{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "global_step": 78750,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02,
      "learning_rate": 4.968253968253969e-05,
      "loss": 0.8671,
      "step": 500
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.936507936507937e-05,
      "loss": 0.2133,
      "step": 1000
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.904761904761905e-05,
      "loss": 0.1414,
      "step": 1500
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.873015873015873e-05,
      "loss": 0.1091,
      "step": 2000
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.841269841269841e-05,
      "loss": 0.0924,
      "step": 2500
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.80952380952381e-05,
      "loss": 0.0813,
      "step": 3000
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7777777777777784e-05,
      "loss": 0.0782,
      "step": 3500
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.746031746031746e-05,
      "loss": 0.0707,
      "step": 4000
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.714285714285714e-05,
      "loss": 0.0672,
      "step": 4500
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.682539682539683e-05,
      "loss": 0.0624,
      "step": 5000
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.6507936507936515e-05,
      "loss": 0.0595,
      "step": 5500
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.6190476190476194e-05,
      "loss": 0.0584,
      "step": 6000
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.587301587301587e-05,
      "loss": 0.0518,
      "step": 6500
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.555555555555556e-05,
      "loss": 0.05,
      "step": 7000
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.523809523809524e-05,
      "loss": 0.0498,
      "step": 7500
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.4920634920634924e-05,
      "loss": 0.0451,
      "step": 8000
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.460317460317461e-05,
      "loss": 0.0513,
      "step": 8500
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.428571428571428e-05,
      "loss": 0.0516,
      "step": 9000
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.396825396825397e-05,
      "loss": 0.0464,
      "step": 9500
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.3650793650793655e-05,
      "loss": 0.0464,
      "step": 10000
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.3333333333333334e-05,
      "loss": 0.0424,
      "step": 10500
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.301587301587302e-05,
      "loss": 0.044,
      "step": 11000
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.26984126984127e-05,
      "loss": 0.0441,
      "step": 11500
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.2380952380952385e-05,
      "loss": 0.0382,
      "step": 12000
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.2063492063492065e-05,
      "loss": 0.0428,
      "step": 12500
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.174603174603175e-05,
      "loss": 0.038,
      "step": 13000
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.1428571428571437e-05,
      "loss": 0.0416,
      "step": 13500
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.111111111111111e-05,
      "loss": 0.0385,
      "step": 14000
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.0793650793650795e-05,
      "loss": 0.0387,
      "step": 14500
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.047619047619048e-05,
      "loss": 0.0392,
      "step": 15000
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.015873015873016e-05,
      "loss": 0.0365,
      "step": 15500
    },
    {
      "epoch": 0.61,
      "learning_rate": 3.984126984126984e-05,
      "loss": 0.0409,
      "step": 16000
    },
    {
      "epoch": 0.63,
      "learning_rate": 3.9523809523809526e-05,
      "loss": 0.0385,
      "step": 16500
    },
    {
