| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 9.952153110047847, | |
| "global_step": 260, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0, | |
| "loss": 27.0117, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 3.333333333333334e-06, | |
| "loss": 21.6211, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 5.283208335737189e-06, | |
| "loss": 21.7207, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 6.666666666666668e-06, | |
| "loss": 11.6904, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 7.739760316291209e-06, | |
| "loss": 7.9741, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.616541669070522e-06, | |
| "loss": 7.1436, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.357849740192014e-06, | |
| "loss": 4.0469, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1e-05, | |
| "loss": 2.0199, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1e-05, | |
| "loss": 1.5365, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 9.960317460317462e-06, | |
| "loss": 1.3037, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 9.920634920634922e-06, | |
| "loss": 1.336, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 9.880952380952381e-06, | |
| "loss": 0.8995, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 9.841269841269842e-06, | |
| "loss": 1.1336, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.801587301587301e-06, | |
| "loss": 1.0137, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 9.761904761904762e-06, | |
| "loss": 0.8937, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 9.722222222222223e-06, | |
| "loss": 0.8082, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 9.682539682539683e-06, | |
| "loss": 1.116, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 9.642857142857144e-06, | |
| "loss": 1.0788, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 9.603174603174605e-06, | |
| "loss": 1.0808, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 9.563492063492064e-06, | |
| "loss": 0.9711, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 9.523809523809525e-06, | |
| "loss": 0.9958, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 9.484126984126984e-06, | |
| "loss": 0.8673, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 9.444444444444445e-06, | |
| "loss": 0.8284, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 9.404761904761905e-06, | |
| "loss": 1.0445, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 9.365079365079366e-06, | |
| "loss": 0.985, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 9.325396825396827e-06, | |
| "loss": 0.8481, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.285714285714288e-06, | |
| "loss": 0.844, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 9.246031746031747e-06, | |
| "loss": 0.8213, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 9.206349206349207e-06, | |
| "loss": 0.8681, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 9.166666666666666e-06, | |
| "loss": 0.9029, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 9.126984126984127e-06, | |
| "loss": 0.8869, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 9.087301587301588e-06, | |
| "loss": 0.8439, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 9.047619047619049e-06, | |
| "loss": 0.7698, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 9.00793650793651e-06, | |
| "loss": 0.8427, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 8.968253968253968e-06, | |
| "loss": 0.8912, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 8.92857142857143e-06, | |
| "loss": 0.7972, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 8.888888888888888e-06, | |
| "loss": 0.9011, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 8.849206349206349e-06, | |
| "loss": 0.7726, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 8.80952380952381e-06, | |
| "loss": 1.0206, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 8.76984126984127e-06, | |
| "loss": 0.763, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 8.730158730158731e-06, | |
| "loss": 0.8353, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 8.690476190476192e-06, | |
| "loss": 0.8116, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 8.650793650793651e-06, | |
| "loss": 0.7707, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 8.611111111111112e-06, | |
| "loss": 0.7291, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 8.571428571428571e-06, | |
| "loss": 0.7479, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 8.531746031746032e-06, | |
| "loss": 0.6102, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 8.492063492063492e-06, | |
| "loss": 0.8495, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 8.452380952380953e-06, | |
| "loss": 0.6822, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 8.412698412698414e-06, | |
| "loss": 0.7712, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 8.373015873015875e-06, | |
| "loss": 0.7723, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 8.333333333333334e-06, | |
