch": 0.57,
      "learning_rate": 1.619730551933942e-05,
      "loss": 0.6264,
      "step": 7000
    },
    {
      "epoch": 0.57,
      "learning_rate": 1.617014341590613e-05,
      "loss": 0.6104,
      "step": 7050
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.614298131247284e-05,
      "loss": 0.5963,
      "step": 7100
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.611581920903955e-05,
      "loss": 0.6044,
      "step": 7150
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.608865710560626e-05,
      "loss": 0.601,
      "step": 7200
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.606149500217297e-05,
      "loss": 0.5882,
      "step": 7250
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.6034332898739678e-05,
      "loss": 0.5976,
      "step": 7300
    },
    {
      "epoch": 0.6,
      "learning_rate": 1.600717079530639e-05,
      "loss": 0.6077,
      "step": 7350
    },
    {
      "epoch": 0.6,
      "learning_rate": 1.5980008691873098e-05,
      "loss": 0.6079,
      "step": 7400
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.595284658843981e-05,
      "loss": 0.6181,
      "step": 7450
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.592568448500652e-05,
      "loss": 0.6186,
      "step": 7500
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.589852238157323e-05,
      "loss": 0.6134,
      "step": 7550
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.587136027813994e-05,
      "loss": 0.6042,
      "step": 7600
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.5844198174706652e-05,
      "loss": 0.621,
      "step": 7650
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.5817036071273363e-05,
      "loss": 0.6249,
      "step": 7700
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.578987396784007e-05,
      "loss": 0.6148,
      "step": 7750
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.576271186440678e-05,
      "loss": 0.6052,
      "step": 7800
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.573554976097349e-05,
      "loss": 0.6207,
      "step": 7850
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.57083876575402e-05,
      "loss": 0.575,
      "step": 7900
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.568122555410691e-05,
      "loss": 0.6081,
      "step": 7950
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.5654063450673622e-05,
      "loss": 0.6148,
      "step": 8000
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.562690134724033e-05,
      "loss": 0.603,
      "step": 8050
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.5599739243807042e-05,
      "loss": 0.5894,
      "step": 8100
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.5572577140373754e-05,
      "loss": 0.6078,
      "step": 8150
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.5545415036940462e-05,
      "loss": 0.5784,
      "step": 8200
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.5518252933507173e-05,
      "loss": 0.5849,
      "step": 8250
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.549109083007388e-05,
      "loss": 0.6066,
      "step": 8300
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.5463928726640593e-05,
      "loss": 0.599,
      "step": 8350
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.54367666232073e-05,
      "loss": 0.6186,
      "step": 8400
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.5409604519774013e-05,
      "loss": 0.5947,
      "step": 8450
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.5382442416340724e-05,
      "loss": 0.6191,
      "step": 8500
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.5355280312907432e-05,
      "loss": 0.5896,
      "step": 8550
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.5328118209474144e-05,
      "loss": 0.6241,
      "step": 8600
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.5300956106040855e-05,
      "loss": 0.5856,
      "step": 8650
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.5273794002607563e-05,
      "loss": 0.5966,
      "step": 8700
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.5246631899174273e-05,
      "loss": 0.602,
      "step": 8750
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.5219469795740985e-05,
      "loss": 0.6188,
      "step": 8800
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.5192307692307693e-05,
      "loss": 0.6342,
      "step": 8850
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.5165145588874404e-05,
      "loss": 0.605,
      "step": 8900
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.5137983485441114e-05,
      "loss": 0.6138,
      "step": 8950
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.5110821382007822e-05,
      "loss": 0.6043,
      "step": 9000
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.5083659278574534e-05,
      "loss": 0.6095,
      "step": 9050
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.5056497175141245e-05,
      "loss": 0.5755,
      "step": 9100
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.5029335071707954e-05,
      "loss": 0.5987,
      "step": 9150
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.5002172968274663e-05,
      "loss": 0.5747,
      "step": 9200
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.4975010864841375e-05,
      "loss": 0.5762,
      "step": 9250
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.4947848761408083e-05,
      "loss": 0.5679,
      "step": 9300
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.4920686657974795e-05,
      "loss": 0.5724,
      "step": 9350
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.4893524554541504e-05,
      "loss": 0.5789,
      "step": 9400
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.4866362451108216e-05,
      "loss": 0.5888,
      "step": 9450
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.4839200347674924e-05,
      "loss": 0.5821,
      "step": 9500
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.4812038244241636e-05,
      "loss": 0.6436,
      "step": 9550
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.4784876140808346e-05,
      "loss": 0.5867,
      "step": 9600
    },
    {
      "epoch": 0.79,
      "learning_rate": 1.4757714037375055e-05,
      "loss": 0.5537,
      "step": 9650
    },
    {
      "epoch": 0.79,
      "learning_rate": 1.4730551933941765e-05,
      "loss": 0.5978,
      "step": 9700
    },
    {
      "epoch": 0.79,
      "learning_rate": 1.4703389830508477e-05,
      "loss": 0.5499,
      "step": 9750
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.4676227727075185e-05,
      "loss": 0.6219,
      "step": 9800
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.4649065623641896e-05,
      "loss": 0.5923,
      "step": 9850
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.4621903520208606e-05,
      "loss": 0.6236,
      "step": 9900
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.4594741416775316e-05,
      "loss": 0.5732,
      "step": 9950
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.4567579313342026e-05,
      "loss": 0.5945,
      "step": 10000
    },
    {
      "epoch": 0.81,
      "eval_accuracy": 0.7680081507896077,
      "eval_loss": 0.5745856761932373,
      "eval_runtime": 16.756,
      "eval_samples_per_second": 585.761,
      "eval_steps_per_second": 36.644,
      "step": 10000
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.4540417209908737e-05,
      "loss": 0.5909,
      "step": 10050
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.4513255106475446e-05,
      "loss": 0.5937,
      "step": 10100
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.4486093003042157e-05,
      "loss": 0.5916,
      "step": 10150
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.4458930899608867e-05,
      "loss": 0.5777,
      "step": 10200
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.4431768796175577e-05,
      "loss": 0.581,
      "step": 10250
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.4404606692742287e-05,
      "loss": 0.6225,
      "step": 10300
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.4377444589308998e-05,
      "loss": 0.6204,
      "step": 10350
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.4350282485875708e-05,
      "loss": 0.5783,
      "step": 10400
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.4323120382442418e-05,
      "loss": 0.5715,
      "step": 10450
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.4295958279009128e-05,
      "loss": 0.601,
      "step": 10500
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.4268796175575839e-05,
      "loss": 0.5964,
      "step": 10550
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.4241634072142547e-05,
      "loss": 0.6087,
      "step": 10600
    },
    {
      "epoch": 0.87,
      "learning_rate": 1.4214471968709259e-05,
      "loss": 0.5917,
      "step": 10650
    },
    {
      "epoch": 0.87,
      "learning_rate": 1.4187309865275969e-05,
      "loss": 0.5593,
      "step": 10700
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.4160147761842677e-05,
      "loss": 0.6105,
      "step": 10750
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.4132985658409388e-05,
      "loss": 0.573,
      "step": 10800