      "epoch": 0.65,
      "learning_rate": 3.9206349206349205e-05,
      "loss": 0.0361,
      "step": 17000
    },
    {
      "epoch": 0.67,
      "learning_rate": 3.888888888888889e-05,
      "loss": 0.0309,
      "step": 17500
    },
    {
      "epoch": 0.69,
      "learning_rate": 3.857142857142858e-05,
      "loss": 0.0348,
      "step": 18000
    },
    {
      "epoch": 0.7,
      "learning_rate": 3.8253968253968256e-05,
      "loss": 0.0321,
      "step": 18500
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.7936507936507935e-05,
      "loss": 0.0331,
      "step": 19000
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.761904761904762e-05,
      "loss": 0.0332,
      "step": 19500
    },
    {
      "epoch": 0.76,
      "learning_rate": 3.730158730158731e-05,
      "loss": 0.0304,
      "step": 20000
    },
    {
      "epoch": 0.78,
      "learning_rate": 3.6984126984126986e-05,
      "loss": 0.0348,
      "step": 20500
    },
    {
      "epoch": 0.8,
      "learning_rate": 3.6666666666666666e-05,
      "loss": 0.0328,
      "step": 21000
    },
    {
      "epoch": 0.82,
      "learning_rate": 3.634920634920635e-05,
      "loss": 0.0297,
      "step": 21500
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.603174603174603e-05,
      "loss": 0.0316,
      "step": 22000
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.571428571428572e-05,
      "loss": 0.0294,
      "step": 22500
    },
    {
      "epoch": 0.88,
      "learning_rate": 3.53968253968254e-05,
      "loss": 0.0311,
      "step": 23000
    },
    {
      "epoch": 0.9,
      "learning_rate": 3.5079365079365075e-05,
      "loss": 0.0281,
      "step": 23500
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.476190476190476e-05,
      "loss": 0.0286,
      "step": 24000
    },
    {
      "epoch": 0.93,
      "learning_rate": 3.444444444444445e-05,
      "loss": 0.0316,
      "step": 24500
    },
    {
      "epoch": 0.95,
      "learning_rate": 3.412698412698413e-05,
      "loss": 0.0308,
      "step": 25000
    },
    {
      "epoch": 0.97,
      "learning_rate": 3.380952380952381e-05,
      "loss": 0.0263,
      "step": 25500
    },
    {
      "epoch": 0.99,
      "learning_rate": 3.349206349206349e-05,
      "loss": 0.0329,
      "step": 26000
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.9930055555555556,
      "eval_f1-score": 0.9929596586748287,
      "eval_loss": 0.024911481887102127,
      "eval_runtime": 2756.559,
      "eval_samples_per_second": 261.195,
      "eval_steps_per_second": 2.041,
      "step": 26250
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.317460317460318e-05,
      "loss": 0.0262,
      "step": 26500
    },
    {
      "epoch": 1.03,
      "learning_rate": 3.285714285714286e-05,
      "loss": 0.0259,
      "step": 27000
    },
    {
      "epoch": 1.05,
      "learning_rate": 3.253968253968254e-05,
      "loss": 0.0283,
      "step": 27500
    },
    {
      "epoch": 1.07,
      "learning_rate": 3.222222222222223e-05,
      "loss": 0.0264,
      "step": 28000
    },
    {
      "epoch": 1.09,
      "learning_rate": 3.19047619047619e-05,
      "loss": 0.0243,
      "step": 28500
    },
    {
      "epoch": 1.1,
      "learning_rate": 3.158730158730159e-05,
      "loss": 0.0275,
      "step": 29000
    },
    {
      "epoch": 1.12,
      "learning_rate": 3.1269841269841274e-05,
      "loss": 0.0284,
      "step": 29500
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.095238095238095e-05,
      "loss": 0.0249,
      "step": 30000
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.063492063492064e-05,
      "loss": 0.0267,
      "step": 30500
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.0317460317460318e-05,
      "loss": 0.0283,
      "step": 31000
    },
    {
      "epoch": 1.2,
      "learning_rate": 3e-05,
      "loss": 0.0223,
      "step": 31500
    },
    {
      "epoch": 1.22,
      "learning_rate": 2.9682539682539683e-05,
      "loss": 0.0267,
