{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.9465648854961832,
  "eval_steps": 500,
  "global_step": 25500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 2.5e-06,
      "loss": 31.9742,
      "step": 50
    },
    {
      "epoch": 0.01,
      "learning_rate": 5e-06,
      "loss": 32.5191,
      "step": 100
    },
    {
      "epoch": 0.01,
      "learning_rate": 7.5e-06,
      "loss": 31.2118,
      "step": 150
    },
    {
      "epoch": 0.02,
      "learning_rate": 1e-05,
      "loss": 28.9569,
      "step": 200
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.25e-05,
      "loss": 27.1361,
      "step": 250
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.5e-05,
      "loss": 24.7898,
      "step": 300
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.75e-05,
      "loss": 22.545,
      "step": 350
    },
    {
      "epoch": 0.03,
      "learning_rate": 2e-05,
      "loss": 21.4837,
      "step": 400
    },
    {
      "epoch": 0.03,
      "learning_rate": 2.25e-05,
      "loss": 18.8854,
      "step": 450
    },
    {
      "epoch": 0.04,
      "learning_rate": 2.5e-05,
      "loss": 16.722,
      "step": 500
    },
    {
      "epoch": 0.04,
      "learning_rate": 2.7500000000000004e-05,
      "loss": 14.4161,
      "step": 550
    },
    {
      "epoch": 0.05,
      "learning_rate": 3e-05,
      "loss": 11.8861,
      "step": 600
    },
    {
      "epoch": 0.05,
      "learning_rate": 3.2500000000000004e-05,
      "loss": 9.5285,
      "step": 650
    },
    {
      "epoch": 0.05,
      "learning_rate": 3.5e-05,
      "loss": 7.503,
      "step": 700
    },
    {
      "epoch": 0.06,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 6.355,
      "step": 750
    },
    {
      "epoch": 0.06,
      "learning_rate": 4e-05,
      "loss": 5.2049,
      "step": 800
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.25e-05,
      "loss": 4.4318,
      "step": 850
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.5e-05,
      "loss": 3.852,
      "step": 900
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.75e-05,
      "loss": 3.4899,
      "step": 950
    },
    {
      "epoch": 0.08,
      "learning_rate": 5e-05,
      "loss": 3.3489,
      "step": 1000
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.990079365079365e-05,
      "loss": 3.1161,
      "step": 1050
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9801587301587306e-05,
      "loss": 2.9771,
      "step": 1100
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9702380952380955e-05,
      "loss": 2.9968,
      "step": 1150
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.960317460317461e-05,
      "loss": 2.9209,
      "step": 1200
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.950396825396826e-05,
      "loss": 2.8767,
      "step": 1250
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.940476190476191e-05,
      "loss": 2.8769,
      "step": 1300
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.930555555555556e-05,
      "loss": 2.8317,
      "step": 1350
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.9206349206349204e-05,
      "loss": 2.8225,
      "step": 1400
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.910714285714286e-05,
      "loss": 2.6981,
      "step": 1450
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.900793650793651e-05,
      "loss": 2.7759,
      "step": 1500
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8908730158730156e-05,
      "loss": 2.631,
      "step": 1550
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.880952380952381e-05,
      "loss": 2.693,
      "step": 1600
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.871031746031746e-05,
      "loss": 2.6437,
      "step": 1650
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8611111111111115e-05,
      "loss": 2.5367,
      "step": 1700
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8511904761904764e-05,
      "loss": 2.6009,
      "step": 1750
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.841269841269841e-05,
      "loss": 2.6043,
      "step": 1800
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.831349206349207e-05,
      "loss": 2.4894,
      "step": 1850
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.8214285714285716e-05,
      "loss": 2.5902,
      "step": 1900
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.811507936507937e-05,
      "loss": 2.5836,
      "step": 1950
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.801587301587302e-05,
      "loss": 2.604,
      "step": 2000
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.791666666666667e-05,
      "loss": 2.4979,
      "step": 2050
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.781746031746032e-05,
      "loss": 2.6036,
      "step": 2100
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.771825396825397e-05,
      "loss": 2.4106,
      "step": 2150
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.761904761904762e-05,
      "loss": 2.4927,
      "step": 2200
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.751984126984127e-05,
      "loss": 2.6297,
      "step": 2250
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.7420634920634924e-05,
      "loss": 2.4696,
      "step": 2300
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.732142857142857e-05,
      "loss": 2.5137,
      "step": 2350
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.722222222222222e-05,
      "loss": 2.4031,
      "step": 2400
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.7123015873015876e-05,
      "loss": 2.5041,
      "step": 2450
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.7023809523809525e-05,
      "loss": 2.4883,
      "step": 2500
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.692460317460317e-05,
      "loss": 2.4844,
      "step": 2550
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.682539682539683e-05,
      "loss": 2.4495,
      "step": 2600
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.672619047619048e-05,
      "loss": 2.4946,
      "step": 2650
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.662698412698413e-05,
      "loss": 2.5235,
      "step": 2700
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.652777777777778e-05,
      "loss": 2.443,
      "step": 2750
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.642857142857143e-05,
      "loss": 2.5334,
      "step": 2800
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.6329365079365085e-05,
      "loss": 2.4367,
      "step": 2850
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.623015873015873e-05,
      "loss": 2.5123,
      "step": 2900
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.613095238095239e-05,
      "loss": 2.3548,
      "step": 2950
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.603174603174603e-05,
      "loss": 2.4387,
      "step": 3000
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.5932539682539685e-05,
      "loss": 2.3441,
      "step": 3050
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.5833333333333334e-05,
      "loss": 2.3135,
      "step": 3100
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.573412698412698e-05,
      "loss": 2.3535,
      "step": 3150
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.563492063492064e-05,
      "loss": 2.3418,
      "step": 3200
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.5535714285714286e-05,
      "loss": 2.466,
      "step": 3250
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.543650793650794e-05,
      "loss": 2.3942,
      "step": 3300
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.533730158730159e-05,