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.41058235549761e-05,
      "loss": 0.615,
      "step": 10850
    },
    {
      "epoch": 0.89,
      "learning_rate": 1.4078661451542808e-05,
      "loss": 0.5854,
      "step": 10900
    },
    {
      "epoch": 0.89,
      "learning_rate": 1.4051499348109518e-05,
      "loss": 0.5828,
      "step": 10950
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.402433724467623e-05,
      "loss": 0.5892,
      "step": 11000
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.3997175141242937e-05,
      "loss": 0.59,
      "step": 11050
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.3970013037809649e-05,
      "loss": 0.5789,
      "step": 11100
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.3942850934376359e-05,
      "loss": 0.5549,
      "step": 11150
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.391568883094307e-05,
      "loss": 0.6024,
      "step": 11200
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.3888526727509778e-05,
      "loss": 0.5597,
      "step": 11250
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.386136462407649e-05,
      "loss": 0.5646,
      "step": 11300
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.38342025206432e-05,
      "loss": 0.5711,
      "step": 11350
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.380704041720991e-05,
      "loss": 0.5555,
      "step": 11400
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.377987831377662e-05,
      "loss": 0.5504,
      "step": 11450
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.3752716210343331e-05,
      "loss": 0.5712,
      "step": 11500
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.372555410691004e-05,
      "loss": 0.5768,
      "step": 11550
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.369839200347675e-05,
      "loss": 0.5877,
      "step": 11600
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.367122990004346e-05,
      "loss": 0.5639,
      "step": 11650
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.364406779661017e-05,
      "loss": 0.5803,
      "step": 11700
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.361690569317688e-05,
      "loss": 0.602,
      "step": 11750
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.3589743589743592e-05,
      "loss": 0.5758,
      "step": 11800
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.35625814863103e-05,
      "loss": 0.5657,
      "step": 11850
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.3535419382877011e-05,
      "loss": 0.5857,
      "step": 11900
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.3508257279443721e-05,
      "loss": 0.5922,
      "step": 11950
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.3481095176010431e-05,
      "loss": 0.6179,
      "step": 12000
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.3453933072577141e-05,
      "loss": 0.5464,
      "step": 12050
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.3426770969143852e-05,
      "loss": 0.5578,
      "step": 12100
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.3399608865710562e-05,
      "loss": 0.6018,
      "step": 12150
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.3372446762277272e-05,
      "loss": 0.5907,
      "step": 12200
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.3345284658843982e-05,
      "loss": 0.5403,
      "step": 12250
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.3318122555410693e-05,
      "loss": 0.5376,
      "step": 12300
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.3290960451977402e-05,
      "loss": 0.5479,
      "step": 12350
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.3263798348544113e-05,
      "loss": 0.5121,
      "step": 12400
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.3236636245110823e-05,
      "loss": 0.4965,
      "step": 12450
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.3209474141677531e-05,
      "loss": 0.481,
      "step": 12500
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.3182312038244243e-05,
      "loss": 0.4933,
      "step": 12550
    },
    {
      "epoch": 1.03,
      "learning_rate": 1.3155149934810954e-05,
      "loss": 0.4907,
      "step": 12600
    },
    {
      "epoch": 1.03,
      "learning_rate": 1.3127987831377662e-05,
      "loss": 0.5178,
      "step": 12650
    },
    {
      "epoch": 1.03,
      "learning_rate": 1.3100825727944372e-05,
      "loss": 0.5114,
      "step": 12700
    },
    {
      "epoch": 1.04,
      "learning_rate": 1.3073663624511084e-05,
      "loss": 0.4874,
      "step": 12750
    },
    {
      "epoch": 1.04,
      "learning_rate": 1.3046501521077792e-05,
      "loss": 0.5061,
      "step": 12800
    },
    {
      "epoch": 1.05,
      "learning_rate": 1.3019339417644503e-05,
      "loss": 0.5151,
      "step": 12850
    },
    {
      "epoch": 1.05,
      "learning_rate": 1.2992177314211213e-05,
      "loss": 0.5208,
      "step": 12900
    },
    {
      "epoch": 1.06,
      "learning_rate": 1.2965015210777925e-05,
      "loss": 0.4972,
      "step": 12950
    },
    {
      "epoch": 1.06,
      "learning_rate": 1.2937853107344633e-05,
      "loss": 0.4916,
      "step": 13000
    },
    {
      "epoch": 1.06,
      "learning_rate": 1.2910691003911344e-05,
      "loss": 0.5134,
      "step": 13050
    },
    {
      "epoch": 1.07,
      "learning_rate": 1.2883528900478056e-05,
      "loss": 0.5218,
      "step": 13100
    },
    {
      "epoch": 1.07,
      "learning_rate": 1.2856366797044764e-05,
      "loss": 0.5096,
      "step": 13150
    },
    {
      "epoch": 1.08,
      "learning_rate": 1.2829204693611474e-05,
      "loss": 0.5038,
      "step": 13200
    },
    {
      "epoch": 1.08,
      "learning_rate": 1.2802042590178185e-05,
      "loss": 0.4812,
      "step": 13250
    },
    {
      "epoch": 1.08,
      "learning_rate": 1.2774880486744893e-05,
      "loss": 0.5195,
      "step": 13300
    },
    {
      "epoch": 1.09,
      "learning_rate": 1.2747718383311605e-05,
      "loss": 0.5023,
      "step": 13350
    },
    {
      "epoch": 1.09,
      "learning_rate": 1.2720556279878315e-05,
      "loss": 0.5112,
      "step": 13400
    },
    {
      "epoch": 1.1,
      "learning_rate": 1.2693394176445025e-05,
      "loss": 0.5131,
      "step": 13450
    },
    {
      "epoch": 1.1,
      "learning_rate": 1.2666232073011735e-05,
      "loss": 0.5317,
      "step": 13500
    },
    {
      "epoch": 1.1,
      "learning_rate": 1.2639069969578446e-05,
      "loss": 0.4836,
      "step": 13550
    },
    {
      "epoch": 1.11,
      "learning_rate": 1.2611907866145154e-05,
      "loss": 0.5583,
      "step": 13600
    },
    {
      "epoch": 1.11,
      "learning_rate": 1.2584745762711866e-05,
      "loss": 0.5195,
      "step": 13650
    },
    {
      "epoch": 1.12,
      "learning_rate": 1.2557583659278576e-05,
      "loss": 0.4933,
      "step": 13700
    },
    {
      "epoch": 1.12,
      "learning_rate": 1.2530421555845285e-05,
      "loss": 0.5127,
      "step": 13750
    },
    {
      "epoch": 1.12,
      "learning_rate": 1.2503259452411995e-05,
      "loss": 0.4924,
      "step": 13800
    },
    {
      "epoch": 1.13,
      "learning_rate": 1.2476097348978707e-05,
      "loss": 0.4972,
      "step": 13850
    },
    {
      "epoch": 1.13,
      "learning_rate": 1.2448935245545417e-05,
      "loss": 0.5336,
      "step": 13900
    },
    {
      "epoch": 1.14,
      "learning_rate": 1.2421773142112126e-05,
      "loss": 0.4995,
      "step": 13950
    },
    {
      "epoch": 1.14,
      "learning_rate": 1.2394611038678836e-05,
      "loss": 0.4911,
      "step": 14000
    },
    {
      "epoch": 1.14,
      "learning_rate": 1.2367448935245548e-05,
      "loss": 0.5177,
      "step": 14050
    },
    {
      "epoch": 1.15,
      "learning_rate": 1.2340286831812256e-05,
      "loss": 0.4603,
      "step": 14100
    },
    {
      "epoch": 1.15,
      "learning_rate": 1.2313124728378967e-05,
      "loss": 0.5058,
      "step": 14150
    },
    {
      "epoch": 1.16,
      "learning_rate": 1.2285962624945677e-05,
      "loss": 0.5239,
      "step": 14200
    },
    {
      "epoch": 1.16,
      "learning_rate": 1.2258800521512385e-05,
      "loss": 0.5143,
      "step": 14250
    },
    {
      "epoch": 1.17,
      "learning_rate": 1.2231638418079097e-05,
      "loss": 0.4909,
      "step": 14300
    },
    {
      "epoch": 1.17,
      "learning_rate": 1.2204476314645808e-05,
      "loss": 0.5104,
      "step": 14350
    },
    {
      "epoch": 1.17,
      "learning_rate": 1.2177314211212517e-05,
      "loss": 0.5263,
      "step": 14400
    },
    {
      "epoch": 1.18,
      "learning_rate": 1.2150152107779226e-05,
      "loss": 0.4828,
      "step": 14450
    },
    {
      "epoch": 1.18,
      "learning_rate": 1.2122990004345938e-05,
      "loss": 0.4891,
      "step": 14500
    },
    {
      "epoch": 1.19,
      "learning_rate": 1.2095827900912646e-05,
      "loss": 0.5003,
      "step": 14550
    },
    {
      "epoch": 1.19,
      "learning_rate": 1.2068665797479358e-05,
      "loss": 0.4827,
      "step": 14600
    },
    {
      "epoch": 1.19,
      "learning_rate": 1.204150369404607e-05,
      "loss": 0.5017,
      "step": 14650