      "step": 32000
    },
    {
      "epoch": 1.24,
      "learning_rate": 2.9365079365079366e-05,
      "loss": 0.0211,
      "step": 32500
    },
    {
      "epoch": 1.26,
      "learning_rate": 2.9047619047619052e-05,
      "loss": 0.0226,
      "step": 33000
    },
    {
      "epoch": 1.28,
      "learning_rate": 2.8730158730158728e-05,
      "loss": 0.0221,
      "step": 33500
    },
    {
      "epoch": 1.3,
      "learning_rate": 2.8412698412698414e-05,
      "loss": 0.0227,
      "step": 34000
    },
    {
      "epoch": 1.31,
      "learning_rate": 2.8095238095238096e-05,
      "loss": 0.0239,
      "step": 34500
    },
    {
      "epoch": 1.33,
      "learning_rate": 2.777777777777778e-05,
      "loss": 0.0254,
      "step": 35000
    },
    {
      "epoch": 1.35,
      "learning_rate": 2.7460317460317465e-05,
      "loss": 0.0236,
      "step": 35500
    },
    {
      "epoch": 1.37,
      "learning_rate": 2.714285714285714e-05,
      "loss": 0.0226,
      "step": 36000
    },
    {
      "epoch": 1.39,
      "learning_rate": 2.6825396825396827e-05,
      "loss": 0.021,
      "step": 36500
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.650793650793651e-05,
      "loss": 0.0259,
      "step": 37000
    },
    {
      "epoch": 1.43,
      "learning_rate": 2.6190476190476192e-05,
      "loss": 0.0241,
      "step": 37500
    },
    {
      "epoch": 1.45,
      "learning_rate": 2.5873015873015878e-05,
      "loss": 0.0226,
      "step": 38000
    },
    {
      "epoch": 1.47,
      "learning_rate": 2.5555555555555554e-05,
      "loss": 0.023,
      "step": 38500
    },
    {
      "epoch": 1.49,
      "learning_rate": 2.523809523809524e-05,
      "loss": 0.0236,
      "step": 39000
    },
    {
      "epoch": 1.5,
      "learning_rate": 2.4920634920634923e-05,
      "loss": 0.0216,
      "step": 39500
    },
    {
      "epoch": 1.52,
      "learning_rate": 2.4603174603174602e-05,
      "loss": 0.0221,
      "step": 40000
    },
    {
      "epoch": 1.54,
      "learning_rate": 2.4285714285714288e-05,
      "loss": 0.0219,
      "step": 40500
    },
    {
      "epoch": 1.56,
      "learning_rate": 2.396825396825397e-05,
      "loss": 0.022,
      "step": 41000
    },
    {
      "epoch": 1.58,
      "learning_rate": 2.365079365079365e-05,
      "loss": 0.024,
      "step": 41500
    },
    {
      "epoch": 1.6,
      "learning_rate": 2.3333333333333336e-05,
      "loss": 0.021,
      "step": 42000
    },
    {
      "epoch": 1.62,
      "learning_rate": 2.3015873015873015e-05,
      "loss": 0.0219,
      "step": 42500
    },
    {
      "epoch": 1.64,
      "learning_rate": 2.2698412698412698e-05,
      "loss": 0.0199,
      "step": 43000
    },
    {
      "epoch": 1.66,
      "learning_rate": 2.2380952380952384e-05,
      "loss": 0.0208,
      "step": 43500
    },
    {
      "epoch": 1.68,
      "learning_rate": 2.2063492063492063e-05,
      "loss": 0.0198,
      "step": 44000
    },
    {
      "epoch": 1.7,
      "learning_rate": 2.174603174603175e-05,
      "loss": 0.0217,
      "step": 44500
    },
    {
      "epoch": 1.71,
      "learning_rate": 2.1428571428571428e-05,
      "loss": 0.0189,
      "step": 45000
    },
    {
      "epoch": 1.73,
      "learning_rate": 2.111111111111111e-05,
      "loss": 0.0199,
      "step": 45500
    },
    {
      "epoch": 1.75,
      "learning_rate": 2.0793650793650797e-05,
      "loss": 0.0208,
      "step": 46000
    },
    {
      "epoch": 1.77,
      "learning_rate": 2.0476190476190476e-05,
      "loss": 0.0196,
      "step": 46500
    },
    {
      "epoch": 1.79,
      "learning_rate": 2.015873015873016e-05,
      "loss": 0.0197,
      "step": 47000
    },
    {
      "epoch": 1.81,
      "learning_rate": 1.984126984126984e-05,
      "loss": 0.0197,
      "step": 47500
    },
    {
      "epoch": 1.83,
      "learning_rate": 1.9523809523809524e-05,
      "loss": 0.0184,
      "step": 48000
    },
    {
      "epoch": 1.85,