| "loss": 0.7771, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 8.293650793650794e-06, | |
| "loss": 0.7649, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 8.253968253968254e-06, | |
| "loss": 0.6148, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 8.214285714285714e-06, | |
| "loss": 0.6307, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 8.174603174603175e-06, | |
| "loss": 0.5723, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 8.134920634920636e-06, | |
| "loss": 0.5753, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 8.095238095238097e-06, | |
| "loss": 0.6041, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 8.055555555555557e-06, | |
| "loss": 0.7362, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 8.015873015873016e-06, | |
| "loss": 0.6931, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 7.976190476190477e-06, | |
| "loss": 0.6417, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 7.936507936507936e-06, | |
| "loss": 0.5284, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 7.896825396825397e-06, | |
| "loss": 0.6719, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 7.857142857142858e-06, | |
| "loss": 0.5167, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 7.817460317460318e-06, | |
| "loss": 0.3668, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 7.77777777777778e-06, | |
| "loss": 0.5614, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 7.738095238095238e-06, | |
| "loss": 0.6468, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 7.698412698412699e-06, | |
| "loss": 0.4868, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 7.65873015873016e-06, | |
| "loss": 0.5214, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 7.61904761904762e-06, | |
| "loss": 0.5433, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 7.5793650793650795e-06, | |
| "loss": 0.5826, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 7.53968253968254e-06, | |
| "loss": 0.6986, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.5377, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 7.460317460317461e-06, | |
| "loss": 0.7216, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 7.420634920634922e-06, | |
| "loss": 0.59, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 7.380952380952382e-06, | |
| "loss": 0.5512, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 7.3412698412698415e-06, | |
| "loss": 0.7005, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 7.301587301587301e-06, | |
| "loss": 0.6035, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 7.261904761904762e-06, | |
| "loss": 0.602, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 7.222222222222223e-06, | |
| "loss": 0.2625, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 7.182539682539683e-06, | |
| "loss": 0.4799, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 7.1428571428571436e-06, | |
| "loss": 0.3642, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 7.103174603174604e-06, | |
| "loss": 0.4635, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 7.063492063492064e-06, | |
| "loss": 0.4869, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 7.023809523809524e-06, | |
| "loss": 0.3282, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 6.984126984126984e-06, | |
| "loss": 0.3473, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 6.944444444444445e-06, | |
| "loss": 0.3549, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 6.9047619047619055e-06, | |
| "loss": 0.3809, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 6.8650793650793654e-06, | |
| "loss": 0.4178, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 6.825396825396826e-06, | |
| "loss": 0.4177, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 6.785714285714287e-06, | |
| "loss": 0.3836, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 6.746031746031747e-06, | |
| "loss": 0.3887, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 6.706349206349207e-06, | |
| "loss": 0.4161, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 0.3486, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 6.626984126984127e-06, | |
| "loss": 0.4358, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "learning_rate": 6.587301587301588e-06, | |
| "loss": 0.3826, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 6.547619047619048e-06, | |
| "loss": 0.5642, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 6.507936507936509e-06, | |
| "loss": 0.4061, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 6.468253968253969e-06, | |
| "loss": 0.4258, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 6.4285714285714295e-06, | |
| "loss": 0.3696, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 6.3888888888888885e-06, | |
| "loss": 0.3485, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 6.349206349206349e-06, | |
| "loss": 0.3755, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 6.30952380952381e-06, | |
| "loss": 0.4713, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "learning_rate": 6.26984126984127e-06, | |