    },
    {
      "epoch": 1.2,
      "learning_rate": 1.2014341590612777e-05,
      "loss": 0.525,
      "step": 14700
    },
    {
      "epoch": 1.2,
      "learning_rate": 1.1987179487179487e-05,
      "loss": 0.5077,
      "step": 14750
    },
    {
      "epoch": 1.21,
      "learning_rate": 1.1960017383746199e-05,
      "loss": 0.5062,
      "step": 14800
    },
    {
      "epoch": 1.21,
      "learning_rate": 1.193285528031291e-05,
      "loss": 0.5098,
      "step": 14850
    },
    {
      "epoch": 1.21,
      "learning_rate": 1.1905693176879618e-05,
      "loss": 0.5366,
      "step": 14900
    },
    {
      "epoch": 1.22,
      "learning_rate": 1.1878531073446328e-05,
      "loss": 0.5097,
      "step": 14950
    },
    {
      "epoch": 1.22,
      "learning_rate": 1.185136897001304e-05,
      "loss": 0.4847,
      "step": 15000
    },
    {
      "epoch": 1.22,
      "eval_accuracy": 0.7772796739684157,
      "eval_loss": 0.5816997289657593,
      "eval_runtime": 16.7546,
      "eval_samples_per_second": 585.81,
      "eval_steps_per_second": 36.647,
      "step": 15000
    },
    {
      "epoch": 1.23,
      "learning_rate": 1.1824206866579748e-05,
      "loss": 0.5372,
      "step": 15050
    },
    {
      "epoch": 1.23,
      "learning_rate": 1.179704476314646e-05,
      "loss": 0.4814,
      "step": 15100
    },
    {
      "epoch": 1.23,
      "learning_rate": 1.176988265971317e-05,
      "loss": 0.4823,
      "step": 15150
    },
    {
      "epoch": 1.24,
      "learning_rate": 1.1742720556279879e-05,
      "loss": 0.5288,
      "step": 15200
    },
    {
      "epoch": 1.24,
      "learning_rate": 1.1715558452846589e-05,
      "loss": 0.5211,
      "step": 15250
    },
    {
      "epoch": 1.25,
      "learning_rate": 1.16883963494133e-05,
      "loss": 0.4856,
      "step": 15300
    },
    {
      "epoch": 1.25,
      "learning_rate": 1.1661234245980009e-05,
      "loss": 0.5162,
      "step": 15350
    },
    {
      "epoch": 1.25,
      "learning_rate": 1.163407214254672e-05,
      "loss": 0.5032,
      "step": 15400
    },
    {
      "epoch": 1.26,
      "learning_rate": 1.160691003911343e-05,
      "loss": 0.5181,
      "step": 15450
    },
    {
      "epoch": 1.26,
      "learning_rate": 1.157974793568014e-05,
      "loss": 0.5042,
      "step": 15500
    },
    {
      "epoch": 1.27,
      "learning_rate": 1.155258583224685e-05,
      "loss": 0.4901,
      "step": 15550
    },
    {
      "epoch": 1.27,
      "learning_rate": 1.1525423728813561e-05,
      "loss": 0.5109,
      "step": 15600
    },
    {
      "epoch": 1.28,
      "learning_rate": 1.1498261625380271e-05,
      "loss": 0.5227,
      "step": 15650
    },
    {
      "epoch": 1.28,
      "learning_rate": 1.147109952194698e-05,
      "loss": 0.4899,
      "step": 15700
    },
    {
      "epoch": 1.28,
      "learning_rate": 1.144393741851369e-05,
      "loss": 0.4738,
      "step": 15750
    },
    {
      "epoch": 1.29,
      "learning_rate": 1.1416775315080402e-05,
      "loss": 0.5072,
      "step": 15800
    },
    {
      "epoch": 1.29,
      "learning_rate": 1.138961321164711e-05,
      "loss": 0.507,
      "step": 15850
    },
    {
      "epoch": 1.3,
      "learning_rate": 1.1362451108213822e-05,
      "loss": 0.5224,
      "step": 15900
    },
    {
      "epoch": 1.3,
      "learning_rate": 1.1335289004780532e-05,
      "loss": 0.5032,
      "step": 15950
    },
    {
      "epoch": 1.3,
      "learning_rate": 1.130812690134724e-05,
      "loss": 0.4855,
      "step": 16000
    },
    {
      "epoch": 1.31,
      "learning_rate": 1.1280964797913951e-05,
      "loss": 0.4908,
      "step": 16050
    },
    {
      "epoch": 1.31,
      "learning_rate": 1.1253802694480663e-05,
      "loss": 0.5129,
      "step": 16100
    },
    {
      "epoch": 1.32,
      "learning_rate": 1.1226640591047371e-05,
      "loss": 0.5153,
      "step": 16150
    },
    {
      "epoch": 1.32,
      "learning_rate": 1.1199478487614082e-05,
      "loss": 0.5167,
      "step": 16200
    },
    {
      "epoch": 1.32,
      "learning_rate": 1.1172316384180792e-05,
      "loss": 0.4818,
      "step": 16250
    },
    {
      "epoch": 1.33,
      "learning_rate": 1.11451542807475e-05,
      "loss": 0.5057,
      "step": 16300
    },
    {
      "epoch": 1.33,
      "learning_rate": 1.1117992177314212e-05,
      "loss": 0.4963,
      "step": 16350
    },
    {
      "epoch": 1.34,
      "learning_rate": 1.1090830073880923e-05,
      "loss": 0.4832,
      "step": 16400
    },
    {
      "epoch": 1.34,
      "learning_rate": 1.1063667970447632e-05,
      "loss": 0.4967,
      "step": 16450
    },
    {
      "epoch": 1.34,
      "learning_rate": 1.1036505867014341e-05,
      "loss": 0.4924,
      "step": 16500
    },
    {
      "epoch": 1.35,
      "learning_rate": 1.1009343763581053e-05,
      "loss": 0.4857,
      "step": 16550
    },
    {
      "epoch": 1.35,
      "learning_rate": 1.0982181660147765e-05,
      "loss": 0.4981,
      "step": 16600
    },
    {
      "epoch": 1.36,
      "learning_rate": 1.0955019556714473e-05,
      "loss": 0.484,
      "step": 16650
    },
    {
      "epoch": 1.36,
      "learning_rate": 1.0927857453281182e-05,
      "loss": 0.5214,
      "step": 16700
    },
    {
      "epoch": 1.36,
      "learning_rate": 1.0900695349847894e-05,
      "loss": 0.5174,
      "step": 16750
    },
    {
      "epoch": 1.37,
      "learning_rate": 1.0873533246414602e-05,
      "loss": 0.5203,
      "step": 16800
    },
    {
      "epoch": 1.37,
      "learning_rate": 1.0846371142981314e-05,
      "loss": 0.4865,
      "step": 16850
    },
    {
      "epoch": 1.38,
      "learning_rate": 1.0819209039548024e-05,
      "loss": 0.501,
      "step": 16900
    },
    {
      "epoch": 1.38,
      "learning_rate": 1.0792046936114733e-05,
      "loss": 0.4807,
      "step": 16950
    },
    {
      "epoch": 1.39,
      "learning_rate": 1.0764884832681443e-05,
      "loss": 0.5103,
      "step": 17000
    },
    {
      "epoch": 1.39,
      "learning_rate": 1.0737722729248155e-05,
      "loss": 0.5467,
      "step": 17050
    },
    {
      "epoch": 1.39,
      "learning_rate": 1.0710560625814863e-05,
      "loss": 0.4653,
      "step": 17100
    },
    {
      "epoch": 1.4,
      "learning_rate": 1.0683398522381574e-05,
      "loss": 0.4821,
      "step": 17150
    },
    {
      "epoch": 1.4,
      "learning_rate": 1.0656236418948284e-05,
      "loss": 0.5136,
      "step": 17200
    },
    {
      "epoch": 1.41,
      "learning_rate": 1.0629074315514994e-05,
      "loss": 0.4993,
      "step": 17250
    },
    {
      "epoch": 1.41,
      "learning_rate": 1.0601912212081704e-05,
      "loss": 0.5035,
      "step": 17300
    },
    {
      "epoch": 1.41,
      "learning_rate": 1.0574750108648415e-05,
      "loss": 0.4993,
      "step": 17350
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.0547588005215125e-05,
      "loss": 0.5187,
      "step": 17400
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.0520425901781835e-05,
      "loss": 0.507,
      "step": 17450
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.0493263798348545e-05,
      "loss": 0.4857,
      "step": 17500
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.0466101694915256e-05,
      "loss": 0.5085,
      "step": 17550
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.0438939591481965e-05,
      "loss": 0.5106,
      "step": 17600
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.0411777488048676e-05,
      "loss": 0.5152,
      "step": 17650
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.0384615384615386e-05,
      "loss": 0.5066,
      "step": 17700
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.0357453281182096e-05,
      "loss": 0.5097,
      "step": 17750
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.0330291177748806e-05,
      "loss": 0.5447,
      "step": 17800
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.0303129074315517e-05,
      "loss": 0.5142,
      "step": 17850
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.0275966970882225e-05,
      "loss": 0.5053,
      "step": 17900
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.0248804867448937e-05,
      "loss": 0.511,
      "step": 17950
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.0221642764015647e-05,
      "loss": 0.5134,
      "step": 18000
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.0194480660582355e-05,
      "loss": 0.5304,
      "step": 18050
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.0167318557149066e-05,
      "loss": 0.4506,
      "step": 18100
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.0140156453715778e-05,
      "loss": 0.5318,
      "step": 18150
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.0112994350282486e-05,
      "loss": 0.4923,
      "step": 18200
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.0085832246849196e-05,
      "loss": 0.4903,
      "step": 18250
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.0058670143415907e-05,
      "loss": 0.4626,
      "step": 18300
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.0031508039982619e-05,
      "loss": 0.4965,
      "step": 18350
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.0004345936549327e-05,