      "learning_rate": 1.920634920634921e-05,
      "loss": 0.0166,
      "step": 48500
    },
    {
      "epoch": 1.87,
      "learning_rate": 1.888888888888889e-05,
      "loss": 0.0217,
      "step": 49000
    },
    {
      "epoch": 1.89,
      "learning_rate": 1.8571428571428572e-05,
      "loss": 0.0189,
      "step": 49500
    },
    {
      "epoch": 1.9,
      "learning_rate": 1.8253968253968254e-05,
      "loss": 0.018,
      "step": 50000
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.7936507936507937e-05,
      "loss": 0.0182,
      "step": 50500
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.761904761904762e-05,
      "loss": 0.0176,
      "step": 51000
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.7301587301587302e-05,
      "loss": 0.0163,
      "step": 51500
    },
    {
      "epoch": 1.98,
      "learning_rate": 1.6984126984126985e-05,
      "loss": 0.0174,
      "step": 52000
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 0.0168,
      "step": 52500
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.9950902777777778,
      "eval_f1-score": 0.99507649644856,
      "eval_loss": 0.01802189275622368,
      "eval_runtime": 2757.205,
      "eval_samples_per_second": 261.134,
      "eval_steps_per_second": 2.04,
      "step": 52500
    },
    {
      "epoch": 2.02,
      "learning_rate": 1.634920634920635e-05,
      "loss": 0.0165,
      "step": 53000
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.6031746031746033e-05,
      "loss": 0.0165,
      "step": 53500
    },
    {
      "epoch": 2.06,
      "learning_rate": 1.5714285714285715e-05,
      "loss": 0.0168,
      "step": 54000
    },
    {
      "epoch": 2.08,
      "learning_rate": 1.5396825396825398e-05,
      "loss": 0.0169,
      "step": 54500
    },
    {
      "epoch": 2.1,
      "learning_rate": 1.5079365079365079e-05,
      "loss": 0.0149,
      "step": 55000
    },
    {
      "epoch": 2.11,
      "learning_rate": 1.4761904761904763e-05,
      "loss": 0.0157,
      "step": 55500
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.4444444444444444e-05,
      "loss": 0.0146,
      "step": 56000
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.4126984126984127e-05,
      "loss": 0.0139,
      "step": 56500
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.3809523809523811e-05,
      "loss": 0.0158,
      "step": 57000
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.3492063492063492e-05,
      "loss": 0.0148,
      "step": 57500
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.3174603174603176e-05,
      "loss": 0.0153,
      "step": 58000
    },
    {
      "epoch": 2.23,
      "learning_rate": 1.2857142857142857e-05,
      "loss": 0.0154,
      "step": 58500
    },
    {
      "epoch": 2.25,
      "learning_rate": 1.253968253968254e-05,
      "loss": 0.016,
      "step": 59000
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.2222222222222222e-05,
      "loss": 0.0146,
      "step": 59500
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.1904761904761905e-05,
      "loss": 0.0162,
      "step": 60000
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.1587301587301588e-05,
      "loss": 0.0156,
      "step": 60500
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.126984126984127e-05,
      "loss": 0.0151,
      "step": 61000
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.0952380952380953e-05,
      "loss": 0.0138,
      "step": 61500
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.0634920634920636e-05,
      "loss": 0.0144,
      "step": 62000
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.0317460317460318e-05,
      "loss": 0.0151,
      "step": 62500
    },
    {
      "epoch": 2.4,
      "learning_rate": 1e-05,
      "loss": 0.0163,
      "step": 63000
    },
    {
      "epoch": 2.42,
      "learning_rate": 9.682539682539683e-06,
      "loss": 0.0131,
      "step": 63500
    },
    {
      "epoch": 2.44,
      "learning_rate": 9.365079365079366e-06,