      "loss": 2.3981,
      "step": 3350
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.523809523809524e-05,
      "loss": 2.4964,
      "step": 3400
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.5138888888888894e-05,
      "loss": 2.4041,
      "step": 3450
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.503968253968254e-05,
      "loss": 2.3344,
      "step": 3500
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.494047619047619e-05,
      "loss": 2.4342,
      "step": 3550
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.4841269841269846e-05,
      "loss": 2.3132,
      "step": 3600
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.4742063492063494e-05,
      "loss": 2.3755,
      "step": 3650
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.464285714285715e-05,
      "loss": 2.3009,
      "step": 3700
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.45436507936508e-05,
      "loss": 2.3669,
      "step": 3750
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.4444444444444447e-05,
      "loss": 2.2948,
      "step": 3800
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.4345238095238095e-05,
      "loss": 2.3885,
      "step": 3850
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.4246031746031744e-05,
      "loss": 2.3917,
      "step": 3900
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.41468253968254e-05,
      "loss": 2.3107,
      "step": 3950
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.404761904761905e-05,
      "loss": 2.2456,
      "step": 4000
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.39484126984127e-05,
      "loss": 2.4836,
      "step": 4050
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.384920634920635e-05,
      "loss": 2.348,
      "step": 4100
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.375e-05,
      "loss": 2.2756,
      "step": 4150
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.3650793650793655e-05,
      "loss": 2.3311,
      "step": 4200
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.35515873015873e-05,
      "loss": 2.3488,
      "step": 4250
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.345238095238096e-05,
      "loss": 2.3493,
      "step": 4300
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.335317460317461e-05,
      "loss": 2.374,
      "step": 4350
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.3253968253968256e-05,
      "loss": 2.3459,
      "step": 4400
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.315476190476191e-05,
      "loss": 2.3194,
      "step": 4450
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.305555555555556e-05,
      "loss": 2.3001,
      "step": 4500
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.295634920634921e-05,
      "loss": 2.3023,
      "step": 4550
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.2857142857142856e-05,
      "loss": 2.3394,
      "step": 4600
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.2757936507936505e-05,
      "loss": 2.3859,
      "step": 4650
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.265873015873016e-05,
      "loss": 2.2853,
      "step": 4700
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.255952380952381e-05,
      "loss": 2.3349,
      "step": 4750
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.2460317460317464e-05,
      "loss": 2.2621,
      "step": 4800
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.236111111111111e-05,
      "loss": 2.3776,
      "step": 4850
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.226190476190476e-05,
      "loss": 2.2746,
      "step": 4900
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.2162698412698416e-05,
      "loss": 2.3579,
      "step": 4950
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.2063492063492065e-05,
      "loss": 2.3256,
      "step": 5000
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.196428571428572e-05,
      "loss": 2.3321,
      "step": 5050
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.186507936507937e-05,
      "loss": 2.2774,
      "step": 5100
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.176587301587302e-05,
      "loss": 2.1933,
      "step": 5150
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.166666666666667e-05,
      "loss": 2.3089,
      "step": 5200
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.156746031746032e-05,
      "loss": 2.3038,
      "step": 5250
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.1468253968253976e-05,
      "loss": 2.327,
      "step": 5300
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.136904761904762e-05,
      "loss": 2.2387,
      "step": 5350
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.126984126984127e-05,
      "loss": 2.2992,
      "step": 5400
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.117063492063492e-05,
      "loss": 2.3167,
      "step": 5450
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.107142857142857e-05,
      "loss": 2.2835,
      "step": 5500
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.0972222222222225e-05,
      "loss": 2.2788,
      "step": 5550
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.0873015873015874e-05,
      "loss": 2.2814,
      "step": 5600
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.077380952380952e-05,
      "loss": 2.2687,
      "step": 5650
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.067460317460318e-05,
      "loss": 2.2743,
      "step": 5700
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.0575396825396826e-05,
      "loss": 2.3038,
      "step": 5750
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.047619047619048e-05,
      "loss": 2.2646,
      "step": 5800
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.037698412698413e-05,
      "loss": 2.2602,
      "step": 5850
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.027777777777778e-05,
      "loss": 2.3499,
      "step": 5900
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.017857142857143e-05,
      "loss": 2.2881,
      "step": 5950
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.007936507936508e-05,
      "loss": 2.277,
      "step": 6000
    },
    {
      "epoch": 0.46,
      "learning_rate": 3.998015873015874e-05,
      "loss": 2.1779,
      "step": 6050
    },
    {
      "epoch": 0.47,
      "learning_rate": 3.9880952380952386e-05,
      "loss": 2.1701,
      "step": 6100
    },
    {
      "epoch": 0.47,
      "learning_rate": 3.9781746031746034e-05,
      "loss": 2.2704,
      "step": 6150
    },
    {
      "epoch": 0.47,
      "learning_rate": 3.968253968253968e-05,
      "loss": 2.2207,
      "step": 6200
    },
    {
      "epoch": 0.48,
      "learning_rate": 3.958333333333333e-05,
      "loss": 2.2198,
      "step": 6250
    },
    {
      "epoch": 0.48,
      "learning_rate": 3.9484126984126986e-05,
      "loss": 2.223,
      "step": 6300
    },
    {
      "epoch": 0.48,
      "learning_rate": 3.9384920634920635e-05,
      "loss": 2.3311,
      "step": 6350
    },
    {
      "epoch": 0.49,
      "learning_rate": 3.928571428571429e-05,
      "loss": 2.1598,
      "step": 6400
    },
    {
      "epoch": 0.49,
      "learning_rate": 3.918650793650794e-05,
      "loss": 2.2098,
      "step": 6450
    },
    {
      "epoch": 0.5,
      "learning_rate": 3.908730158730159e-05,
      "loss": 2.2645,
      "step": 6500
    },
    {
      "epoch": 0.5,
      "learning_rate": 3.898809523809524e-05,
      "loss": 2.2462,
      "step": 6550
    },
    {
      "epoch": 0.5,
      "learning_rate": 3.888888888888889e-05,
      "loss": 2.3286,
      "step": 6600