| "loss": 0.4767, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 6.230158730158731e-06, | |
| "loss": 0.5167, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 6.1904761904761914e-06, | |
| "loss": 0.4248, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 6.150793650793651e-06, | |
| "loss": 0.2968, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 6.111111111111112e-06, | |
| "loss": 0.2642, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 6.071428571428571e-06, | |
| "loss": 0.2437, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 6.031746031746032e-06, | |
| "loss": 0.2496, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 4.21, | |
| "learning_rate": 5.992063492063493e-06, | |
| "loss": 0.3475, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 4.25, | |
| "learning_rate": 5.9523809523809525e-06, | |
| "loss": 0.2465, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 4.29, | |
| "learning_rate": 5.912698412698413e-06, | |
| "loss": 0.2471, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 4.33, | |
| "learning_rate": 5.873015873015874e-06, | |
| "loss": 0.331, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 4.36, | |
| "learning_rate": 5.833333333333334e-06, | |
| "loss": 0.2714, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "learning_rate": 5.793650793650795e-06, | |
| "loss": 0.3384, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "learning_rate": 5.753968253968254e-06, | |
| "loss": 0.2718, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 4.48, | |
| "learning_rate": 5.7142857142857145e-06, | |
| "loss": 0.1803, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 4.52, | |
| "learning_rate": 5.674603174603175e-06, | |
| "loss": 0.2611, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 4.56, | |
| "learning_rate": 5.634920634920635e-06, | |
| "loss": 0.2818, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 4.59, | |
| "learning_rate": 5.595238095238096e-06, | |
| "loss": 0.3179, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 4.63, | |
| "learning_rate": 5.555555555555557e-06, | |
| "loss": 0.2274, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 4.67, | |
| "learning_rate": 5.5158730158730166e-06, | |
| "loss": 0.1747, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 4.71, | |
| "learning_rate": 5.476190476190477e-06, | |
| "loss": 0.2825, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 4.75, | |
| "learning_rate": 5.436507936507936e-06, | |
| "loss": 0.249, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "learning_rate": 5.396825396825397e-06, | |
| "loss": 0.1721, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 4.82, | |
| "learning_rate": 5.357142857142857e-06, | |
| "loss": 0.266, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 4.86, | |
| "learning_rate": 5.317460317460318e-06, | |
| "loss": 0.2893, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 4.9, | |
| "learning_rate": 5.2777777777777785e-06, | |
| "loss": 0.2114, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 4.94, | |
| "learning_rate": 5.2380952380952384e-06, | |
| "loss": 0.3086, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "learning_rate": 5.198412698412699e-06, | |
| "loss": 0.2471, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 5.15873015873016e-06, | |
| "loss": 0.205, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 5.05, | |
| "learning_rate": 5.119047619047619e-06, | |
| "loss": 0.2055, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "learning_rate": 5.07936507936508e-06, | |
| "loss": 0.191, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 5.13, | |
| "learning_rate": 5.03968253968254e-06, | |
| "loss": 0.1575, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 5.17, | |
| "learning_rate": 5e-06, | |
| "loss": 0.1273, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 5.21, | |
| "learning_rate": 4.960317460317461e-06, | |
| "loss": 0.1414, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 5.24, | |
| "learning_rate": 4.920634920634921e-06, | |
| "loss": 0.1245, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 5.28, | |
| "learning_rate": 4.880952380952381e-06, | |
| "loss": 0.1597, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 5.32, | |
| "learning_rate": 4.841269841269842e-06, | |
| "loss": 0.1187, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 5.36, | |
| "learning_rate": 4.8015873015873025e-06, | |
| "loss": 0.1787, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 5.4, | |
| "learning_rate": 4.761904761904762e-06, | |
| "loss": 0.2021, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 5.44, | |
| "learning_rate": 4.722222222222222e-06, | |
| "loss": 0.1358, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 5.47, | |
| "learning_rate": 4.682539682539683e-06, | |
| "loss": 0.1262, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 5.51, | |
| "learning_rate": 4.642857142857144e-06, | |
| "loss": 0.1312, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 5.55, | |