      "loss": 0.015,
      "step": 64000
    },
    {
      "epoch": 2.46,
      "learning_rate": 9.047619047619047e-06,
      "loss": 0.0142,
      "step": 64500
    },
    {
      "epoch": 2.48,
      "learning_rate": 8.73015873015873e-06,
      "loss": 0.0145,
      "step": 65000
    },
    {
      "epoch": 2.5,
      "learning_rate": 8.412698412698414e-06,
      "loss": 0.0147,
      "step": 65500
    },
    {
      "epoch": 2.51,
      "learning_rate": 8.095238095238097e-06,
      "loss": 0.0163,
      "step": 66000
    },
    {
      "epoch": 2.53,
      "learning_rate": 7.777777777777777e-06,
      "loss": 0.0138,
      "step": 66500
    },
    {
      "epoch": 2.55,
      "learning_rate": 7.460317460317461e-06,
      "loss": 0.0145,
      "step": 67000
    },
    {
      "epoch": 2.57,
      "learning_rate": 7.142857142857143e-06,
      "loss": 0.0135,
      "step": 67500
    },
    {
      "epoch": 2.59,
      "learning_rate": 6.825396825396825e-06,
      "loss": 0.0141,
      "step": 68000
    },
    {
      "epoch": 2.61,
      "learning_rate": 6.507936507936509e-06,
      "loss": 0.013,
      "step": 68500
    },
    {
      "epoch": 2.63,
      "learning_rate": 6.190476190476191e-06,
      "loss": 0.0125,
      "step": 69000
    },
    {
      "epoch": 2.65,
      "learning_rate": 5.873015873015873e-06,
      "loss": 0.0116,
      "step": 69500
    },
    {
      "epoch": 2.67,
      "learning_rate": 5.555555555555556e-06,
      "loss": 0.0165,
      "step": 70000
    },
    {
      "epoch": 2.69,
      "learning_rate": 5.2380952380952384e-06,
      "loss": 0.0135,
      "step": 70500
    },
    {
      "epoch": 2.7,
      "learning_rate": 4.920634920634921e-06,
      "loss": 0.012,
      "step": 71000
    },
    {
      "epoch": 2.72,
      "learning_rate": 4.603174603174604e-06,
      "loss": 0.0115,
      "step": 71500
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.285714285714286e-06,
      "loss": 0.0137,
      "step": 72000
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.968253968253968e-06,
      "loss": 0.0141,
      "step": 72500
    },
    {
      "epoch": 2.78,
      "learning_rate": 3.650793650793651e-06,
      "loss": 0.0112,
      "step": 73000
    },
    {
      "epoch": 2.8,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 0.014,
      "step": 73500
    },
    {
      "epoch": 2.82,
      "learning_rate": 3.015873015873016e-06,
      "loss": 0.0123,
      "step": 74000
    },
    {
      "epoch": 2.84,
      "learning_rate": 2.6984126984126986e-06,
      "loss": 0.0151,
      "step": 74500
    },
    {
      "epoch": 2.86,
      "learning_rate": 2.3809523809523808e-06,
      "loss": 0.0141,
      "step": 75000
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.0634920634920634e-06,
      "loss": 0.0134,
      "step": 75500
    },
    {
      "epoch": 2.9,
      "learning_rate": 1.7460317460317462e-06,
      "loss": 0.0124,
      "step": 76000
    },
    {
      "epoch": 2.91,
      "learning_rate": 1.4285714285714286e-06,
      "loss": 0.011,
      "step": 76500
    },
    {
      "epoch": 2.93,
      "learning_rate": 1.1111111111111112e-06,
      "loss": 0.0113,
      "step": 77000
    },
    {
      "epoch": 2.95,
      "learning_rate": 7.936507936507937e-07,
      "loss": 0.0123,
      "step": 77500
    },
    {
      "epoch": 2.97,
      "learning_rate": 4.761904761904763e-07,
      "loss": 0.0119,
      "step": 78000
    },
    {
      "epoch": 2.99,
      "learning_rate": 1.5873015873015874e-07,
      "loss": 0.011,
      "step": 78500
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.9963097222222222,
      "eval_f1-score": 0.9963028995135582,
      "eval_loss": 0.013718551024794579,
      "eval_runtime": 2760.1832,
      "eval_samples_per_second": 260.852,
      "eval_steps_per_second": 2.038,
      "step": 78750
    }
  ],
  "max_steps": 78750,
  "num_train_epochs": 3,
  "total_flos": 6.6789762859008e+17,
  "trial_name": null,
  "trial_params": null
}