    },
    {
      "epoch": 0.51,
      "learning_rate": 3.878968253968254e-05,
      "loss": 2.2331,
      "step": 6650
    },
    {
      "epoch": 0.51,
      "learning_rate": 3.8690476190476195e-05,
      "loss": 2.2508,
      "step": 6700
    },
    {
      "epoch": 0.52,
      "learning_rate": 3.859126984126984e-05,
      "loss": 2.2066,
      "step": 6750
    },
    {
      "epoch": 0.52,
      "learning_rate": 3.84920634920635e-05,
      "loss": 2.2877,
      "step": 6800
    },
    {
      "epoch": 0.52,
      "learning_rate": 3.839285714285715e-05,
      "loss": 2.2437,
      "step": 6850
    },
    {
      "epoch": 0.53,
      "learning_rate": 3.8293650793650795e-05,
      "loss": 2.2342,
      "step": 6900
    },
    {
      "epoch": 0.53,
      "learning_rate": 3.8194444444444444e-05,
      "loss": 2.3143,
      "step": 6950
    },
    {
      "epoch": 0.53,
      "learning_rate": 3.809523809523809e-05,
      "loss": 2.2252,
      "step": 7000
    },
    {
      "epoch": 0.54,
      "learning_rate": 3.799603174603175e-05,
      "loss": 2.2422,
      "step": 7050
    },
    {
      "epoch": 0.54,
      "learning_rate": 3.7896825396825396e-05,
      "loss": 2.1808,
      "step": 7100
    },
    {
      "epoch": 0.55,
      "learning_rate": 3.779761904761905e-05,
      "loss": 2.1898,
      "step": 7150
    },
    {
      "epoch": 0.55,
      "learning_rate": 3.76984126984127e-05,
      "loss": 2.1189,
      "step": 7200
    },
    {
      "epoch": 0.55,
      "learning_rate": 3.759920634920635e-05,
      "loss": 2.2667,
      "step": 7250
    },
    {
      "epoch": 0.56,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 2.1627,
      "step": 7300
    },
    {
      "epoch": 0.56,
      "learning_rate": 3.740079365079365e-05,
      "loss": 2.2402,
      "step": 7350
    },
    {
      "epoch": 0.56,
      "learning_rate": 3.730158730158731e-05,
      "loss": 2.203,
      "step": 7400
    },
    {
      "epoch": 0.57,
      "learning_rate": 3.7202380952380956e-05,
      "loss": 2.2202,
      "step": 7450
    },
    {
      "epoch": 0.57,
      "learning_rate": 3.7103174603174604e-05,
      "loss": 2.2406,
      "step": 7500
    },
    {
      "epoch": 0.58,
      "learning_rate": 3.700396825396826e-05,
      "loss": 2.2132,
      "step": 7550
    },
    {
      "epoch": 0.58,
      "learning_rate": 3.690476190476191e-05,
      "loss": 2.1831,
      "step": 7600
    },
    {
      "epoch": 0.58,
      "learning_rate": 3.6805555555555556e-05,
      "loss": 2.1939,
      "step": 7650
    },
    {
      "epoch": 0.59,
      "learning_rate": 3.6706349206349205e-05,
      "loss": 2.1544,
      "step": 7700
    },
    {
      "epoch": 0.59,
      "learning_rate": 3.6607142857142853e-05,
      "loss": 2.2235,
      "step": 7750
    },
    {
      "epoch": 0.6,
      "learning_rate": 3.650793650793651e-05,
      "loss": 2.2568,
      "step": 7800
    },
    {
      "epoch": 0.6,
      "learning_rate": 3.640873015873016e-05,
      "loss": 2.2168,
      "step": 7850
    },
    {
      "epoch": 0.6,
      "learning_rate": 3.630952380952381e-05,
      "loss": 2.1741,
      "step": 7900
    },
    {
      "epoch": 0.61,
      "learning_rate": 3.621031746031746e-05,
      "loss": 2.2061,
      "step": 7950
    },
    {
      "epoch": 0.61,
      "learning_rate": 3.611111111111111e-05,
      "loss": 2.2602,
      "step": 8000
    },
    {
      "epoch": 0.61,
      "learning_rate": 3.6011904761904765e-05,
      "loss": 2.2055,
      "step": 8050
    },
    {
      "epoch": 0.62,
      "learning_rate": 3.591269841269841e-05,
      "loss": 2.2411,
      "step": 8100
    },
    {
      "epoch": 0.62,
      "learning_rate": 3.581349206349207e-05,
      "loss": 2.2783,
      "step": 8150
    },
    {
      "epoch": 0.63,
      "learning_rate": 3.571428571428572e-05,
      "loss": 2.1591,
      "step": 8200
    },
    {
      "epoch": 0.63,
      "learning_rate": 3.5615079365079365e-05,
      "loss": 2.1479,
      "step": 8250
    },
    {
      "epoch": 0.63,
      "learning_rate": 3.551587301587302e-05,
      "loss": 2.1781,
      "step": 8300
    },
    {
      "epoch": 0.64,
      "learning_rate": 3.541666666666667e-05,
      "loss": 2.1612,
      "step": 8350
    },
    {
      "epoch": 0.64,
      "learning_rate": 3.5317460317460324e-05,
      "loss": 2.1134,
      "step": 8400
    },
    {
      "epoch": 0.65,
      "learning_rate": 3.521825396825397e-05,
      "loss": 2.1386,
      "step": 8450
    },
    {
      "epoch": 0.65,
      "learning_rate": 3.511904761904762e-05,
      "loss": 2.0929,
      "step": 8500
    },
    {
      "epoch": 0.65,
      "learning_rate": 3.501984126984127e-05,
      "loss": 2.2324,
      "step": 8550
    },
    {
      "epoch": 0.66,
      "learning_rate": 3.492063492063492e-05,
      "loss": 2.1589,
      "step": 8600
    },
    {
      "epoch": 0.66,
      "learning_rate": 3.4821428571428574e-05,
      "loss": 2.1681,
      "step": 8650
    },
    {
      "epoch": 0.66,
      "learning_rate": 3.472222222222222e-05,
      "loss": 2.1765,
      "step": 8700
    },
    {
      "epoch": 0.67,
      "learning_rate": 3.462301587301587e-05,
      "loss": 2.1753,
      "step": 8750
    },
    {
      "epoch": 0.67,
      "learning_rate": 3.4523809523809526e-05,
      "loss": 2.249,
      "step": 8800
    },
    {
      "epoch": 0.68,
      "learning_rate": 3.4424603174603174e-05,
      "loss": 2.1518,
      "step": 8850
    },
    {
      "epoch": 0.68,
      "learning_rate": 3.432539682539683e-05,
      "loss": 2.2355,
      "step": 8900
    },
    {
      "epoch": 0.68,
      "learning_rate": 3.422619047619048e-05,
      "loss": 2.1606,
      "step": 8950
    },
    {
      "epoch": 0.69,
      "learning_rate": 3.412698412698413e-05,
      "loss": 2.2142,
      "step": 9000
    },
    {
      "epoch": 0.69,
      "learning_rate": 3.402777777777778e-05,
      "loss": 2.1308,
      "step": 9050
    },
    {
      "epoch": 0.69,
      "learning_rate": 3.392857142857143e-05,
      "loss": 2.1826,
      "step": 9100
    },
    {
      "epoch": 0.7,
      "learning_rate": 3.3829365079365086e-05,
      "loss": 2.2073,
      "step": 9150
    },
    {
      "epoch": 0.7,
      "learning_rate": 3.3730158730158734e-05,
      "loss": 2.1768,
      "step": 9200
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.363095238095238e-05,
      "loss": 2.2085,
      "step": 9250
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.353174603174603e-05,
      "loss": 2.1844,
      "step": 9300
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.343253968253968e-05,
      "loss": 2.1136,
      "step": 9350
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 2.1827,
      "step": 9400
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.3234126984126983e-05,
      "loss": 2.084,
      "step": 9450
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.313492063492064e-05,
      "loss": 2.1826,
      "step": 9500
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.303571428571429e-05,
      "loss": 2.0812,
      "step": 9550
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.2936507936507936e-05,
      "loss": 2.3111,
      "step": 9600
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.283730158730159e-05,
      "loss": 2.2154,
      "step": 9650
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.273809523809524e-05,
      "loss": 2.1717,
      "step": 9700
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.263888888888889e-05,
      "loss": 2.1668,
      "step": 9750
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.253968253968254e-05,
      "loss": 2.1209,
      "step": 9800
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.244047619047619e-05,
      "loss": 2.0419,
      "step": 9850
    },
    {
      "epoch": 0.76,
      "learning_rate": 3.234126984126985e-05,
      "loss": 2.2324,
      "step": 9900
    },
    {
      "epoch": 0.76,
      "learning_rate": 3.2242063492063495e-05,
      "loss": 2.2163,
      "step": 9950
    },
    {
      "epoch": 0.76,
      "learning_rate": 3.2142857142857144e-05,
      "loss": 2.2406,
      "step": 10000
    },
    {
      "epoch": 0.77,
      "learning_rate": 3.20436507936508e-05,
      "loss": 2.1455,
      "step": 10050
    },
    {
      "epoch": 0.77,
      "learning_rate": 3.194444444444444e-05,
      "loss": 2.0851,
      "step": 10100
    },
    {
      "epoch": 0.77,
      "learning_rate": 3.1845238095238096e-05,
      "loss": 2.1666,
      "step": 10150
    },
    {
      "epoch": 0.78,
      "learning_rate": 3.1746031746031745e-05,
      "loss": 2.2034,
      "step": 10200
    },
    {
      "epoch": 0.78,
      "learning_rate": 3.16468253968254e-05,
      "loss": 2.0919,
      "step": 10250
    },
    {
      "epoch": 0.79,
      "learning_rate": 3.154761904761905e-05,
      "loss": 2.0807,
      "step": 10300
    },
    {
      "epoch": 0.79,
      "learning_rate": 3.14484126984127e-05,
      "loss": 2.1648,
      "step": 10350
    },
    {
      "epoch": 0.79,
      "learning_rate": 3.134920634920635e-05,
      "loss": 2.1216,
      "step": 10400
    },
    {
      "epoch": 0.8,
      "learning_rate": 3.125e-05,
      "loss": 2.1722,
      "step": 10450
    },
    {
      "epoch": 0.8,
      "learning_rate": 3.1150793650793656e-05,
      "loss": 2.2017,
      "step": 10500
    },
    {
      "epoch": 0.81,
      "learning_rate": 3.1051587301587304e-05,
      "loss": 2.076,
      "step": 10550
    },
    {
      "epoch": 0.81,
      "learning_rate": 3.095238095238095e-05,
      "loss": 2.0688,
      "step": 10600
    },
    {
      "epoch": 0.81,
      "learning_rate": 3.085317460317461e-05,
      "loss": 2.0956,
      "step": 10650
    },
    {
      "epoch": 0.82,
      "learning_rate": 3.075396825396826e-05,
      "loss": 2.1461,
      "step": 10700
    },
    {
      "epoch": 0.82,
      "learning_rate": 3.0654761904761905e-05,
      "loss": 2.1046,
      "step": 10750
    },
    {
      "epoch": 0.82,
      "learning_rate": 3.055555555555556e-05,
      "loss": 2.163,
      "step": 10800
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.0456349206349206e-05,
      "loss": 2.1368,
      "step": 10850
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.0357142857142857e-05,
      "loss": 2.1615,
      "step": 10900
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.0257936507936506e-05,
      "loss": 2.1717,
      "step": 10950
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.0158730158730158e-05,
      "loss": 2.0984,
      "step": 11000
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.005952380952381e-05,
      "loss": 2.1134,
      "step": 11050
    },
    {
      "epoch": 0.85,
      "learning_rate": 2.996031746031746e-05,
      "loss": 2.1184,
      "step": 11100
    },
    {
      "epoch": 0.85,
      "learning_rate": 2.9861111111111113e-05,
      "loss": 2.1654,
      "step": 11150
    },
    {
      "epoch": 0.85,
      "learning_rate": 2.9761904761904762e-05,
      "loss": 2.1137,
      "step": 11200
    },
    {
      "epoch": 0.86,
      "learning_rate": 2.9662698412698414e-05,
      "loss": 2.1224,
      "step": 11250
    },
    {
      "epoch": 0.86,
      "learning_rate": 2.9563492063492066e-05,
      "loss": 2.1394,
      "step": 11300
    },
    {
      "epoch": 0.87,
      "learning_rate": 2.9464285714285718e-05,
      "loss": 2.1456,
      "step": 11350
    },
    {
      "epoch": 0.87,
      "learning_rate": 2.9365079365079366e-05,
      "loss": 2.0706,
      "step": 11400
    },
    {
      "epoch": 0.87,
      "learning_rate": 2.9265873015873018e-05,
      "loss": 2.1334,
      "step": 11450
    },
    {
      "epoch": 0.88,
      "learning_rate": 2.916666666666667e-05,
      "loss": 2.0187,
      "step": 11500
    },
    {
      "epoch": 0.88,
      "learning_rate": 2.906746031746032e-05,
      "loss": 2.1122,
      "step": 11550
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.8968253968253974e-05,
      "loss": 2.2484,
      "step": 11600
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.886904761904762e-05,
      "loss": 2.1894,
      "step": 11650
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.876984126984127e-05,
      "loss": 2.2025,
      "step": 11700
    },
    {
      "epoch": 0.9,
      "learning_rate": 2.867063492063492e-05,
      "loss": 2.183,
      "step": 11750
    },
    {
      "epoch": 0.9,
      "learning_rate": 2.857142857142857e-05,
      "loss": 2.0762,
      "step": 11800
    },
    {
      "epoch": 0.9,
      "learning_rate": 2.8472222222222223e-05,
      "loss": 2.1949,
      "step": 11850
    },
    {
      "epoch": 0.91,
      "learning_rate": 2.8373015873015875e-05,
      "loss": 2.0965,
      "step": 11900
    },
    {
      "epoch": 0.91,
      "learning_rate": 2.8273809523809523e-05,
      "loss": 2.0698,
      "step": 11950
    },
    {
      "epoch": 0.92,
      "learning_rate": 2.8174603174603175e-05,
      "loss": 2.1163,
      "step": 12000
    },
    {
      "epoch": 0.92,
      "learning_rate": 2.8075396825396827e-05,
      "loss": 2.1259,
      "step": 12050
    },
    {
      "epoch": 0.92,
      "learning_rate": 2.797619047619048e-05,
      "loss": 2.1668,
      "step": 12100
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.787698412698413e-05,
      "loss": 2.1295,
      "step": 12150
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.777777777777778e-05,
      "loss": 2.1924,
      "step": 12200
    },
    {
      "epoch": 0.94,
      "learning_rate": 2.767857142857143e-05,
      "loss": 2.2458,
      "step": 12250
    },
    {
      "epoch": 0.94,
      "learning_rate": 2.7579365079365083e-05,
      "loss": 2.0815,
      "step": 12300
    },
    {
      "epoch": 0.94,
      "learning_rate": 2.7480158730158735e-05,
      "loss": 2.0327,
      "step": 12350
    },
    {
      "epoch": 0.95,
      "learning_rate": 2.7380952380952383e-05,
      "loss": 2.1338,
      "step": 12400
    },
    {
      "epoch": 0.95,
      "learning_rate": 2.7281746031746032e-05,
      "loss": 2.1107,
      "step": 12450
    },
    {
      "epoch": 0.95,
      "learning_rate": 2.718253968253968e-05,
      "loss": 2.0855,
      "step": 12500
    },
    {
      "epoch": 0.96,
      "learning_rate": 2.7083333333333332e-05,
      "loss": 2.1685,
      "step": 12550
    },
    {
      "epoch": 0.96,
      "learning_rate": 2.6984126984126984e-05,
      "loss": 2.1238,
      "step": 12600
    },
    {
      "epoch": 0.97,
      "learning_rate": 2.6884920634920636e-05,
      "loss": 2.1076,
      "step": 12650
    },
    {
      "epoch": 0.97,
      "learning_rate": 2.6785714285714288e-05,
      "loss": 2.1635,
      "step": 12700
    },
    {
      "epoch": 0.97,
      "learning_rate": 2.6686507936507936e-05,
      "loss": 2.1277,
      "step": 12750
    },
    {
      "epoch": 0.98,
      "learning_rate": 2.6587301587301588e-05,
      "loss": 2.1048,
      "step": 12800
    },
    {
      "epoch": 0.98,
      "learning_rate": 2.648809523809524e-05,
      "loss": 2.1246,
      "step": 12850
    },
    {
      "epoch": 0.98,
      "learning_rate": 2.6388888888888892e-05,
      "loss": 2.1518,
      "step": 12900
    },
    {
      "epoch": 0.99,
      "learning_rate": 2.628968253968254e-05,
      "loss": 2.0936,
      "step": 12950