| "learning_rate": 4.603174603174604e-06, | |
| "loss": 0.1239, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 5.59, | |
| "learning_rate": 4.563492063492064e-06, | |
| "loss": 0.1453, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 5.63, | |
| "learning_rate": 4.523809523809524e-06, | |
| "loss": 0.1433, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 5.67, | |
| "learning_rate": 4.484126984126984e-06, | |
| "loss": 0.1165, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 5.7, | |
| "learning_rate": 4.444444444444444e-06, | |
| "loss": 0.1955, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 5.74, | |
| "learning_rate": 4.404761904761905e-06, | |
| "loss": 0.163, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 5.78, | |
| "learning_rate": 4.365079365079366e-06, | |
| "loss": 0.1306, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 5.82, | |
| "learning_rate": 4.3253968253968256e-06, | |
| "loss": 0.142, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 5.86, | |
| "learning_rate": 4.2857142857142855e-06, | |
| "loss": 0.1747, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 5.89, | |
| "learning_rate": 4.246031746031746e-06, | |
| "loss": 0.1584, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 5.93, | |
| "learning_rate": 4.206349206349207e-06, | |
| "loss": 0.0868, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 5.97, | |
| "learning_rate": 4.166666666666667e-06, | |
| "loss": 0.181, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 4.126984126984127e-06, | |
| "loss": 0.0737, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 6.05, | |
| "learning_rate": 4.0873015873015875e-06, | |
| "loss": 0.0959, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 6.09, | |
| "learning_rate": 4.047619047619048e-06, | |
| "loss": 0.1235, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 6.12, | |
| "learning_rate": 4.007936507936508e-06, | |
| "loss": 0.0853, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 6.16, | |
| "learning_rate": 3.968253968253968e-06, | |
| "loss": 0.0876, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 6.2, | |
| "learning_rate": 3.928571428571429e-06, | |
| "loss": 0.0424, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 6.24, | |
| "learning_rate": 3.88888888888889e-06, | |
| "loss": 0.128, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 6.28, | |
| "learning_rate": 3.8492063492063495e-06, | |
| "loss": 0.0874, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 6.32, | |
| "learning_rate": 3.80952380952381e-06, | |
| "loss": 0.0785, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 6.35, | |
| "learning_rate": 3.76984126984127e-06, | |
| "loss": 0.0692, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 6.39, | |
| "learning_rate": 3.7301587301587305e-06, | |
| "loss": 0.0595, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 6.43, | |
| "learning_rate": 3.690476190476191e-06, | |
| "loss": 0.0856, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 6.47, | |
| "learning_rate": 3.6507936507936507e-06, | |
| "loss": 0.1139, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 6.51, | |
| "learning_rate": 3.6111111111111115e-06, | |
| "loss": 0.0859, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 6.55, | |
| "learning_rate": 3.5714285714285718e-06, | |
| "loss": 0.0626, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 6.58, | |
| "learning_rate": 3.531746031746032e-06, | |
| "loss": 0.0843, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 6.62, | |
| "learning_rate": 3.492063492063492e-06, | |
| "loss": 0.0991, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 6.66, | |
| "learning_rate": 3.4523809523809528e-06, | |
| "loss": 0.1023, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 6.7, | |
| "learning_rate": 3.412698412698413e-06, | |
| "loss": 0.0686, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 6.74, | |
| "learning_rate": 3.3730158730158734e-06, | |
| "loss": 0.0896, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 6.78, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 0.084, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 6.81, | |
| "learning_rate": 3.293650793650794e-06, | |
| "loss": 0.1006, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 6.85, | |
| "learning_rate": 3.2539682539682544e-06, | |
| "loss": 0.0901, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 6.89, | |
| "learning_rate": 3.2142857142857147e-06, | |
| "loss": 0.082, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 6.93, | |
| "learning_rate": 3.1746031746031746e-06, | |
| "loss": 0.0704, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 6.97, | |
| "learning_rate": 3.134920634920635e-06, | |
| "loss": 0.071, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 3.0952380952380957e-06, | |
| "loss": 0.0812, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 7.04, | |
| "learning_rate": 3.055555555555556e-06, | |
| "loss": 0.0371, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 7.08, | |