      "loss": 0.4839,
      "step": 18400
    },
    {
      "epoch": 1.5,
      "learning_rate": 9.977183833116037e-06,
      "loss": 0.5238,
      "step": 18450
    },
    {
      "epoch": 1.51,
      "learning_rate": 9.950021729682747e-06,
      "loss": 0.5056,
      "step": 18500
    },
    {
      "epoch": 1.51,
      "learning_rate": 9.922859626249458e-06,
      "loss": 0.5162,
      "step": 18550
    },
    {
      "epoch": 1.52,
      "learning_rate": 9.895697522816168e-06,
      "loss": 0.4718,
      "step": 18600
    },
    {
      "epoch": 1.52,
      "learning_rate": 9.868535419382878e-06,
      "loss": 0.5132,
      "step": 18650
    },
    {
      "epoch": 1.52,
      "learning_rate": 9.841373315949588e-06,
      "loss": 0.5185,
      "step": 18700
    },
    {
      "epoch": 1.53,
      "learning_rate": 9.814211212516298e-06,
      "loss": 0.4927,
      "step": 18750
    },
    {
      "epoch": 1.53,
      "learning_rate": 9.787049109083007e-06,
      "loss": 0.4992,
      "step": 18800
    },
    {
      "epoch": 1.54,
      "learning_rate": 9.759887005649719e-06,
      "loss": 0.4462,
      "step": 18850
    },
    {
      "epoch": 1.54,
      "learning_rate": 9.732724902216429e-06,
      "loss": 0.4955,
      "step": 18900
    },
    {
      "epoch": 1.54,
      "learning_rate": 9.705562798783139e-06,
      "loss": 0.5153,
      "step": 18950
    },
    {
      "epoch": 1.55,
      "learning_rate": 9.678400695349848e-06,
      "loss": 0.4776,
      "step": 19000
    },
    {
      "epoch": 1.55,
      "learning_rate": 9.651238591916558e-06,
      "loss": 0.518,
      "step": 19050
    },
    {
      "epoch": 1.56,
      "learning_rate": 9.62407648848327e-06,
      "loss": 0.5094,
      "step": 19100
    },
    {
      "epoch": 1.56,
      "learning_rate": 9.59691438504998e-06,
      "loss": 0.5092,
      "step": 19150
    },
    {
      "epoch": 1.56,
      "learning_rate": 9.56975228161669e-06,
      "loss": 0.5016,
      "step": 19200
    },
    {
      "epoch": 1.57,
      "learning_rate": 9.5425901781834e-06,
      "loss": 0.5357,
      "step": 19250
    },
    {
      "epoch": 1.57,
      "learning_rate": 9.515428074750109e-06,
      "loss": 0.521,
      "step": 19300
    },
    {
      "epoch": 1.58,
      "learning_rate": 9.488265971316819e-06,
      "loss": 0.488,
      "step": 19350
    },
    {
      "epoch": 1.58,
      "learning_rate": 9.46110386788353e-06,
      "loss": 0.4856,
      "step": 19400
    },
    {
      "epoch": 1.58,
      "learning_rate": 9.43394176445024e-06,
      "loss": 0.506,
      "step": 19450
    },
    {
      "epoch": 1.59,
      "learning_rate": 9.40677966101695e-06,
      "loss": 0.5038,
      "step": 19500
    },
    {
      "epoch": 1.59,
      "learning_rate": 9.37961755758366e-06,
      "loss": 0.471,
      "step": 19550
    },
    {
      "epoch": 1.6,
      "learning_rate": 9.35245545415037e-06,
      "loss": 0.4886,
      "step": 19600
    },
    {
      "epoch": 1.6,
      "learning_rate": 9.325293350717081e-06,
      "loss": 0.5131,
      "step": 19650
    },
    {
      "epoch": 1.61,
      "learning_rate": 9.298131247283791e-06,
      "loss": 0.5082,
      "step": 19700
    },
    {
      "epoch": 1.61,
      "learning_rate": 9.2709691438505e-06,
      "loss": 0.5021,
      "step": 19750
    },
    {
      "epoch": 1.61,
      "learning_rate": 9.24380704041721e-06,
      "loss": 0.4695,
      "step": 19800
    },
    {
      "epoch": 1.62,
      "learning_rate": 9.21664493698392e-06,
      "loss": 0.4883,
      "step": 19850
    },
    {
      "epoch": 1.62,
      "learning_rate": 9.189482833550632e-06,
      "loss": 0.4833,
      "step": 19900
    },
    {
      "epoch": 1.63,
      "learning_rate": 9.162320730117342e-06,
      "loss": 0.4642,
      "step": 19950
    },
    {
      "epoch": 1.63,
      "learning_rate": 9.13515862668405e-06,
      "loss": 0.5109,
      "step": 20000
    },
    {
      "epoch": 1.63,
      "eval_accuracy": 0.7790117167600611,
      "eval_loss": 0.5679929852485657,
      "eval_runtime": 16.5969,
      "eval_samples_per_second": 591.374,
      "eval_steps_per_second": 36.995,
      "step": 20000
    },
    {
      "epoch": 1.63,
      "learning_rate": 9.107996523250762e-06,
      "loss": 0.5284,
      "step": 20050
    },
    {
      "epoch": 1.64,
      "learning_rate": 9.080834419817471e-06,
      "loss": 0.4903,
      "step": 20100
    },
    {
      "epoch": 1.64,
      "learning_rate": 9.053672316384181e-06,
      "loss": 0.5407,
      "step": 20150
    },
    {
      "epoch": 1.65,
      "learning_rate": 9.026510212950891e-06,
      "loss": 0.4949,
      "step": 20200
    },
    {
      "epoch": 1.65,
      "learning_rate": 8.999348109517601e-06,
      "loss": 0.5127,
      "step": 20250
    },
    {
      "epoch": 1.65,
      "learning_rate": 8.972186006084312e-06,
      "loss": 0.4523,
      "step": 20300
    },
    {
      "epoch": 1.66,
      "learning_rate": 8.945023902651022e-06,
      "loss": 0.4655,
      "step": 20350
    },
    {
      "epoch": 1.66,
      "learning_rate": 8.917861799217732e-06,
      "loss": 0.5071,
      "step": 20400
    },
    {
      "epoch": 1.67,
      "learning_rate": 8.890699695784442e-06,
      "loss": 0.4697,
      "step": 20450
    },
    {
      "epoch": 1.67,
      "learning_rate": 8.863537592351152e-06,
      "loss": 0.499,
      "step": 20500
    },
    {
      "epoch": 1.67,
      "learning_rate": 8.836375488917862e-06,
      "loss": 0.5156,
      "step": 20550
    },
    {
      "epoch": 1.68,
      "learning_rate": 8.809213385484573e-06,
      "loss": 0.5244,
      "step": 20600
    },
    {
      "epoch": 1.68,
      "learning_rate": 8.782051282051283e-06,
      "loss": 0.5259,
      "step": 20650
    },
    {
      "epoch": 1.69,
      "learning_rate": 8.754889178617993e-06,
      "loss": 0.4975,
      "step": 20700
    },
    {
      "epoch": 1.69,
      "learning_rate": 8.727727075184703e-06,
      "loss": 0.4861,
      "step": 20750
    },
    {
      "epoch": 1.69,
      "learning_rate": 8.700564971751413e-06,
      "loss": 0.5068,
      "step": 20800
    },
    {
      "epoch": 1.7,
      "learning_rate": 8.673402868318124e-06,
      "loss": 0.477,
      "step": 20850
    },
    {
      "epoch": 1.7,
      "learning_rate": 8.646240764884834e-06,
      "loss": 0.4909,
      "step": 20900
    },
    {
      "epoch": 1.71,
      "learning_rate": 8.619078661451544e-06,
      "loss": 0.5286,
      "step": 20950
    },
    {
      "epoch": 1.71,
      "learning_rate": 8.591916558018254e-06,
      "loss": 0.5003,
      "step": 21000
    },
    {
      "epoch": 1.72,
      "learning_rate": 8.564754454584963e-06,
      "loss": 0.4592,
      "step": 21050
    },
    {
      "epoch": 1.72,
      "learning_rate": 8.537592351151673e-06,
      "loss": 0.4945,
      "step": 21100
    },
    {
      "epoch": 1.72,
      "learning_rate": 8.510430247718385e-06,
      "loss": 0.5153,
      "step": 21150
    },
    {
      "epoch": 1.73,
      "learning_rate": 8.483268144285095e-06,
      "loss": 0.488,
      "step": 21200
    },
    {
      "epoch": 1.73,
      "learning_rate": 8.456106040851804e-06,
      "loss": 0.5066,
      "step": 21250
    },
    {
      "epoch": 1.74,
      "learning_rate": 8.428943937418514e-06,
      "loss": 0.506,
      "step": 21300
    },
    {
      "epoch": 1.74,
      "learning_rate": 8.401781833985224e-06,
      "loss": 0.5118,
      "step": 21350
    },
    {
      "epoch": 1.74,
      "learning_rate": 8.374619730551936e-06,
      "loss": 0.4863,
      "step": 21400
    },
    {
      "epoch": 1.75,
      "learning_rate": 8.347457627118645e-06,
      "loss": 0.4974,
      "step": 21450
    },
    {
      "epoch": 1.75,
      "learning_rate": 8.320295523685355e-06,
      "loss": 0.4872,
      "step": 21500
    },
    {
      "epoch": 1.76,
      "learning_rate": 8.293133420252065e-06,
      "loss": 0.4901,
      "step": 21550
    },
    {
      "epoch": 1.76,
      "learning_rate": 8.265971316818775e-06,
      "loss": 0.5113,
      "step": 21600
    },
    {
      "epoch": 1.76,
      "learning_rate": 8.238809213385486e-06,
      "loss": 0.4773,
      "step": 21650
    },
    {
      "epoch": 1.77,
      "learning_rate": 8.211647109952196e-06,
      "loss": 0.4777,
      "step": 21700
    },
    {
      "epoch": 1.77,
      "learning_rate": 8.184485006518904e-06,
      "loss": 0.5108,
      "step": 21750
    },
    {
      "epoch": 1.78,
      "learning_rate": 8.157322903085616e-06,
      "loss": 0.4692,
      "step": 21800
    },
    {
      "epoch": 1.78,
      "learning_rate": 8.130160799652326e-06,
      "loss": 0.5289,
      "step": 21850
    },
    {
      "epoch": 1.78,
      "learning_rate": 8.102998696219036e-06,
      "loss": 0.5053,
      "step": 21900
    },
    {
      "epoch": 1.79,
      "learning_rate": 8.075836592785745e-06,
      "loss": 0.4812,
      "step": 21950
    },
    {
      "epoch": 1.79,
      "learning_rate": 8.048674489352455e-06,
      "loss": 0.4772,
      "step": 22000
    },
    {
      "epoch": 1.8,
      "learning_rate": 8.021512385919165e-06,
      "loss": 0.5205,
      "step": 22050
    },
    {
      "epoch": 1.8,
      "learning_rate": 7.994350282485877e-06,
      "loss": 0.4581,
      "step": 22100
    },
    {
      "epoch": 1.8,
      "learning_rate": 7.967188179052586e-06,
      "loss": 0.4899,
      "step": 22150
    },
    {
      "epoch": 1.81,
      "learning_rate": 7.940026075619296e-06,
      "loss": 0.5192,
      "step": 22200
    },
    {
      "epoch": 1.81,
      "learning_rate": 7.912863972186006e-06,
      "loss": 0.4652,
      "step": 22250
    },
    {
      "epoch": 1.82,
      "learning_rate": 7.885701868752716e-06,
      "loss": 0.4865,
      "step": 22300
    },
    {
      "epoch": 1.82,
      "learning_rate": 7.858539765319428e-06,
      "loss": 0.5024,
      "step": 22350
    },
    {
      "epoch": 1.83,
      "learning_rate": 7.831377661886137e-06,
      "loss": 0.5156,
      "step": 22400
    },
    {
      "epoch": 1.83,
      "learning_rate": 7.804215558452847e-06,
      "loss": 0.5068,
      "step": 22450
    },
    {
      "epoch": 1.83,
      "learning_rate": 7.777053455019557e-06,
      "loss": 0.4798,
      "step": 22500
    },
    {
      "epoch": 1.84,
      "learning_rate": 7.749891351586267e-06,
      "loss": 0.4908,
      "step": 22550
    },
    {
      "epoch": 1.84,
      "learning_rate": 7.722729248152978e-06,
      "loss": 0.4852,
      "step": 22600
    },
    {
      "epoch": 1.85,
      "learning_rate": 7.695567144719688e-06,
      "loss": 0.5014,
      "step": 22650
    },
    {
      "epoch": 1.85,
      "learning_rate": 7.668405041286398e-06,
      "loss": 0.4777,
      "step": 22700
    },
    {
      "epoch": 1.85,
      "learning_rate": 7.641242937853108e-06,
      "loss": 0.4924,
      "step": 22750
    },
    {
      "epoch": 1.86,
      "learning_rate": 7.614080834419818e-06,
      "loss": 0.4686,
      "step": 22800
    },
    {
      "epoch": 1.86,
      "learning_rate": 7.5869187309865275e-06,
      "loss": 0.4535,
      "step": 22850
    },
    {
      "epoch": 1.87,
      "learning_rate": 7.559756627553238e-06,
      "loss": 0.488,
      "step": 22900
    },
    {
      "epoch": 1.87,
      "learning_rate": 7.532594524119948e-06,
      "loss": 0.4881,
      "step": 22950
    },
    {
      "epoch": 1.87,
      "learning_rate": 7.505432420686659e-06,
      "loss": 0.4768,
      "step": 23000
    },
    {
      "epoch": 1.88,
      "learning_rate": 7.4782703172533686e-06,
      "loss": 0.4947,
      "step": 23050
    },
    {
      "epoch": 1.88,
      "learning_rate": 7.451108213820078e-06,
      "loss": 0.4965,
      "step": 23100
    },
    {
      "epoch": 1.89,
      "learning_rate": 7.423946110386789e-06,
      "loss": 0.5233,
      "step": 23150
    },
    {
      "epoch": 1.89,
      "learning_rate": 7.396784006953499e-06,
      "loss": 0.4837,
      "step": 23200
    },
    {
      "epoch": 1.89,
      "learning_rate": 7.369621903520209e-06,
      "loss": 0.4922,
      "step": 23250
    },
    {
      "epoch": 1.9,
      "learning_rate": 7.3424598000869194e-06,
      "loss": 0.5411,
      "step": 23300
    },
    {
      "epoch": 1.9,
      "learning_rate": 7.315297696653629e-06,
      "loss": 0.5368,
      "step": 23350
    },
    {
      "epoch": 1.91,
      "learning_rate": 7.288135593220339e-06,
      "loss": 0.479,
      "step": 23400
    },
    {
      "epoch": 1.91,
      "learning_rate": 7.26097348978705e-06,
      "loss": 0.438,
      "step": 23450
    },
    {
      "epoch": 1.91,
      "learning_rate": 7.23381138635376e-06,
      "loss": 0.4863,
      "step": 23500
    },
    {
      "epoch": 1.92,
      "learning_rate": 7.20664928292047e-06,
      "loss": 0.4888,
      "step": 23550
    },
    {
      "epoch": 1.92,
      "learning_rate": 7.17948717948718e-06,
      "loss": 0.4937,
      "step": 23600
    },
    {
      "epoch": 1.93,
      "learning_rate": 7.15232507605389e-06,
      "loss": 0.4749,
      "step": 23650
    },
    {
      "epoch": 1.93,
      "learning_rate": 7.125162972620601e-06,
      "loss": 0.4869,
      "step": 23700
    },
    {
      "epoch": 1.94,
      "learning_rate": 7.0980008691873105e-06,
      "loss": 0.4876,
      "step": 23750
    },
    {
      "epoch": 1.94,
      "learning_rate": 7.07083876575402e-06,
      "loss": 0.4814,
      "step": 23800
    },
    {
      "epoch": 1.94,
      "learning_rate": 7.043676662320731e-06,
      "loss": 0.4893,
      "step": 23850
    },
    {
      "epoch": 1.95,
      "learning_rate": 7.016514558887441e-06,
      "loss": 0.4679,
      "step": 23900
    },
    {
      "epoch": 1.95,
      "learning_rate": 6.9893524554541515e-06,
      "loss": 0.46,
      "step": 23950
    },
    {
      "epoch": 1.96,
      "learning_rate": 6.962190352020861e-06,
      "loss": 0.4786,
      "step": 24000
    },
    {
      "epoch": 1.96,
      "learning_rate": 6.935028248587571e-06,
      "loss": 0.5118,
      "step": 24050
    },
    {
      "epoch": 1.96,
      "learning_rate": 6.907866145154282e-06,
      "loss": 0.4952,
      "step": 24100
    },
    {
      "epoch": 1.97,
      "learning_rate": 6.880704041720992e-06,
      "loss": 0.4985,
      "step": 24150
    },
    {
      "epoch": 1.97,
      "learning_rate": 6.8535419382877015e-06,
      "loss": 0.49,
      "step": 24200
    },
    {
      "epoch": 1.98,
      "learning_rate": 6.826379834854412e-06,
      "loss": 0.4873,
      "step": 24250
    },
    {
      "epoch": 1.98,
      "learning_rate": 6.799217731421122e-06,
      "loss": 0.4672,
      "step": 24300
    },
    {
      "epoch": 1.98,
      "learning_rate": 6.772055627987833e-06,
      "loss": 0.4655,
      "step": 24350
    },
    {
      "epoch": 1.99,
      "learning_rate": 6.7448935245545425e-06,
      "loss": 0.482,
      "step": 24400
    },
    {
      "epoch": 1.99,
      "learning_rate": 6.7177314211212515e-06,
      "loss": 0.4529,
      "step": 24450
    },
    {
      "epoch": 2.0,
      "learning_rate": 6.690569317687963e-06,
      "loss": 0.5119,
      "step": 24500
    },
    {
      "epoch": 2.0,
      "learning_rate": 6.663407214254672e-06,
      "loss": 0.4338,
      "step": 24550
    },
    {
      "epoch": 2.0,
      "learning_rate": 6.636245110821382e-06,
      "loss": 0.4019,
      "step": 24600
    },
    {
      "epoch": 2.01,
      "learning_rate": 6.6090830073880925e-06,
      "loss": 0.3948,
      "step": 24650
    },
    {
      "epoch": 2.01,
      "learning_rate": 6.581920903954802e-06,
      "loss": 0.395,
      "step": 24700
    },
    {
      "epoch": 2.02,
      "learning_rate": 6.554758800521513e-06,
      "loss": 0.4065,
      "step": 24750
    },
    {
      "epoch": 2.02,
      "learning_rate": 6.527596697088223e-06,
      "loss": 0.3928,
      "step": 24800
    },
    {
      "epoch": 2.02,
      "learning_rate": 6.500434593654933e-06,
      "loss": 0.3977,
      "step": 24850
    },
    {
      "epoch": 2.03,
      "learning_rate": 6.473272490221643e-06,
      "loss": 0.3697,
      "step": 24900
    },
    {
      "epoch": 2.03,
      "learning_rate": 6.446110386788353e-06,
      "loss": 0.4261,
      "step": 24950
    },
    {
      "epoch": 2.04,
      "learning_rate": 6.418948283355063e-06,
      "loss": 0.3754,
      "step": 25000
    },
    {
      "epoch": 2.04,
      "eval_accuracy": 0.7889964340295467,
      "eval_loss": 0.5795576572418213,
      "eval_runtime": 16.708,
      "eval_samples_per_second": 587.442,
      "eval_steps_per_second": 36.749,
      "step": 25000
    },
    {
      "epoch": 2.04,
      "learning_rate": 6.391786179921774e-06,
      "loss": 0.4236,
      "step": 25050
    },
    {
      "epoch": 2.05,
      "learning_rate": 6.364624076488484e-06,
      "loss": 0.4392,
      "step": 25100
    },
    {
      "epoch": 2.05,
      "learning_rate": 6.337461973055193e-06,
      "loss": 0.3784,
      "step": 25150
    },
    {
      "epoch": 2.05,
      "learning_rate": 6.310299869621904e-06,
      "loss": 0.3771,
      "step": 25200
    },
    {
      "epoch": 2.06,
      "learning_rate": 6.283137766188614e-06,
      "loss": 0.4049,
      "step": 25250
    },
    {
      "epoch": 2.06,
      "learning_rate": 6.255975662755325e-06,
      "loss": 0.4441,
      "step": 25300
    },
    {
      "epoch": 2.07,
      "learning_rate": 6.2288135593220344e-06,
      "loss": 0.3952,
      "step": 25350
    },
    {
      "epoch": 2.07,
      "learning_rate": 6.201651455888744e-06,
      "loss": 0.4196,
      "step": 25400
    },
    {
      "epoch": 2.07,
      "learning_rate": 6.174489352455455e-06,
      "loss": 0.3848,
      "step": 25450
    },
    {
      "epoch": 2.08,
      "learning_rate": 6.147327249022165e-06,
      "loss": 0.396,
      "step": 25500
    },
    {
      "epoch": 2.08,
      "learning_rate": 6.120165145588875e-06,
      "loss": 0.4099,
      "step": 25550
    },
    {
      "epoch": 2.09,
      "learning_rate": 6.093003042155585e-06,
      "loss": 0.3977,
      "step": 25600
    },
    {
      "epoch": 2.09,
      "learning_rate": 6.065840938722295e-06,
      "loss": 0.4118,
      "step": 25650
    },
    {
      "epoch": 2.09,
      "learning_rate": 6.038678835289006e-06,
      "loss": 0.378,
      "step": 25700
    },
    {
      "epoch": 2.1,
      "learning_rate": 6.011516731855716e-06,
      "loss": 0.3767,
      "step": 25750
    },
    {
      "epoch": 2.1,
      "learning_rate": 5.9843546284224255e-06,
      "loss": 0.4023,
      "step": 25800
    },
    {
      "epoch": 2.11,
      "learning_rate": 5.957192524989136e-06,
      "loss": 0.437,
      "step": 25850
    },
    {
      "epoch": 2.11,
      "learning_rate": 5.930030421555846e-06,
      "loss": 0.4221,
      "step": 25900
    },
    {
      "epoch": 2.11,
      "learning_rate": 5.902868318122556e-06,
      "loss": 0.4053,
      "step": 25950
    },
    {
      "epoch": 2.12,
      "learning_rate": 5.8757062146892665e-06,
      "loss": 0.3794,
      "step": 26000
    },
    {
      "epoch": 2.12,
      "learning_rate": 5.848544111255976e-06,
      "loss": 0.3832,
      "step": 26050
    },
    {
      "epoch": 2.13,
      "learning_rate": 5.821382007822687e-06,
      "loss": 0.373,
      "step": 26100
    },
    {
      "epoch": 2.13,
      "learning_rate": 5.794219904389397e-06,
      "loss": 0.3986,
      "step": 26150
    },
    {
      "epoch": 2.13,
      "learning_rate": 5.767057800956106e-06,
      "loss": 0.4046,
      "step": 26200
    },
    {
      "epoch": 2.14,
      "learning_rate": 5.739895697522817e-06,
      "loss": 0.404,
      "step": 26250
    },
    {
      "epoch": 2.14,
      "learning_rate": 5.712733594089526e-06,
      "loss": 0.3856,
      "step": 26300
    },
    {
      "epoch": 2.15,
      "learning_rate": 5.685571490656236e-06,
      "loss": 0.3753,
      "step": 26350
    },
    {
      "epoch": 2.15,
      "learning_rate": 5.658409387222948e-06,
      "loss": 0.4296,
      "step": 26400
    },
    {
      "epoch": 2.16,
      "learning_rate": 5.631247283789657e-06,
      "loss": 0.3694,
      "step": 26450
    },
    {
      "epoch": 2.16,
      "learning_rate": 5.6040851803563665e-06,
      "loss": 0.4106,
      "step": 26500
    },
    {
      "epoch": 2.16,
      "learning_rate": 5.576923076923077e-06,
      "loss": 0.3862,
      "step": 26550
    },
    {
      "epoch": 2.17,
      "learning_rate": 5.549760973489787e-06,
      "loss": 0.4276,
      "step": 26600
    },
    {
      "epoch": 2.17,
      "learning_rate": 5.522598870056498e-06,
      "loss": 0.4068,
      "step": 26650
    },
    {
      "epoch": 2.18,
      "learning_rate": 5.4954367666232076e-06,
      "loss": 0.3879,
      "step": 26700
    },
    {
      "epoch": 2.18,
      "learning_rate": 5.468274663189917e-06,
      "loss": 0.422,
      "step": 26750
    },
    {
      "epoch": 2.18,
      "learning_rate": 5.441112559756628e-06,
      "loss": 0.3924,
      "step": 26800
    },
    {
      "epoch": 2.19,
      "learning_rate": 5.413950456323338e-06,
      "loss": 0.4107,
      "step": 26850
    },
    {
      "epoch": 2.19,
      "learning_rate": 5.386788352890048e-06,
      "loss": 0.4056,
      "step": 26900
    },
    {
      "epoch": 2.2,
      "learning_rate": 5.359626249456758e-06,
      "loss": 0.3873,
      "step": 26950
    },
    {
      "epoch": 2.2,
      "learning_rate": 5.332464146023468e-06,
      "loss": 0.409,
      "step": 27000
    },
    {
      "epoch": 2.2,
      "learning_rate": 5.305302042590179e-06,
      "loss": 0.4301,
      "step": 27050
    },
    {
      "epoch": 2.21,
      "learning_rate": 5.278139939156889e-06,
      "loss": 0.409,
      "step": 27100
    },
    {
      "epoch": 2.21,
      "learning_rate": 5.250977835723599e-06,
      "loss": 0.4243,
      "step": 27150
    },
    {
      "epoch": 2.22,
      "learning_rate": 5.223815732290309e-06,
      "loss": 0.4252,
      "step": 27200
    },
    {
      "epoch": 2.22,
      "learning_rate": 5.196653628857019e-06,
      "loss": 0.405,
      "step": 27250
    },
    {
      "epoch": 2.22,
      "learning_rate": 5.169491525423729e-06,
      "loss": 0.374,
      "step": 27300
    },
    {
      "epoch": 2.23,
      "learning_rate": 5.14232942199044e-06,
      "loss": 0.4005,
      "step": 27350
    },
    {
      "epoch": 2.23,
      "learning_rate": 5.1151673185571495e-06,
      "loss": 0.4009,
      "step": 27400
    },
    {
      "epoch": 2.24,
      "learning_rate": 5.08800521512386e-06,
      "loss": 0.413,
      "step": 27450
    },
    {
      "epoch": 2.24,
      "learning_rate": 5.06084311169057e-06,
      "loss": 0.4043,
      "step": 27500
    },
    {
      "epoch": 2.24,
      "learning_rate": 5.03368100825728e-06,
      "loss": 0.3998,
      "step": 27550
    },
    {
      "epoch": 2.25,
      "learning_rate": 5.0065189048239905e-06,
      "loss": 0.441,
      "step": 27600
    },
    {
      "epoch": 2.25,
      "learning_rate": 4.9793568013907e-06,
      "loss": 0.4,
      "step": 27650
    },
    {
      "epoch": 2.26,
      "learning_rate": 4.95219469795741e-06,
      "loss": 0.3921,
      "step": 27700
    },
    {
      "epoch": 2.26,
      "learning_rate": 4.92503259452412e-06,
      "loss": 0.3924,
      "step": 27750
    },
    {
      "epoch": 2.27,
      "learning_rate": 4.897870491090831e-06,
      "loss": 0.4186,
      "step": 27800
    },
    {
      "epoch": 2.27,
      "learning_rate": 4.8707083876575405e-06,
      "loss": 0.3972,
      "step": 27850
    },
    {
      "epoch": 2.27,
      "learning_rate": 4.843546284224251e-06,
      "loss": 0.425,
      "step": 27900
    },
    {
      "epoch": 2.28,
      "learning_rate": 4.816384180790961e-06,
      "loss": 0.4067,
      "step": 27950
    },
    {
      "epoch": 2.28,
      "learning_rate": 4.789222077357671e-06,
      "loss": 0.4474,
      "step": 28000
    },
    {
      "epoch": 2.29,
      "learning_rate": 4.7620599739243815e-06,
      "loss": 0.4214,
      "step": 28050
    },
    {
      "epoch": 2.29,
      "learning_rate": 4.734897870491091e-06,
      "loss": 0.3861,
      "step": 28100
    },
    {
      "epoch": 2.29,
      "learning_rate": 4.707735767057801e-06,
      "loss": 0.4119,
      "step": 28150
    },
    {
      "epoch": 2.3,
      "learning_rate": 4.680573663624511e-06,
      "loss": 0.4046,
      "step": 28200
    },
    {
      "epoch": 2.3,
      "learning_rate": 4.653411560191222e-06,
      "loss": 0.3846,
      "step": 28250
    },
    {
      "epoch": 2.31,
      "learning_rate": 4.6262494567579315e-06,
      "loss": 0.4105,
      "step": 28300
    },
    {
      "epoch": 2.31,
      "learning_rate": 4.599087353324641e-06,
      "loss": 0.4202,
      "step": 28350
    },
    {
      "epoch": 2.31,
      "learning_rate": 4.571925249891352e-06,
      "loss": 0.3995,
      "step": 28400
    },
    {
      "epoch": 2.32,
      "learning_rate": 4.544763146458062e-06,
      "loss": 0.4221,
      "step": 28450
    },
    {
      "epoch": 2.32,
      "learning_rate": 4.5176010430247726e-06,
      "loss": 0.4233,
      "step": 28500
    },
    {
      "epoch": 2.33,
      "learning_rate": 4.490438939591482e-06,
      "loss": 0.4016,
      "step": 28550
    },
    {
      "epoch": 2.33,
      "learning_rate": 4.463276836158192e-06,
      "loss": 0.3899,
      "step": 28600
    },
    {
      "epoch": 2.33,
      "learning_rate": 4.436114732724903e-06,
      "loss": 0.4105,
      "step": 28650
    },
    {
      "epoch": 2.34,
      "learning_rate": 4.408952629291613e-06,
      "loss": 0.3675,
      "step": 28700
    },
    {
      "epoch": 2.34,
      "learning_rate": 4.381790525858323e-06,
      "loss": 0.3944,
      "step": 28750
    },
    {
      "epoch": 2.35,
      "learning_rate": 4.354628422425033e-06,
      "loss": 0.4149,
      "step": 28800
    },
    {
      "epoch": 2.35,
      "learning_rate": 4.327466318991743e-06,
      "loss": 0.3901,
      "step": 28850
    },
    {
      "epoch": 2.35,
      "learning_rate": 4.300304215558454e-06,
      "loss": 0.3982,
      "step": 28900
    },
    {
      "epoch": 2.36,
      "learning_rate": 4.273142112125163e-06,
      "loss": 0.4135,
      "step": 28950
    },
    {
      "epoch": 2.36,
      "learning_rate": 4.2459800086918734e-06,
      "loss": 0.4122,
      "step": 29000
    },
    {
      "epoch": 2.37,
      "learning_rate": 4.218817905258584e-06,
      "loss": 0.3816,
      "step": 29050
    },
    {
      "epoch": 2.37,
      "learning_rate": 4.191655801825294e-06,
      "loss": 0.3875,
      "step": 29100
    },
    {
      "epoch": 2.38,
      "learning_rate": 4.164493698392004e-06,
      "loss": 0.4247,
      "step": 29150
    },
    {
      "epoch": 2.38,
      "learning_rate": 4.137331594958714e-06,
      "loss": 0.4301,
      "step": 29200
    },
    {
      "epoch": 2.38,
      "learning_rate": 4.110169491525424e-06,
      "loss": 0.4442,
      "step": 29250
    },
    {
      "epoch": 2.39,
      "learning_rate": 4.083007388092134e-06,
      "loss": 0.4154,
      "step": 29300
    },
    {
      "epoch": 2.39,
      "learning_rate": 4.055845284658844e-06,
      "loss": 0.4089,
      "step": 29350
    },
    {
      "epoch": 2.4,
      "learning_rate": 4.028683181225555e-06,
      "loss": 0.3797,
      "step": 29400
    },
    {
      "epoch": 2.4,
      "learning_rate": 4.0015210777922645e-06,
      "loss": 0.4433,
      "step": 29450
    },
    {
      "epoch": 2.4,
      "learning_rate": 3.974358974358974e-06,
      "loss": 0.3922,
      "step": 29500
    },
    {
      "epoch": 2.41,
      "learning_rate": 3.947196870925685e-06,
      "loss": 0.4504,
      "step": 29550
    },
    {
      "epoch": 2.41,
      "learning_rate": 3.920034767492395e-06,