    },
    {
      "epoch": 0.99,
      "learning_rate": 2.6190476190476192e-05,
      "loss": 2.096,
      "step": 13000
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.6091269841269844e-05,
      "loss": 2.1602,
      "step": 13050
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.5992063492063496e-05,
      "loss": 2.1038,
      "step": 13100
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.5892857142857148e-05,
      "loss": 2.1308,
      "step": 13150
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.5793650793650796e-05,
      "loss": 2.1161,
      "step": 13200
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.5694444444444445e-05,
      "loss": 2.0994,
      "step": 13250
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.5595238095238093e-05,
      "loss": 2.1299,
      "step": 13300
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.5496031746031745e-05,
      "loss": 2.0676,
      "step": 13350
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.5396825396825397e-05,
      "loss": 2.0956,
      "step": 13400
    },
    {
      "epoch": 1.03,
      "learning_rate": 2.529761904761905e-05,
      "loss": 2.1986,
      "step": 13450
    },
    {
      "epoch": 1.03,
      "learning_rate": 2.5198412698412697e-05,
      "loss": 2.0667,
      "step": 13500
    },
    {
      "epoch": 1.03,
      "learning_rate": 2.509920634920635e-05,
      "loss": 1.989,
      "step": 13550
    },
    {
      "epoch": 1.04,
      "learning_rate": 2.5e-05,
      "loss": 2.0445,
      "step": 13600
    },
    {
      "epoch": 1.04,
      "learning_rate": 2.4900793650793653e-05,
      "loss": 2.0726,
      "step": 13650
    },
    {
      "epoch": 1.05,
      "learning_rate": 2.4801587301587305e-05,
      "loss": 2.1361,
      "step": 13700
    },
    {
      "epoch": 1.05,
      "learning_rate": 2.4702380952380953e-05,
      "loss": 2.0984,
      "step": 13750
    },
    {
      "epoch": 1.05,
      "learning_rate": 2.4603174603174602e-05,
      "loss": 2.1363,
      "step": 13800
    },
    {
      "epoch": 1.06,
      "learning_rate": 2.4503968253968254e-05,
      "loss": 2.0652,
      "step": 13850
    },
    {
      "epoch": 1.06,
      "learning_rate": 2.4404761904761906e-05,
      "loss": 2.087,
      "step": 13900
    },
    {
      "epoch": 1.06,
      "learning_rate": 2.4305555555555558e-05,
      "loss": 2.1069,
      "step": 13950
    },
    {
      "epoch": 1.07,
      "learning_rate": 2.4206349206349206e-05,
      "loss": 2.1175,
      "step": 14000
    },
    {
      "epoch": 1.07,
      "learning_rate": 2.4107142857142858e-05,
      "loss": 2.071,
      "step": 14050
    },
    {
      "epoch": 1.08,
      "learning_rate": 2.400793650793651e-05,
      "loss": 2.0809,
      "step": 14100
    },
    {
      "epoch": 1.08,
      "learning_rate": 2.390873015873016e-05,
      "loss": 2.0943,
      "step": 14150
    },
    {
      "epoch": 1.08,
      "learning_rate": 2.380952380952381e-05,
      "loss": 2.1503,
      "step": 14200
    },
    {
      "epoch": 1.09,
      "learning_rate": 2.3710317460317462e-05,
      "loss": 2.1241,
      "step": 14250
    },
    {
      "epoch": 1.09,
      "learning_rate": 2.361111111111111e-05,
      "loss": 2.1259,
      "step": 14300
    },
    {
      "epoch": 1.1,
      "learning_rate": 2.3511904761904762e-05,
      "loss": 2.0531,
      "step": 14350
    },
    {
      "epoch": 1.1,
      "learning_rate": 2.3412698412698414e-05,
      "loss": 2.2189,
      "step": 14400
    },
    {
      "epoch": 1.1,
      "learning_rate": 2.3313492063492066e-05,
      "loss": 2.0976,
      "step": 14450
    },
    {
      "epoch": 1.11,
      "learning_rate": 2.3214285714285715e-05,
      "loss": 2.1033,
      "step": 14500
    },
    {
      "epoch": 1.11,
      "learning_rate": 2.3115079365079367e-05,
      "loss": 2.1572,
      "step": 14550
    },
    {
      "epoch": 1.11,
      "learning_rate": 2.3015873015873015e-05,
      "loss": 2.0534,
      "step": 14600
    },
    {
      "epoch": 1.12,
      "learning_rate": 2.2916666666666667e-05,
      "loss": 2.1607,
      "step": 14650
    },
    {
      "epoch": 1.12,
      "learning_rate": 2.281746031746032e-05,
      "loss": 1.9848,
      "step": 14700
    },
    {
      "epoch": 1.13,
      "learning_rate": 2.271825396825397e-05,
      "loss": 2.0368,
      "step": 14750
    },
    {
      "epoch": 1.13,
      "learning_rate": 2.261904761904762e-05,
      "loss": 2.1026,
      "step": 14800
    },
    {
      "epoch": 1.13,
      "learning_rate": 2.251984126984127e-05,
      "loss": 2.1793,
      "step": 14850
    },
    {
      "epoch": 1.14,
      "learning_rate": 2.2420634920634923e-05,
      "loss": 2.0969,
      "step": 14900
    },
    {
      "epoch": 1.14,
      "learning_rate": 2.2321428571428575e-05,
      "loss": 2.0451,
      "step": 14950
    },
    {
      "epoch": 1.15,
      "learning_rate": 2.2222222222222223e-05,
      "loss": 1.9796,
      "step": 15000
    },
    {
      "epoch": 1.15,
      "learning_rate": 2.2123015873015872e-05,
      "loss": 2.0302,
      "step": 15050
    },
    {
      "epoch": 1.15,
      "learning_rate": 2.2023809523809524e-05,
      "loss": 2.06,
      "step": 15100
    },
    {
      "epoch": 1.16,
      "learning_rate": 2.1924603174603176e-05,
      "loss": 2.1026,
      "step": 15150
    },
    {
      "epoch": 1.16,
      "learning_rate": 2.1825396825396827e-05,
      "loss": 2.0849,
      "step": 15200
    },
    {
      "epoch": 1.16,
      "learning_rate": 2.172619047619048e-05,
      "loss": 2.1208,
      "step": 15250
    },
    {
      "epoch": 1.17,
      "learning_rate": 2.1626984126984128e-05,
      "loss": 2.0993,
      "step": 15300
    },
    {
      "epoch": 1.17,
      "learning_rate": 2.152777777777778e-05,
      "loss": 2.0567,
      "step": 15350
    },
    {
      "epoch": 1.18,
      "learning_rate": 2.1428571428571428e-05,
      "loss": 2.0269,
      "step": 15400
    },
    {
      "epoch": 1.18,
      "learning_rate": 2.132936507936508e-05,
      "loss": 2.0395,
      "step": 15450
    },
    {
      "epoch": 1.18,
      "learning_rate": 2.1230158730158732e-05,
      "loss": 2.0267,
      "step": 15500
    },
    {
      "epoch": 1.19,
      "learning_rate": 2.113095238095238e-05,
      "loss": 2.0488,
      "step": 15550
    },
    {
      "epoch": 1.19,
      "learning_rate": 2.1031746031746032e-05,
      "loss": 2.0858,
      "step": 15600
    },
    {
      "epoch": 1.19,
      "learning_rate": 2.0932539682539684e-05,
      "loss": 2.0541,
      "step": 15650
    },
    {
      "epoch": 1.2,
      "learning_rate": 2.0833333333333336e-05,
      "loss": 2.046,
      "step": 15700
    },
    {
      "epoch": 1.2,
      "learning_rate": 2.0734126984126988e-05,
      "loss": 1.9805,
      "step": 15750
    },
    {
      "epoch": 1.21,
      "learning_rate": 2.0634920634920636e-05,
      "loss": 2.0991,
      "step": 15800
    },
    {
      "epoch": 1.21,
      "learning_rate": 2.0535714285714285e-05,
      "loss": 2.1884,
      "step": 15850
    },
    {
      "epoch": 1.21,
      "learning_rate": 2.0436507936507937e-05,
      "loss": 2.1337,
      "step": 15900
    },
    {
      "epoch": 1.22,
      "learning_rate": 2.033730158730159e-05,
      "loss": 2.0913,
      "step": 15950
    },
    {
      "epoch": 1.22,
      "learning_rate": 2.023809523809524e-05,
      "loss": 2.1773,
      "step": 16000
    },
    {
      "epoch": 1.23,
      "learning_rate": 2.013888888888889e-05,
      "loss": 2.1711,
      "step": 16050
    },
    {
      "epoch": 1.23,
      "learning_rate": 2.003968253968254e-05,
      "loss": 2.0521,
      "step": 16100
    },
    {
      "epoch": 1.23,
      "learning_rate": 1.9940476190476193e-05,
      "loss": 2.0473,
      "step": 16150
    },
    {
      "epoch": 1.24,
      "learning_rate": 1.984126984126984e-05,
      "loss": 2.058,
      "step": 16200
    },
    {
      "epoch": 1.24,
      "learning_rate": 1.9742063492063493e-05,
      "loss": 2.0332,
      "step": 16250
    },
    {
      "epoch": 1.24,
      "learning_rate": 1.9642857142857145e-05,
      "loss": 2.049,
      "step": 16300
    },
    {
      "epoch": 1.25,
      "learning_rate": 1.9543650793650793e-05,
      "loss": 2.106,
      "step": 16350
    },
    {
      "epoch": 1.25,
      "learning_rate": 1.9444444444444445e-05,
      "loss": 2.0819,
      "step": 16400
    },
    {
      "epoch": 1.26,
      "learning_rate": 1.9345238095238097e-05,
      "loss": 2.0966,
      "step": 16450
    },
    {
      "epoch": 1.26,
      "learning_rate": 1.924603174603175e-05,
      "loss": 2.1137,
      "step": 16500
    },
    {
      "epoch": 1.26,
      "learning_rate": 1.9146825396825398e-05,
      "loss": 2.0342,
      "step": 16550
    },
    {
      "epoch": 1.27,
      "learning_rate": 1.9047619047619046e-05,
      "loss": 2.0423,
      "step": 16600
    },
    {
      "epoch": 1.27,
      "learning_rate": 1.8948412698412698e-05,
      "loss": 2.0843,
      "step": 16650
    },
    {
      "epoch": 1.27,
      "learning_rate": 1.884920634920635e-05,
      "loss": 2.0687,
      "step": 16700
    },
    {
      "epoch": 1.28,
      "learning_rate": 1.8750000000000002e-05,
      "loss": 2.0647,
      "step": 16750
    },
    {
      "epoch": 1.28,
      "learning_rate": 1.8650793650793654e-05,
      "loss": 2.0359,
      "step": 16800
    },
    {
      "epoch": 1.29,
      "learning_rate": 1.8551587301587302e-05,
      "loss": 2.0275,
      "step": 16850
    },
    {
      "epoch": 1.29,
      "learning_rate": 1.8452380952380954e-05,
      "loss": 2.1197,
      "step": 16900
    },
    {
      "epoch": 1.29,
      "learning_rate": 1.8353174603174602e-05,
      "loss": 2.054,
      "step": 16950
    },
    {
      "epoch": 1.3,
      "learning_rate": 1.8253968253968254e-05,
      "loss": 2.0854,
      "step": 17000
    },
    {
      "epoch": 1.3,
      "learning_rate": 1.8154761904761906e-05,
      "loss": 2.0493,
      "step": 17050
    },
    {
      "epoch": 1.31,
      "learning_rate": 1.8055555555555555e-05,
      "loss": 2.1042,
      "step": 17100
    },
    {
      "epoch": 1.31,
      "learning_rate": 1.7956349206349207e-05,
      "loss": 1.913,
      "step": 17150
    },
    {
      "epoch": 1.31,
      "learning_rate": 1.785714285714286e-05,
      "loss": 2.1167,
      "step": 17200
    },
    {
      "epoch": 1.32,
      "learning_rate": 1.775793650793651e-05,
      "loss": 2.0636,
      "step": 17250
    },
    {
      "epoch": 1.32,
      "learning_rate": 1.7658730158730162e-05,
      "loss": 2.0994,
      "step": 17300
    },
    {
      "epoch": 1.32,
      "learning_rate": 1.755952380952381e-05,
      "loss": 2.116,
      "step": 17350
    },
    {
      "epoch": 1.33,
      "learning_rate": 1.746031746031746e-05,
      "loss": 2.0227,
      "step": 17400
    },
    {
      "epoch": 1.33,
      "learning_rate": 1.736111111111111e-05,
      "loss": 2.102,
      "step": 17450
    },
    {
      "epoch": 1.34,
      "learning_rate": 1.7261904761904763e-05,
      "loss": 2.1564,
      "step": 17500
    },
    {
      "epoch": 1.34,
      "learning_rate": 1.7162698412698415e-05,
      "loss": 2.0394,
      "step": 17550
    },
    {
      "epoch": 1.34,
      "learning_rate": 1.7063492063492063e-05,
      "loss": 2.0377,
      "step": 17600
    },
    {
      "epoch": 1.35,
      "learning_rate": 1.6964285714285715e-05,
      "loss": 2.119,
      "step": 17650
    },
    {
      "epoch": 1.35,
      "learning_rate": 1.6865079365079367e-05,
      "loss": 2.0541,
      "step": 17700
    },
    {
      "epoch": 1.35,
      "learning_rate": 1.6765873015873016e-05,
      "loss": 2.1053,
      "step": 17750
    },
    {
      "epoch": 1.36,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 2.114,
      "step": 17800
    },
    {
      "epoch": 1.36,
      "learning_rate": 1.656746031746032e-05,
      "loss": 2.0926,
      "step": 17850
    },
    {
      "epoch": 1.37,
      "learning_rate": 1.6468253968253968e-05,
      "loss": 2.1273,
      "step": 17900
    },
    {
      "epoch": 1.37,
      "learning_rate": 1.636904761904762e-05,
      "loss": 2.027,
      "step": 17950
    },
    {
      "epoch": 1.37,
      "learning_rate": 1.626984126984127e-05,
      "loss": 2.1071,
      "step": 18000
    },
    {
      "epoch": 1.38,
      "learning_rate": 1.6170634920634923e-05,
      "loss": 2.0269,
      "step": 18050
    },
    {
      "epoch": 1.38,
      "learning_rate": 1.6071428571428572e-05,
      "loss": 2.0236,
      "step": 18100
    },
    {
      "epoch": 1.39,
      "learning_rate": 1.597222222222222e-05,
      "loss": 2.0724,
      "step": 18150
    },
    {
      "epoch": 1.39,
      "learning_rate": 1.5873015873015872e-05,
      "loss": 2.0522,
      "step": 18200
    },
    {
      "epoch": 1.39,
      "learning_rate": 1.5773809523809524e-05,
      "loss": 2.0657,
      "step": 18250
    },
    {
      "epoch": 1.4,
      "learning_rate": 1.5674603174603176e-05,
      "loss": 2.0205,
      "step": 18300
    },
    {
      "epoch": 1.4,
      "learning_rate": 1.5575396825396828e-05,
      "loss": 1.9731,
      "step": 18350
    },
    {
      "epoch": 1.4,
      "learning_rate": 1.5476190476190476e-05,
      "loss": 2.0794,
      "step": 18400
    },
    {
      "epoch": 1.41,
      "learning_rate": 1.537698412698413e-05,
      "loss": 2.1129,
      "step": 18450
    },
    {
      "epoch": 1.41,
      "learning_rate": 1.527777777777778e-05,
      "loss": 2.0713,
      "step": 18500
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.5178571428571429e-05,
      "loss": 2.0227,
      "step": 18550
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.5079365079365079e-05,
      "loss": 2.0003,
      "step": 18600
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.498015873015873e-05,
      "loss": 1.961,
      "step": 18650
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.4880952380952381e-05,
      "loss": 1.9847,
      "step": 18700
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.4781746031746033e-05,
      "loss": 2.0144,
      "step": 18750
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.4682539682539683e-05,
      "loss": 2.0327,
      "step": 18800
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.4583333333333335e-05,
      "loss": 1.9285,
      "step": 18850
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.4484126984126987e-05,
      "loss": 2.0534,
      "step": 18900
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.4384920634920635e-05,
      "loss": 2.123,
      "step": 18950
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.4285714285714285e-05,
      "loss": 2.0127,
      "step": 19000
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.4186507936507937e-05,
      "loss": 2.1508,
      "step": 19050
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.4087301587301587e-05,
      "loss": 2.1037,
      "step": 19100
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.398809523809524e-05,
      "loss": 2.0962,
      "step": 19150
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.388888888888889e-05,
      "loss": 2.0785,
      "step": 19200
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.3789682539682541e-05,
      "loss": 2.1065,
      "step": 19250
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.3690476190476192e-05,
      "loss": 2.1285,
      "step": 19300
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.359126984126984e-05,
      "loss": 2.0568,
      "step": 19350
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.3492063492063492e-05,
      "loss": 2.1088,
      "step": 19400
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.3392857142857144e-05,
      "loss": 2.0946,
      "step": 19450
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.3293650793650794e-05,
      "loss": 2.0457,
      "step": 19500
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.3194444444444446e-05,
      "loss": 2.0999,
      "step": 19550
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.3095238095238096e-05,
      "loss": 2.0653,
      "step": 19600
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.2996031746031748e-05,
      "loss": 2.0811,
      "step": 19650
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.2896825396825398e-05,
      "loss": 2.0284,
      "step": 19700
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.2797619047619047e-05,
      "loss": 2.0423,
      "step": 19750
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.2698412698412699e-05,
      "loss": 1.9918,
      "step": 19800
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.2599206349206349e-05,
      "loss": 2.0226,
      "step": 19850
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.25e-05,
      "loss": 2.0711,
      "step": 19900
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.2400793650793652e-05,
      "loss": 2.0474,
      "step": 19950
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.2301587301587301e-05,
      "loss": 2.1422,
      "step": 20000
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.2202380952380953e-05,
      "loss": 2.0376,
      "step": 20050
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.2103174603174603e-05,
      "loss": 2.056,
      "step": 20100
    },
    {
      "epoch": 1.54,
      "learning_rate": 1.2003968253968255e-05,
      "loss": 2.1421,
      "step": 20150
    },
    {
      "epoch": 1.54,
      "learning_rate": 1.1904761904761905e-05,
      "loss": 2.1107,
      "step": 20200
    },
    {
      "epoch": 1.55,
      "learning_rate": 1.1805555555555555e-05,
      "loss": 2.0642,
      "step": 20250
    },
    {
      "epoch": 1.55,
      "learning_rate": 1.1706349206349207e-05,
      "loss": 2.0525,
      "step": 20300
    },
    {
      "epoch": 1.55,
      "learning_rate": 1.1607142857142857e-05,
      "loss": 2.0263,
      "step": 20350
    },
    {
      "epoch": 1.56,
      "learning_rate": 1.1507936507936508e-05,
      "loss": 2.1769,
      "step": 20400
    },
    {
      "epoch": 1.56,
      "learning_rate": 1.140873015873016e-05,
      "loss": 2.0337,
      "step": 20450
    },
    {
      "epoch": 1.56,
      "learning_rate": 1.130952380952381e-05,
      "loss": 1.9124,
      "step": 20500
    },
    {
      "epoch": 1.57,
      "learning_rate": 1.1210317460317461e-05,
      "loss": 2.03,
      "step": 20550
    },
    {
      "epoch": 1.57,
      "learning_rate": 1.1111111111111112e-05,
      "loss": 2.0838,
      "step": 20600
    },
    {
      "epoch": 1.58,
      "learning_rate": 1.1011904761904762e-05,
      "loss": 2.0309,
      "step": 20650
    },
    {
      "epoch": 1.58,
      "learning_rate": 1.0912698412698414e-05,
      "loss": 2.1253,
      "step": 20700
    },
    {
      "epoch": 1.58,
      "learning_rate": 1.0813492063492064e-05,
      "loss": 2.0027,
      "step": 20750
    },
    {
      "epoch": 1.59,
      "learning_rate": 1.0714285714285714e-05,
      "loss": 2.1645,
      "step": 20800
    },
    {
      "epoch": 1.59,
      "learning_rate": 1.0615079365079366e-05,
      "loss": 1.9856,
      "step": 20850
    },
    {
      "epoch": 1.6,
      "learning_rate": 1.0515873015873016e-05,
      "loss": 2.0447,
      "step": 20900
    },
    {
      "epoch": 1.6,
      "learning_rate": 1.0416666666666668e-05,
      "loss": 2.0944,
      "step": 20950
    },
    {
      "epoch": 1.6,
      "learning_rate": 1.0317460317460318e-05,
      "loss": 2.0401,
      "step": 21000
    },
    {
      "epoch": 1.61,
      "learning_rate": 1.0218253968253968e-05,
      "loss": 2.0073,
      "step": 21050
    },
    {
      "epoch": 1.61,
      "learning_rate": 1.011904761904762e-05,
      "loss": 1.947,
      "step": 21100
    },
    {
      "epoch": 1.61,
      "learning_rate": 1.001984126984127e-05,
      "loss": 2.049,
      "step": 21150
    },
    {
      "epoch": 1.62,
      "learning_rate": 9.92063492063492e-06,
      "loss": 2.0946,
      "step": 21200
    },
    {
      "epoch": 1.62,
      "learning_rate": 9.821428571428573e-06,
      "loss": 2.1091,
      "step": 21250
    },
    {
      "epoch": 1.63,
      "learning_rate": 9.722222222222223e-06,
      "loss": 2.0092,
      "step": 21300
    },
    {
      "epoch": 1.63,
      "learning_rate": 9.623015873015875e-06,
      "loss": 1.972,
      "step": 21350
    },
    {
      "epoch": 1.63,
      "learning_rate": 9.523809523809523e-06,
      "loss": 2.0306,
      "step": 21400
    },
    {
      "epoch": 1.64,
      "learning_rate": 9.424603174603175e-06,
      "loss": 2.0901,
      "step": 21450
    },
    {
      "epoch": 1.64,
      "learning_rate": 9.325396825396827e-06,
      "loss": 2.0199,
      "step": 21500
    },
    {
      "epoch": 1.65,
      "learning_rate": 9.226190476190477e-06,
      "loss": 1.9885,
      "step": 21550
    },
    {
      "epoch": 1.65,
      "learning_rate": 9.126984126984127e-06,
      "loss": 2.0567,
      "step": 21600
    },
    {
      "epoch": 1.65,
      "learning_rate": 9.027777777777777e-06,
      "loss": 2.1078,
      "step": 21650
    },
    {
      "epoch": 1.66,
      "learning_rate": 8.92857142857143e-06,
      "loss": 2.0263,
      "step": 21700
    },
    {
      "epoch": 1.66,
      "learning_rate": 8.829365079365081e-06,
      "loss": 2.0338,
      "step": 21750
    },
    {
      "epoch": 1.66,
      "learning_rate": 8.73015873015873e-06,
      "loss": 2.024,
      "step": 21800
    },
    {
      "epoch": 1.67,
      "learning_rate": 8.630952380952381e-06,
      "loss": 2.0244,
      "step": 21850
    },
    {
      "epoch": 1.67,
      "learning_rate": 8.531746031746032e-06,
      "loss": 2.0738,
      "step": 21900
    },
    {
      "epoch": 1.68,
      "learning_rate": 8.432539682539684e-06,
      "loss": 2.0932,
      "step": 21950
    },
    {
      "epoch": 1.68,
      "learning_rate": 8.333333333333334e-06,
      "loss": 2.0231,
      "step": 22000
    },
    {
      "epoch": 1.68,
      "learning_rate": 8.234126984126984e-06,
      "loss": 2.0454,
      "step": 22050
    },
    {
      "epoch": 1.69,
      "learning_rate": 8.134920634920636e-06,
      "loss": 2.0424,
      "step": 22100
    },
    {
      "epoch": 1.69,
      "learning_rate": 8.035714285714286e-06,
      "loss": 2.0762,
      "step": 22150
    },
    {
      "epoch": 1.69,
      "learning_rate": 7.936507936507936e-06,
      "loss": 2.0292,
      "step": 22200
    },
    {
      "epoch": 1.7,
      "learning_rate": 7.837301587301588e-06,
      "loss": 2.0349,
      "step": 22250
    },
    {
      "epoch": 1.7,
      "learning_rate": 7.738095238095238e-06,
      "loss": 2.0433,
      "step": 22300
    },
    {
      "epoch": 1.71,
      "learning_rate": 7.63888888888889e-06,
      "loss": 2.0813,
      "step": 22350
    },
    {
      "epoch": 1.71,
      "learning_rate": 7.5396825396825394e-06,
      "loss": 2.0036,
      "step": 22400
    },
    {
      "epoch": 1.71,
      "learning_rate": 7.4404761904761905e-06,
      "loss": 1.9523,
      "step": 22450
    },
    {
      "epoch": 1.72,
      "learning_rate": 7.3412698412698415e-06,
      "loss": 2.0789,
      "step": 22500
    },
    {
      "epoch": 1.72,
      "learning_rate": 7.242063492063493e-06,
      "loss": 1.9412,
      "step": 22550
    },
    {
      "epoch": 1.73,
      "learning_rate": 7.142857142857143e-06,
      "loss": 1.9535,
      "step": 22600
    },
    {
      "epoch": 1.73,
      "learning_rate": 7.043650793650794e-06,
      "loss": 2.0737,
      "step": 22650
    },
    {
      "epoch": 1.73,
      "learning_rate": 6.944444444444445e-06,
      "loss": 2.0372,
      "step": 22700
    },
    {
      "epoch": 1.74,
      "learning_rate": 6.845238095238096e-06,
      "loss": 1.9616,
      "step": 22750
    },
    {
      "epoch": 1.74,
      "learning_rate": 6.746031746031746e-06,
      "loss": 2.0529,
      "step": 22800
    },
    {
      "epoch": 1.74,
      "learning_rate": 6.646825396825397e-06,
      "loss": 1.9972,
      "step": 22850
    },
    {
      "epoch": 1.75,
      "learning_rate": 6.547619047619048e-06,
      "loss": 2.1145,
      "step": 22900
    },
    {
      "epoch": 1.75,
      "learning_rate": 6.448412698412699e-06,
      "loss": 1.9787,
      "step": 22950
    },
    {
      "epoch": 1.76,
      "learning_rate": 6.349206349206349e-06,
      "loss": 2.0294,
      "step": 23000
    },
    {
      "epoch": 1.76,
      "learning_rate": 6.25e-06,
      "loss": 1.9645,
      "step": 23050
    },
    {
      "epoch": 1.76,
      "learning_rate": 6.1507936507936505e-06,
      "loss": 1.9659,
      "step": 23100
    },
    {
      "epoch": 1.77,
      "learning_rate": 6.0515873015873015e-06,
      "loss": 1.9515,
      "step": 23150
    },
    {
      "epoch": 1.77,
      "learning_rate": 5.9523809523809525e-06,
      "loss": 2.0043,
      "step": 23200
    },
    {
      "epoch": 1.77,
      "learning_rate": 5.8531746031746036e-06,
      "loss": 1.9432,
      "step": 23250
    },
    {
      "epoch": 1.78,
      "learning_rate": 5.753968253968254e-06,
      "loss": 1.9595,
      "step": 23300
    },
    {
      "epoch": 1.78,
      "learning_rate": 5.654761904761905e-06,
      "loss": 2.172,
      "step": 23350
    },
    {
      "epoch": 1.79,
      "learning_rate": 5.555555555555556e-06,
      "loss": 2.0694,
      "step": 23400
    },
    {
      "epoch": 1.79,
      "learning_rate": 5.456349206349207e-06,
      "loss": 1.9371,
      "step": 23450
    },
    {
      "epoch": 1.79,
      "learning_rate": 5.357142857142857e-06,
      "loss": 1.9168,
      "step": 23500
    },
    {
      "epoch": 1.8,
      "learning_rate": 5.257936507936508e-06,
      "loss": 2.0086,
      "step": 23550
    },
    {
      "epoch": 1.8,
      "learning_rate": 5.158730158730159e-06,
      "loss": 2.0677,
      "step": 23600
    },
    {
      "epoch": 1.81,
      "learning_rate": 5.05952380952381e-06,
      "loss": 1.9763,
      "step": 23650
    },
    {
      "epoch": 1.81,
      "learning_rate": 4.96031746031746e-06,
      "loss": 2.016,
      "step": 23700
    },
    {
      "epoch": 1.81,
      "learning_rate": 4.861111111111111e-06,
      "loss": 2.0365,
      "step": 23750
    },
    {
      "epoch": 1.82,
      "learning_rate": 4.7619047619047615e-06,
      "loss": 1.9687,
      "step": 23800
    },
    {
      "epoch": 1.82,
      "learning_rate": 4.662698412698413e-06,
      "loss": 1.9616,
      "step": 23850
    },
    {
      "epoch": 1.82,
      "learning_rate": 4.563492063492064e-06,
      "loss": 2.0307,
      "step": 23900
    },
    {
      "epoch": 1.83,
      "learning_rate": 4.464285714285715e-06,
      "loss": 2.0172,
      "step": 23950
    },
    {
      "epoch": 1.83,
      "learning_rate": 4.365079365079365e-06,
      "loss": 2.0068,
      "step": 24000
    },
    {
      "epoch": 1.84,
      "learning_rate": 4.265873015873016e-06,
      "loss": 2.042,
      "step": 24050
    },
    {
      "epoch": 1.84,
      "learning_rate": 4.166666666666667e-06,
      "loss": 1.9672,
      "step": 24100
    },
    {
      "epoch": 1.84,
      "learning_rate": 4.067460317460318e-06,
      "loss": 2.1522,
      "step": 24150
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.968253968253968e-06,
      "loss": 1.9359,
      "step": 24200
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.869047619047619e-06,
      "loss": 2.0953,
      "step": 24250
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.7698412698412697e-06,
      "loss": 2.036,
      "step": 24300
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.6706349206349208e-06,
      "loss": 1.8848,
      "step": 24350
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.5714285714285714e-06,
      "loss": 2.0225,
      "step": 24400
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.4722222222222224e-06,
      "loss": 1.994,
      "step": 24450
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.373015873015873e-06,
      "loss": 2.0587,
      "step": 24500
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.273809523809524e-06,
      "loss": 2.0663,
      "step": 24550
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.1746031746031746e-06,
      "loss": 1.9933,
      "step": 24600
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.0753968253968252e-06,
      "loss": 2.0767,
      "step": 24650
    },
    {
      "epoch": 1.89,
      "learning_rate": 2.9761904761904763e-06,
      "loss": 2.0258,
      "step": 24700
    },
    {
      "epoch": 1.89,
      "learning_rate": 2.876984126984127e-06,
      "loss": 1.9668,
      "step": 24750
    },
    {
      "epoch": 1.89,
      "learning_rate": 2.777777777777778e-06,
      "loss": 2.1309,
      "step": 24800
    },
    {
      "epoch": 1.9,
      "learning_rate": 2.6785714285714285e-06,
      "loss": 1.9892,
      "step": 24850
    },
    {
      "epoch": 1.9,
      "learning_rate": 2.5793650793650795e-06,
      "loss": 2.0322,
      "step": 24900
    },
    {
      "epoch": 1.9,
      "learning_rate": 2.48015873015873e-06,
      "loss": 2.0309,
      "step": 24950
    },
    {
      "epoch": 1.91,
      "learning_rate": 2.3809523809523808e-06,
      "loss": 2.0231,
      "step": 25000
    },
    {
      "epoch": 1.91,
      "learning_rate": 2.281746031746032e-06,
      "loss": 2.0231,
      "step": 25050
    },
    {
      "epoch": 1.92,
      "learning_rate": 2.1825396825396824e-06,
      "loss": 1.9826,
      "step": 25100
    },
    {
      "epoch": 1.92,
      "learning_rate": 2.0833333333333334e-06,
      "loss": 1.9962,
      "step": 25150
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.984126984126984e-06,
      "loss": 2.1486,
      "step": 25200
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.8849206349206349e-06,
      "loss": 2.0243,
      "step": 25250
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.7857142857142857e-06,
      "loss": 2.1495,
      "step": 25300
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.6865079365079365e-06,
      "loss": 2.0119,
      "step": 25350
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.5873015873015873e-06,
      "loss": 2.0651,
      "step": 25400
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.4880952380952381e-06,
      "loss": 1.987,
      "step": 25450
    },
    {
      "epoch": 1.95,
      "learning_rate": 1.388888888888889e-06,
      "loss": 2.077,
      "step": 25500
    }
  ],
  "logging_steps": 50,
  "max_steps": 26200,
  "num_train_epochs": 2,
  "save_steps": 1500,
  "total_flos": 5.393246846976e+16,
  "trial_name": null,
  "trial_params": null
}