| "learning_rate": 3.015873015873016e-06, | |
| "loss": 0.0393, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 7.12, | |
| "learning_rate": 2.9761904761904763e-06, | |
| "loss": 0.0253, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 7.16, | |
| "learning_rate": 2.936507936507937e-06, | |
| "loss": 0.0624, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 7.2, | |
| "learning_rate": 2.8968253968253974e-06, | |
| "loss": 0.0625, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 7.23, | |
| "learning_rate": 2.8571428571428573e-06, | |
| "loss": 0.0464, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 7.27, | |
| "learning_rate": 2.8174603174603176e-06, | |
| "loss": 0.0354, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 7.31, | |
| "learning_rate": 2.7777777777777783e-06, | |
| "loss": 0.0516, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 7.35, | |
| "learning_rate": 2.7380952380952387e-06, | |
| "loss": 0.0695, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 7.39, | |
| "learning_rate": 2.6984126984126986e-06, | |
| "loss": 0.0405, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 7.43, | |
| "learning_rate": 2.658730158730159e-06, | |
| "loss": 0.036, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 7.46, | |
| "learning_rate": 2.6190476190476192e-06, | |
| "loss": 0.0474, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 7.5, | |
| "learning_rate": 2.57936507936508e-06, | |
| "loss": 0.0445, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 7.54, | |
| "learning_rate": 2.53968253968254e-06, | |
| "loss": 0.0546, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 7.58, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.0553, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 7.62, | |
| "learning_rate": 2.4603174603174605e-06, | |
| "loss": 0.0137, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 7.66, | |
| "learning_rate": 2.420634920634921e-06, | |
| "loss": 0.0406, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 7.69, | |
| "learning_rate": 2.380952380952381e-06, | |
| "loss": 0.0333, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 7.73, | |
| "learning_rate": 2.3412698412698415e-06, | |
| "loss": 0.067, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 7.77, | |
| "learning_rate": 2.301587301587302e-06, | |
| "loss": 0.0636, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 7.81, | |
| "learning_rate": 2.261904761904762e-06, | |
| "loss": 0.0476, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 7.85, | |
| "learning_rate": 2.222222222222222e-06, | |
| "loss": 0.0397, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 7.89, | |
| "learning_rate": 2.182539682539683e-06, | |
| "loss": 0.0251, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 7.92, | |
| "learning_rate": 2.1428571428571427e-06, | |
| "loss": 0.0443, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 7.96, | |
| "learning_rate": 2.1031746031746035e-06, | |
| "loss": 0.0356, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "learning_rate": 2.0634920634920634e-06, | |
| "loss": 0.0444, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 8.04, | |
| "learning_rate": 2.023809523809524e-06, | |
| "loss": 0.0432, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 8.08, | |
| "learning_rate": 1.984126984126984e-06, | |
| "loss": 0.0397, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 8.11, | |
| "learning_rate": 1.944444444444445e-06, | |
| "loss": 0.022, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 8.15, | |
| "learning_rate": 1.904761904761905e-06, | |
| "loss": 0.0377, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 8.19, | |
| "learning_rate": 1.8650793650793652e-06, | |
| "loss": 0.0178, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 8.23, | |
| "learning_rate": 1.8253968253968254e-06, | |
| "loss": 0.0205, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 8.27, | |
| "learning_rate": 1.7857142857142859e-06, | |
| "loss": 0.0285, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 8.31, | |
| "learning_rate": 1.746031746031746e-06, | |
| "loss": 0.0262, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 8.34, | |
| "learning_rate": 1.7063492063492065e-06, | |
| "loss": 0.0196, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 8.38, | |
| "learning_rate": 1.6666666666666667e-06, | |
| "loss": 0.0265, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 8.42, | |
| "learning_rate": 1.6269841269841272e-06, | |
| "loss": 0.0333, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 8.46, | |
| "learning_rate": 1.5873015873015873e-06, | |
| "loss": 0.0202, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 8.5, | |
| "learning_rate": 1.5476190476190479e-06, | |
| "loss": 0.0205, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 8.54, | |
| "learning_rate": 1.507936507936508e-06, | |
| "loss": 0.0234, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 8.57, | |
| "learning_rate": 1.4682539682539685e-06, | |