      "loss": 0.3998,
      "step": 29600
    },
    {
      "epoch": 2.42,
      "learning_rate": 3.8928726640591055e-06,
      "loss": 0.3771,
      "step": 29650
    },
    {
      "epoch": 2.42,
      "learning_rate": 3.865710560625815e-06,
      "loss": 0.4342,
      "step": 29700
    },
    {
      "epoch": 2.42,
      "learning_rate": 3.838548457192525e-06,
      "loss": 0.4114,
      "step": 29750
    },
    {
      "epoch": 2.43,
      "learning_rate": 3.8113863537592354e-06,
      "loss": 0.3966,
      "step": 29800
    },
    {
      "epoch": 2.43,
      "learning_rate": 3.7842242503259457e-06,
      "loss": 0.411,
      "step": 29850
    },
    {
      "epoch": 2.44,
      "learning_rate": 3.7570621468926555e-06,
      "loss": 0.4119,
      "step": 29900
    },
    {
      "epoch": 2.44,
      "learning_rate": 3.7299000434593658e-06,
      "loss": 0.4471,
      "step": 29950
    },
    {
      "epoch": 2.44,
      "learning_rate": 3.702737940026076e-06,
      "loss": 0.3989,
      "step": 30000
    },
    {
      "epoch": 2.44,
      "eval_accuracy": 0.7892002037697402,
      "eval_loss": 0.5580592751502991,
      "eval_runtime": 16.7363,
      "eval_samples_per_second": 586.449,
      "eval_steps_per_second": 36.687,
      "step": 30000
    },
    {
      "epoch": 2.45,
      "learning_rate": 3.6755758365927863e-06,
      "loss": 0.3757,
      "step": 30050
    },
    {
      "epoch": 2.45,
      "learning_rate": 3.648413733159496e-06,
      "loss": 0.3962,
      "step": 30100
    },
    {
      "epoch": 2.46,
      "learning_rate": 3.6212516297262064e-06,
      "loss": 0.4181,
      "step": 30150
    },
    {
      "epoch": 2.46,
      "learning_rate": 3.5940895262929166e-06,
      "loss": 0.4102,
      "step": 30200
    },
    {
      "epoch": 2.46,
      "learning_rate": 3.566927422859627e-06,
      "loss": 0.4232,
      "step": 30250
    },
    {
      "epoch": 2.47,
      "learning_rate": 3.5397653194263363e-06,
      "loss": 0.4225,
      "step": 30300
    },
    {
      "epoch": 2.47,
      "learning_rate": 3.5126032159930466e-06,
      "loss": 0.4211,
      "step": 30350
    },
    {
      "epoch": 2.48,
      "learning_rate": 3.4854411125597572e-06,
      "loss": 0.4046,
      "step": 30400
    },
    {
      "epoch": 2.48,
      "learning_rate": 3.4582790091264675e-06,
      "loss": 0.424,
      "step": 30450
    },
    {
      "epoch": 2.49,
      "learning_rate": 3.431116905693177e-06,
      "loss": 0.4039,
      "step": 30500
    },
    {
      "epoch": 2.49,
      "learning_rate": 3.403954802259887e-06,
      "loss": 0.4225,
      "step": 30550
    },
    {
      "epoch": 2.49,
      "learning_rate": 3.3767926988265974e-06,
      "loss": 0.3794,
      "step": 30600
    },
    {
      "epoch": 2.5,
      "learning_rate": 3.3496305953933073e-06,
      "loss": 0.4136,
      "step": 30650
    },
    {
      "epoch": 2.5,
      "learning_rate": 3.3224684919600175e-06,
      "loss": 0.3917,
      "step": 30700
    },
    {
      "epoch": 2.51,
      "learning_rate": 3.2953063885267278e-06,
      "loss": 0.3823,
      "step": 30750
    },
    {
      "epoch": 2.51,
      "learning_rate": 3.268144285093438e-06,
      "loss": 0.3955,
      "step": 30800
    },
    {
      "epoch": 2.51,
      "learning_rate": 3.240982181660148e-06,
      "loss": 0.4004,
      "step": 30850
    },
    {
      "epoch": 2.52,
      "learning_rate": 3.213820078226858e-06,
      "loss": 0.4052,
      "step": 30900
    },
    {
      "epoch": 2.52,
      "learning_rate": 3.1866579747935684e-06,
      "loss": 0.4154,
      "step": 30950
    },
    {
      "epoch": 2.53,
      "learning_rate": 3.1594958713602786e-06,
      "loss": 0.3727,
      "step": 31000
    },
    {
      "epoch": 2.53,
      "learning_rate": 3.1323337679269885e-06,
      "loss": 0.3971,
      "step": 31050
    },
    {
      "epoch": 2.53,
      "learning_rate": 3.1051716644936987e-06,
      "loss": 0.3984,
      "step": 31100
    },
    {
      "epoch": 2.54,
      "learning_rate": 3.078009561060409e-06,
      "loss": 0.3918,
      "step": 31150
    },
    {
      "epoch": 2.54,
      "learning_rate": 3.0508474576271192e-06,
      "loss": 0.4079,
      "step": 31200
    },
    {
      "epoch": 2.55,
      "learning_rate": 3.0236853541938286e-06,
      "loss": 0.3879,
      "step": 31250
    },
    {
      "epoch": 2.55,
      "learning_rate": 2.996523250760539e-06,
      "loss": 0.3808,
      "step": 31300
    },
    {
      "epoch": 2.55,
      "learning_rate": 2.969361147327249e-06,
      "loss": 0.4068,
      "step": 31350
    },
    {
      "epoch": 2.56,
      "learning_rate": 2.9421990438939594e-06,
      "loss": 0.4029,
      "step": 31400
    },
    {
      "epoch": 2.56,
      "learning_rate": 2.9150369404606692e-06,
      "loss": 0.3524,
      "step": 31450
    },
    {
      "epoch": 2.57,
      "learning_rate": 2.8878748370273795e-06,
      "loss": 0.4021,
      "step": 31500
    },
    {
      "epoch": 2.57,
      "learning_rate": 2.8607127335940898e-06,
      "loss": 0.3924,
      "step": 31550
    },
    {
      "epoch": 2.57,
      "learning_rate": 2.8335506301608e-06,
      "loss": 0.4303,
      "step": 31600
    },
    {
      "epoch": 2.58,
      "learning_rate": 2.80638852672751e-06,
      "loss": 0.3993,
      "step": 31650
    },
    {
      "epoch": 2.58,
      "learning_rate": 2.77922642329422e-06,
      "loss": 0.3992,
      "step": 31700
    },
    {
      "epoch": 2.59,
      "learning_rate": 2.7520643198609304e-06,
      "loss": 0.4215,
      "step": 31750
    },
    {
      "epoch": 2.59,
      "learning_rate": 2.7249022164276406e-06,
      "loss": 0.4107,
      "step": 31800
    },
    {
      "epoch": 2.6,
      "learning_rate": 2.6977401129943505e-06,
      "loss": 0.3876,
      "step": 31850
    },
    {
      "epoch": 2.6,
      "learning_rate": 2.6705780095610607e-06,
      "loss": 0.3873,
      "step": 31900
    },
    {
      "epoch": 2.6,
      "learning_rate": 2.643415906127771e-06,
      "loss": 0.4126,
      "step": 31950
    },
    {
      "epoch": 2.61,
      "learning_rate": 2.6162538026944812e-06,
      "loss": 0.4173,
      "step": 32000
    },
    {
      "epoch": 2.61,
      "learning_rate": 2.589091699261191e-06,
      "loss": 0.409,
      "step": 32050
    },
    {
      "epoch": 2.62,
      "learning_rate": 2.5619295958279013e-06,
      "loss": 0.4129,
      "step": 32100
    },
    {
      "epoch": 2.62,
      "learning_rate": 2.5347674923946116e-06,
      "loss": 0.3676,
      "step": 32150
    },
    {
      "epoch": 2.62,
      "learning_rate": 2.507605388961321e-06,
      "loss": 0.3939,
      "step": 32200
    },
    {
      "epoch": 2.63,
      "learning_rate": 2.4804432855280312e-06,
      "loss": 0.3828,
      "step": 32250
    },
    {
      "epoch": 2.63,
      "learning_rate": 2.4532811820947415e-06,
      "loss": 0.418,
      "step": 32300
    },
    {
      "epoch": 2.64,
      "learning_rate": 2.4261190786614517e-06,
      "loss": 0.4142,
      "step": 32350
    },
    {
      "epoch": 2.64,
      "learning_rate": 2.398956975228162e-06,
      "loss": 0.398,
      "step": 32400
    },
    {
      "epoch": 2.64,
      "learning_rate": 2.371794871794872e-06,
      "loss": 0.4077,
      "step": 32450
    },
    {
      "epoch": 2.65,
      "learning_rate": 2.344632768361582e-06,
      "loss": 0.412,
      "step": 32500
    },
    {
      "epoch": 2.65,
      "learning_rate": 2.3174706649282924e-06,
      "loss": 0.4163,
      "step": 32550
    },
    {
      "epoch": 2.66,
      "learning_rate": 2.2903085614950026e-06,
      "loss": 0.4202,
      "step": 32600
    },
    {
      "epoch": 2.66,
      "learning_rate": 2.2631464580617124e-06,
      "loss": 0.4121,
      "step": 32650
    },
    {
      "epoch": 2.66,
      "learning_rate": 2.2359843546284227e-06,
      "loss": 0.4,
      "step": 32700
    },
    {
      "epoch": 2.67,
      "learning_rate": 2.2088222511951325e-06,
      "loss": 0.4008,
      "step": 32750
    },
    {
      "epoch": 2.67,
      "learning_rate": 2.1816601477618428e-06,
      "loss": 0.4103,
      "step": 32800
    },
    {
      "epoch": 2.68,
      "learning_rate": 2.154498044328553e-06,
      "loss": 0.3995,
      "step": 32850
    },
    {
      "epoch": 2.68,
      "learning_rate": 2.1273359408952633e-06,
      "loss": 0.3924,
      "step": 32900
    },
    {
      "epoch": 2.68,
      "learning_rate": 2.100173837461973e-06,
      "loss": 0.431,
      "step": 32950
    },
    {
      "epoch": 2.69,
      "learning_rate": 2.0730117340286834e-06,
      "loss": 0.4029,
      "step": 33000
    },
    {
      "epoch": 2.69,
      "learning_rate": 2.0458496305953932e-06,
      "loss": 0.3988,
      "step": 33050
    },
    {
      "epoch": 2.7,
      "learning_rate": 2.018687527162104e-06,
      "loss": 0.4012,
      "step": 33100
    },
    {
      "epoch": 2.7,
      "learning_rate": 1.9915254237288137e-06,
      "loss": 0.3842,
      "step": 33150
    },
    {
      "epoch": 2.71,
      "learning_rate": 1.964363320295524e-06,
      "loss": 0.3976,
      "step": 33200
    },
    {
      "epoch": 2.71,
      "learning_rate": 1.937201216862234e-06,
      "loss": 0.4081,
      "step": 33250
    },
    {
      "epoch": 2.71,
      "learning_rate": 1.910039113428944e-06,
      "loss": 0.3973,
      "step": 33300
    },
    {
      "epoch": 2.72,
      "learning_rate": 1.8828770099956541e-06,
      "loss": 0.4087,
      "step": 33350
    },
    {
      "epoch": 2.72,
      "learning_rate": 1.8557149065623644e-06,
      "loss": 0.387,
      "step": 33400
    },
    {
      "epoch": 2.73,
      "learning_rate": 1.8285528031290744e-06,
      "loss": 0.3997,
      "step": 33450
    },
    {
      "epoch": 2.73,
      "learning_rate": 1.8013906996957845e-06,
      "loss": 0.3971,
      "step": 33500
    },
    {
      "epoch": 2.73,
      "learning_rate": 1.7742285962624947e-06,
      "loss": 0.3768,
      "step": 33550
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.7470664928292048e-06,
      "loss": 0.364,
      "step": 33600
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.719904389395915e-06,
      "loss": 0.4112,
      "step": 33650
    },
    {
      "epoch": 2.75,
      "learning_rate": 1.6927422859626249e-06,
      "loss": 0.4038,
      "step": 33700
    },
    {
      "epoch": 2.75,
      "learning_rate": 1.6655801825293353e-06,
      "loss": 0.365,
      "step": 33750
    },
    {
      "epoch": 2.75,
      "learning_rate": 1.6384180790960452e-06,
      "loss": 0.4264,
      "step": 33800
    },
    {
      "epoch": 2.76,
      "learning_rate": 1.6112559756627554e-06,
      "loss": 0.4099,
      "step": 33850
    },
    {
      "epoch": 2.76,
      "learning_rate": 1.5840938722294655e-06,
      "loss": 0.3823,
      "step": 33900
    },
    {
      "epoch": 2.77,
      "learning_rate": 1.5569317687961757e-06,
      "loss": 0.427,
      "step": 33950
    },
    {
      "epoch": 2.77,
      "learning_rate": 1.5297696653628858e-06,
      "loss": 0.3966,
      "step": 34000
    },
    {
      "epoch": 2.77,
      "learning_rate": 1.502607561929596e-06,
      "loss": 0.4186,
      "step": 34050
    },
    {
      "epoch": 2.78,
      "learning_rate": 1.475445458496306e-06,
      "loss": 0.3757,
      "step": 34100
    },
    {
      "epoch": 2.78,
      "learning_rate": 1.4482833550630163e-06,
      "loss": 0.3756,
      "step": 34150
    },
    {
      "epoch": 2.79,
      "learning_rate": 1.4211212516297262e-06,
      "loss": 0.3846,
      "step": 34200
    },
    {
      "epoch": 2.79,
      "learning_rate": 1.3939591481964364e-06,
      "loss": 0.425,
      "step": 34250
    },
    {
      "epoch": 2.79,
      "learning_rate": 1.3667970447631465e-06,
      "loss": 0.4,
      "step": 34300
    },
    {
      "epoch": 2.8,
      "learning_rate": 1.3396349413298567e-06,
      "loss": 0.4014,
      "step": 34350
    },
    {
      "epoch": 2.8,
      "learning_rate": 1.3124728378965668e-06,
      "loss": 0.4189,
      "step": 34400
    },
    {
      "epoch": 2.81,
      "learning_rate": 1.285310734463277e-06,
      "loss": 0.4223,
      "step": 34450
    },
    {
      "epoch": 2.81,
      "learning_rate": 1.258148631029987e-06,
      "loss": 0.3971,
      "step": 34500
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.2309865275966971e-06,
      "loss": 0.4106,
      "step": 34550
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.2038244241634074e-06,
      "loss": 0.4021,
      "step": 34600
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.1766623207301174e-06,
      "loss": 0.4022,
      "step": 34650
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.1495002172968275e-06,
      "loss": 0.3845,
      "step": 34700
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.1223381138635377e-06,
      "loss": 0.3963,
      "step": 34750
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.0951760104302478e-06,
      "loss": 0.4259,
      "step": 34800
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.068013906996958e-06,
      "loss": 0.434,
      "step": 34850
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.040851803563668e-06,
      "loss": 0.3829,
      "step": 34900
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.0136897001303781e-06,
      "loss": 0.3953,
      "step": 34950
    },
    {
      "epoch": 2.85,
      "learning_rate": 9.865275966970884e-07,
      "loss": 0.4013,
      "step": 35000
    },
    {
      "epoch": 2.85,
      "eval_accuracy": 0.7955170657157412,
      "eval_loss": 0.5501273274421692,
      "eval_runtime": 16.8148,
      "eval_samples_per_second": 583.712,
      "eval_steps_per_second": 36.515,
      "step": 35000
    },
    {
      "epoch": 2.86,
      "learning_rate": 9.593654932637984e-07,
      "loss": 0.3919,
      "step": 35050
    },
    {
      "epoch": 2.86,
      "learning_rate": 9.322033898305086e-07,
      "loss": 0.4299,
      "step": 35100
    },
    {
      "epoch": 2.86,
      "learning_rate": 9.050412863972187e-07,
      "loss": 0.4219,
      "step": 35150
    },
    {
      "epoch": 2.87,
      "learning_rate": 8.778791829639289e-07,
      "loss": 0.3997,
      "step": 35200
    },
    {
      "epoch": 2.87,
      "learning_rate": 8.507170795306389e-07,
      "loss": 0.4138,
      "step": 35250
    },
    {
      "epoch": 2.88,
      "learning_rate": 8.235549760973491e-07,
      "loss": 0.3842,
      "step": 35300
    },
    {
      "epoch": 2.88,
      "learning_rate": 7.963928726640592e-07,
      "loss": 0.4033,
      "step": 35350
    },
    {
      "epoch": 2.88,
      "learning_rate": 7.692307692307694e-07,
      "loss": 0.4052,
      "step": 35400
    },
    {
      "epoch": 2.89,
      "learning_rate": 7.420686657974795e-07,
      "loss": 0.4042,
      "step": 35450
    },
    {
      "epoch": 2.89,
      "learning_rate": 7.149065623641896e-07,
      "loss": 0.4045,
      "step": 35500
    },
    {
      "epoch": 2.9,
      "learning_rate": 6.877444589308997e-07,
      "loss": 0.4073,
      "step": 35550
    },
    {
      "epoch": 2.9,
      "learning_rate": 6.605823554976099e-07,
      "loss": 0.4266,
      "step": 35600
    },
    {
      "epoch": 2.9,
      "learning_rate": 6.3342025206432e-07,
      "loss": 0.3942,
      "step": 35650
    },
    {
      "epoch": 2.91,
      "learning_rate": 6.0625814863103e-07,
      "loss": 0.4103,
      "step": 35700
    },
    {
      "epoch": 2.91,
      "learning_rate": 5.790960451977402e-07,
      "loss": 0.4249,
      "step": 35750
    },
    {
      "epoch": 2.92,
      "learning_rate": 5.519339417644502e-07,
      "loss": 0.4074,
      "step": 35800
    },
    {
      "epoch": 2.92,
      "learning_rate": 5.247718383311604e-07,
      "loss": 0.4009,
      "step": 35850
    },
    {
      "epoch": 2.93,
      "learning_rate": 4.976097348978705e-07,
      "loss": 0.3856,
      "step": 35900
    },
    {
      "epoch": 2.93,
      "learning_rate": 4.704476314645807e-07,
      "loss": 0.3941,
      "step": 35950
    },
    {
      "epoch": 2.93,
      "learning_rate": 4.432855280312908e-07,
      "loss": 0.4133,
      "step": 36000
    },
    {
      "epoch": 2.94,
      "learning_rate": 4.1612342459800084e-07,
      "loss": 0.4005,
      "step": 36050
    },
    {
      "epoch": 2.94,
      "learning_rate": 3.88961321164711e-07,
      "loss": 0.368,
      "step": 36100
    },
    {
      "epoch": 2.95,
      "learning_rate": 3.6179921773142114e-07,
      "loss": 0.4314,
      "step": 36150
    },
    {
      "epoch": 2.95,
      "learning_rate": 3.3463711429813124e-07,
      "loss": 0.4145,
      "step": 36200
    },
    {
      "epoch": 2.95,
      "learning_rate": 3.074750108648414e-07,
      "loss": 0.4122,
      "step": 36250
    },
    {
      "epoch": 2.96,
      "learning_rate": 2.803129074315515e-07,
      "loss": 0.4063,
      "step": 36300
    },
    {
      "epoch": 2.96,
      "learning_rate": 2.5315080399826164e-07,
      "loss": 0.415,
      "step": 36350
    },
    {
      "epoch": 2.97,
      "learning_rate": 2.2598870056497177e-07,
      "loss": 0.3831,
      "step": 36400
    },
    {
      "epoch": 2.97,
      "learning_rate": 1.988265971316819e-07,
      "loss": 0.3935,
      "step": 36450
    },
    {
      "epoch": 2.97,
      "learning_rate": 1.7166449369839202e-07,
      "loss": 0.3924,
      "step": 36500
    },
    {
      "epoch": 2.98,
      "learning_rate": 1.4450239026510214e-07,
      "loss": 0.375,
      "step": 36550
    },
    {
      "epoch": 2.98,
      "learning_rate": 1.1734028683181226e-07,
      "loss": 0.4107,
      "step": 36600
    },
    {
      "epoch": 2.99,
      "learning_rate": 9.017818339852239e-08,
      "loss": 0.4141,
      "step": 36650
    },
    {
      "epoch": 2.99,
      "learning_rate": 6.301607996523251e-08,
      "loss": 0.4065,
      "step": 36700
    },
    {
      "epoch": 2.99,
      "learning_rate": 3.585397653194264e-08,
      "loss": 0.3686,
      "step": 36750
    },
    {
      "epoch": 3.0,
      "learning_rate": 8.691873098652761e-09,
      "loss": 0.4018,
      "step": 36800
    },
    {
      "epoch": 3.0,
      "step": 36816,
      "total_flos": 7.237278554846584e+16,
      "train_loss": 0.5138862587508302,
      "train_runtime": 6461.7512,
      "train_samples_per_second": 182.32,
      "train_steps_per_second": 5.698
    }
  ],
  "logging_steps": 50,
  "max_steps": 36816,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 5000,
  "total_flos": 7.237278554846584e+16,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}