| "loss": 0.0156, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 8.61, | |
| "learning_rate": 1.4285714285714286e-06, | |
| "loss": 0.0229, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 8.65, | |
| "learning_rate": 1.3888888888888892e-06, | |
| "loss": 0.0247, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 8.69, | |
| "learning_rate": 1.3492063492063493e-06, | |
| "loss": 0.0227, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 8.73, | |
| "learning_rate": 1.3095238095238096e-06, | |
| "loss": 0.02, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 8.77, | |
| "learning_rate": 1.26984126984127e-06, | |
| "loss": 0.0244, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 8.8, | |
| "learning_rate": 1.2301587301587303e-06, | |
| "loss": 0.0264, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 8.84, | |
| "learning_rate": 1.1904761904761906e-06, | |
| "loss": 0.0247, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 8.88, | |
| "learning_rate": 1.150793650793651e-06, | |
| "loss": 0.0275, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 8.92, | |
| "learning_rate": 1.111111111111111e-06, | |
| "loss": 0.0246, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 8.96, | |
| "learning_rate": 1.0714285714285714e-06, | |
| "loss": 0.0216, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "learning_rate": 1.0317460317460317e-06, | |
| "loss": 0.0204, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 9.03, | |
| "learning_rate": 9.92063492063492e-07, | |
| "loss": 0.02, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 9.07, | |
| "learning_rate": 9.523809523809525e-07, | |
| "loss": 0.0113, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 9.11, | |
| "learning_rate": 9.126984126984127e-07, | |
| "loss": 0.0187, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 9.15, | |
| "learning_rate": 8.73015873015873e-07, | |
| "loss": 0.018, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 9.19, | |
| "learning_rate": 8.333333333333333e-07, | |
| "loss": 0.0183, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 9.22, | |
| "learning_rate": 7.936507936507937e-07, | |
| "loss": 0.0173, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 9.26, | |
| "learning_rate": 7.53968253968254e-07, | |
| "loss": 0.0113, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 9.3, | |
| "learning_rate": 7.142857142857143e-07, | |
| "loss": 0.0124, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 9.34, | |
| "learning_rate": 6.746031746031746e-07, | |
| "loss": 0.0148, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 9.38, | |
| "learning_rate": 6.34920634920635e-07, | |
| "loss": 0.0132, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 9.42, | |
| "learning_rate": 5.952380952380953e-07, | |
| "loss": 0.0118, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 9.45, | |
| "learning_rate": 5.555555555555555e-07, | |
| "loss": 0.0116, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 9.49, | |
| "learning_rate": 5.158730158730158e-07, | |
| "loss": 0.0113, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 9.53, | |
| "learning_rate": 4.7619047619047623e-07, | |
| "loss": 0.0143, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 9.57, | |
| "learning_rate": 4.365079365079365e-07, | |
| "loss": 0.0145, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 9.61, | |
| "learning_rate": 3.9682539682539683e-07, | |
| "loss": 0.0175, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 9.65, | |
| "learning_rate": 3.5714285714285716e-07, | |
| "loss": 0.0275, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 9.68, | |
| "learning_rate": 3.174603174603175e-07, | |
| "loss": 0.0106, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 9.72, | |
| "learning_rate": 2.7777777777777776e-07, | |
| "loss": 0.0178, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 9.76, | |
| "learning_rate": 2.3809523809523811e-07, | |
| "loss": 0.0124, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 9.8, | |
| "learning_rate": 1.9841269841269841e-07, | |
| "loss": 0.0111, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 9.84, | |
| "learning_rate": 1.5873015873015874e-07, | |
| "loss": 0.0181, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 9.88, | |
| "learning_rate": 1.1904761904761906e-07, | |
| "loss": 0.0244, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 9.91, | |
| "learning_rate": 7.936507936507937e-08, | |
| "loss": 0.0194, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 9.95, | |
| "learning_rate": 3.9682539682539686e-08, | |
| "loss": 0.0209, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 9.95, | |
| "step": 260, | |
| "total_flos": 41946461159424.0, | |
| "train_loss": 0.7100576841415694, | |
| "train_runtime": 17873.7755, | |
| "train_samples_per_second": 1.871, | |
| "train_steps_per_second": 0.015 | |
| } | |
| ], | |
| "max_steps": 260, | |
| "num_train_epochs": 10, | |
| "total_flos": 41946461159424.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |