| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "global_step": 446, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0, | |
| "loss": 9.0661, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0, | |
| "loss": 8.5613, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0, | |
| "loss": 8.8408, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.000000000000001e-08, | |
| "loss": 8.8901, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.2000000000000002e-07, | |
| "loss": 8.6498, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.8e-07, | |
| "loss": 8.8394, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 2.4000000000000003e-07, | |
| "loss": 8.7049, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.0000000000000004e-07, | |
| "loss": 8.5633, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6e-07, | |
| "loss": 8.5671, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.2e-07, | |
| "loss": 8.666, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.800000000000001e-07, | |
| "loss": 8.5021, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5.4e-07, | |
| "loss": 8.4256, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 6.000000000000001e-07, | |
| "loss": 8.3533, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 6.000000000000001e-07, | |
| "loss": 8.5756, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 6.599999999999999e-07, | |
| "loss": 8.268, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 7.2e-07, | |
| "loss": 8.5085, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 7.799999999999999e-07, | |
| "loss": 8.2511, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.4e-07, | |
| "loss": 8.2268, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9e-07, | |
| "loss": 8.2555, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.600000000000001e-07, | |
| "loss": 8.2484, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.0200000000000002e-06, | |
| "loss": 8.4371, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.08e-06, | |
| "loss": 8.2564, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.14e-06, | |
| "loss": 8.1644, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.2000000000000002e-06, | |
| "loss": 8.202, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.26e-06, | |
| "loss": 8.0417, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.3199999999999999e-06, | |
| "loss": 8.0118, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.38e-06, | |
| "loss": 7.8806, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.44e-06, | |
| "loss": 7.8138, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.5e-06, | |
| "loss": 7.9453, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.5599999999999999e-06, | |
| "loss": 7.7751, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.62e-06, | |
| "loss": 7.7891, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.68e-06, | |
| "loss": 7.8177, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.74e-06, | |
| "loss": 7.726, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.8e-06, | |
| "loss": 7.7002, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.86e-06, | |
| "loss": 7.7217, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.9200000000000003e-06, | |
| "loss": 7.7386, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.98e-06, | |
| "loss": 7.6806, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 2.0400000000000004e-06, | |
| "loss": 7.6063, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 2.1000000000000002e-06, | |
| "loss": 7.6285, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 2.16e-06, | |
| "loss": 7.5746, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 2.22e-06, | |
| "loss": 7.6043, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 2.28e-06, | |
| "loss": 7.6903, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 2.34e-06, | |
| "loss": 7.4152, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 7.6104, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 2.46e-06, | |
| "loss": 7.6849, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 2.52e-06, | |
| "loss": 7.6706, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 2.58e-06, | |
| "loss": 7.6569, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 2.6399999999999997e-06, | |
| "loss": 7.7951, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 2.7e-06, | |
| "loss": 7.6621, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 2.76e-06, | |
| "loss": 7.7049, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 2.82e-06, | |
| "loss": 7.4315, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 2.88e-06, | |
| "loss": 7.3429, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 2.9400000000000002e-06, | |
| "loss": 7.4178, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 3e-06, | |
| "loss": 7.3289, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 3.06e-06, | |
| "loss": 7.2022, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 3.1199999999999998e-06, | |
| "loss": 7.3158, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 3.18e-06, | |
| "loss": 7.3564, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 3.24e-06, | |
| "loss": 7.4193, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 3.3e-06, | |
| "loss": 7.2878, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 3.36e-06, | |
| "loss": 7.2734, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 3.4200000000000003e-06, | |
| "loss": 7.2506, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 3.48e-06, | |
| "loss": 7.2649, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 3.54e-06, | |
| "loss": 7.2061, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 3.6e-06, | |
| "loss": 7.2911, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 3.66e-06, | |
| "loss": 7.2255, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 3.72e-06, | |
| "loss": 7.1992, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 3.7800000000000002e-06, | |
| "loss": 7.2259, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 3.8400000000000005e-06, | |
| "loss": 7.2138, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 3.9e-06, | |
| "loss": 7.2534, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 3.96e-06, | |
| "loss": 7.1484, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.0200000000000005e-06, | |
| "loss": 7.2939, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.080000000000001e-06, | |
| "loss": 7.2613, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.14e-06, | |
| "loss": 7.2353, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.2000000000000004e-06, | |
| "loss": 7.2668, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.26e-06, | |
| "loss": 7.2238, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.32e-06, | |
| "loss": 7.2507, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.3799999999999996e-06, | |
| "loss": 7.2616, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.44e-06, | |
| "loss": 7.2783, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.5e-06, | |
| "loss": 7.1368, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.56e-06, | |
| "loss": 7.2477, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.62e-06, | |
| "loss": 7.2056, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.68e-06, | |
| "loss": 7.2238, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.74e-06, | |
| "loss": 7.2219, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.800000000000001e-06, | |
| "loss": 7.2008, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.86e-06, | |
| "loss": 7.1812, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.92e-06, | |
| "loss": 7.2346, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.980000000000001e-06, | |
| "loss": 7.1994, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 5.04e-06, | |
| "loss": 7.1441, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 5.1e-06, | |
| "loss": 7.1018, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 5.16e-06, | |
| "loss": 7.2426, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 5.22e-06, | |
| "loss": 7.1448, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 5.279999999999999e-06, | |
| "loss": 7.3725, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 5.34e-06, | |
| "loss": 7.2283, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 5.4e-06, | |
| "loss": 7.0783, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 5.46e-06, | |
| "loss": 6.9872, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 5.52e-06, | |
| "loss": 7.219, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 5.58e-06, | |
| "loss": 7.1056, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 5.64e-06, | |
| "loss": 6.8522, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 5.7000000000000005e-06, | |
| "loss": 6.9854, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 5.76e-06, | |
| "loss": 6.6152, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 5.82e-06, | |
| "loss": 7.2604, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 5.8800000000000005e-06, | |
| "loss": 7.2338, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 5.940000000000001e-06, | |
| "loss": 7.1187, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 6e-06, | |
| "loss": 7.1481, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 6.0600000000000004e-06, | |
| "loss": 7.157, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 6.12e-06, | |
| "loss": 7.1806, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 6.18e-06, | |
| "loss": 7.1208, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 6.2399999999999995e-06, | |
| "loss": 7.1334, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 6.3e-06, | |
| "loss": 7.0156, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 6.36e-06, | |
| "loss": 7.1107, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 6.42e-06, | |
| "loss": 7.0161, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 6.48e-06, | |
| "loss": 7.0205, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 6.54e-06, | |
| "loss": 7.0943, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 6.6e-06, | |
| "loss": 7.1342, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 6.660000000000001e-06, | |
| "loss": 7.066, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 6.72e-06, | |
| "loss": 7.1983, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 6.78e-06, | |
| "loss": 7.1548, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 6.840000000000001e-06, | |
| "loss": 7.1237, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 6.900000000000001e-06, | |
| "loss": 7.0314, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 6.96e-06, | |
| "loss": 7.1185, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 7.0200000000000006e-06, | |
| "loss": 7.059, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 7.08e-06, | |
| "loss": 7.0087, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 7.14e-06, | |
| "loss": 7.0584, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 7.2e-06, | |
| "loss": 7.0146, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 7.26e-06, | |
| "loss": 7.1049, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 7.32e-06, | |
| "loss": 7.1239, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 7.3800000000000005e-06, | |
| "loss": 7.1031, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 7.44e-06, | |
| "loss": 7.0613, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 7.5e-06, | |
| "loss": 7.127, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 7.5600000000000005e-06, | |
| "loss": 7.1652, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 7.62e-06, | |
| "loss": 6.9731, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 7.680000000000001e-06, | |
| "loss": 7.1539, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 7.74e-06, | |
| "loss": 7.0214, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 7.8e-06, | |
| "loss": 7.0548, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 7.860000000000001e-06, | |
| "loss": 7.1714, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 7.92e-06, | |
| "loss": 6.9496, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 7.98e-06, | |
| "loss": 7.1553, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 8.040000000000001e-06, | |
| "loss": 7.0663, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 8.1e-06, | |
| "loss": 7.0589, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 8.160000000000001e-06, | |
| "loss": 7.1641, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 8.220000000000001e-06, | |
| "loss": 7.1462, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 8.28e-06, | |
| "loss": 7.1284, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 8.340000000000001e-06, | |
| "loss": 6.9173, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 8.400000000000001e-06, | |
| "loss": 6.9755, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 8.459999999999999e-06, | |
| "loss": 6.9108, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 8.52e-06, | |
| "loss": 6.9447, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 8.58e-06, | |
| "loss": 7.0189, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 8.64e-06, | |
| "loss": 7.0141, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 8.7e-06, | |
| "loss": 6.795, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 8.759999999999999e-06, | |
| "loss": 6.657, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 8.82e-06, | |
| "loss": 7.1582, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 8.88e-06, | |
| "loss": 7.1828, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 8.939999999999999e-06, | |
| "loss": 7.1247, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 9e-06, | |
| "loss": 7.0606, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 9.06e-06, | |
| "loss": 7.0153, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 9.12e-06, | |
| "loss": 7.0451, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 9.18e-06, | |
| "loss": 7.1135, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 9.24e-06, | |
| "loss": 7.0684, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 9.3e-06, | |
| "loss": 7.0523, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 9.36e-06, | |
| "loss": 7.0882, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 9.42e-06, | |
| "loss": 7.0388, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 9.48e-06, | |
| "loss": 7.0899, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 9.54e-06, | |
| "loss": 7.0785, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 9.600000000000001e-06, | |
| "loss": 7.069, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 9.66e-06, | |
| "loss": 7.0036, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 9.72e-06, | |
| "loss": 7.0883, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 9.780000000000001e-06, | |
| "loss": 6.9732, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 9.84e-06, | |
| "loss": 6.9378, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 9.9e-06, | |
| "loss": 6.9888, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 9.960000000000001e-06, | |
| "loss": 7.1208, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.002e-05, | |
| "loss": 7.0391, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.008e-05, | |
| "loss": 7.0709, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.0140000000000001e-05, | |
| "loss": 7.0227, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.02e-05, | |
| "loss": 7.081, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.0260000000000002e-05, | |
| "loss": 7.0427, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.032e-05, | |
| "loss": 6.9713, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.0379999999999999e-05, | |
| "loss": 6.9645, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.044e-05, | |
| "loss": 7.0354, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.05e-05, | |
| "loss": 7.0338, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.0559999999999999e-05, | |
| "loss": 7.095, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.062e-05, | |
| "loss": 7.0756, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.068e-05, | |
| "loss": 6.9755, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.074e-05, | |
| "loss": 7.0655, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.08e-05, | |
| "loss": 7.1194, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.086e-05, | |
| "loss": 7.0337, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.092e-05, | |
| "loss": 6.9139, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.098e-05, | |
| "loss": 7.0081, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.104e-05, | |
| "loss": 6.9888, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.11e-05, | |
| "loss": 7.0655, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.116e-05, | |
| "loss": 7.1192, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.1220000000000001e-05, | |
| "loss": 7.1267, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.128e-05, | |
| "loss": 6.9655, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.134e-05, | |
| "loss": 7.0361, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.1400000000000001e-05, | |
| "loss": 6.9628, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.146e-05, | |
| "loss": 6.9358, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.152e-05, | |
| "loss": 6.8373, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.1580000000000001e-05, | |
| "loss": 6.7758, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.164e-05, | |
| "loss": 6.9332, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.1700000000000001e-05, | |
| "loss": 6.7629, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.1760000000000001e-05, | |
| "loss": 6.5633, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.182e-05, | |
| "loss": 7.2408, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.1880000000000001e-05, | |
| "loss": 7.2488, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.1940000000000001e-05, | |
| "loss": 7.2284, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.2e-05, | |
| "loss": 7.1491, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.2060000000000001e-05, | |
| "loss": 7.1344, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.2120000000000001e-05, | |
| "loss": 7.1017, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.2180000000000002e-05, | |
| "loss": 6.9784, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.224e-05, | |
| "loss": 7.1243, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.2299999999999999e-05, | |
| "loss": 7.0496, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.236e-05, | |
| "loss": 7.0962, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.242e-05, | |
| "loss": 7.0022, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.2479999999999999e-05, | |
| "loss": 7.0542, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.254e-05, | |
| "loss": 6.9093, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.26e-05, | |
| "loss": 7.009, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.2659999999999999e-05, | |
| "loss": 6.9863, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.272e-05, | |
| "loss": 7.0156, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.278e-05, | |
| "loss": 6.9996, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.284e-05, | |
| "loss": 7.0404, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.29e-05, | |
| "loss": 7.0057, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.296e-05, | |
| "loss": 7.0278, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.302e-05, | |
| "loss": 7.0697, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.308e-05, | |
| "loss": 7.0499, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.314e-05, | |
| "loss": 6.9859, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.32e-05, | |
| "loss": 6.9388, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.326e-05, | |
| "loss": 7.1033, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.3320000000000001e-05, | |
| "loss": 7.0169, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.338e-05, | |
| "loss": 7.1146, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.344e-05, | |
| "loss": 7.0342, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.3500000000000001e-05, | |
| "loss": 6.9501, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.356e-05, | |
| "loss": 7.0036, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.362e-05, | |
| "loss": 7.0218, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.3680000000000001e-05, | |
| "loss": 6.8725, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.374e-05, | |
| "loss": 6.9275, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.3800000000000002e-05, | |
| "loss": 6.9323, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.3860000000000001e-05, | |
| "loss": 7.0464, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.392e-05, | |
| "loss": 6.9281, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.3980000000000002e-05, | |
| "loss": 7.0814, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.4040000000000001e-05, | |
| "loss": 6.9639, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.4099999999999999e-05, | |
| "loss": 6.8773, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.416e-05, | |
| "loss": 7.1066, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.422e-05, | |
| "loss": 6.9166, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.428e-05, | |
| "loss": 6.9288, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.434e-05, | |
| "loss": 7.0485, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.44e-05, | |
| "loss": 6.8353, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.446e-05, | |
| "loss": 6.8549, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.452e-05, | |
| "loss": 6.8157, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.458e-05, | |
| "loss": 6.5884, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.464e-05, | |
| "loss": 6.8573, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.47e-05, | |
| "loss": 6.7128, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.4760000000000001e-05, | |
| "loss": 6.4373, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.482e-05, | |
| "loss": 7.2887, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 1.488e-05, | |
| "loss": 7.2864, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 1.4940000000000001e-05, | |
| "loss": 7.2313, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 1.5e-05, | |
| "loss": 7.1999, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 1.506e-05, | |
| "loss": 7.1424, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 1.5120000000000001e-05, | |
| "loss": 7.0273, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.518e-05, | |
| "loss": 6.9975, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.524e-05, | |
| "loss": 7.0017, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.53e-05, | |
| "loss": 6.9275, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.5360000000000002e-05, | |
| "loss": 6.965, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.542e-05, | |
| "loss": 6.9751, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.548e-05, | |
| "loss": 6.944, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.554e-05, | |
| "loss": 7.0152, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.56e-05, | |
| "loss": 7.002, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.5660000000000003e-05, | |
| "loss": 7.0258, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.5720000000000002e-05, | |
| "loss": 6.8102, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.578e-05, | |
| "loss": 6.9953, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.584e-05, | |
| "loss": 6.8559, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.59e-05, | |
| "loss": 6.9861, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.596e-05, | |
| "loss": 7.0232, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.6020000000000002e-05, | |
| "loss": 6.8916, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.6080000000000002e-05, | |
| "loss": 6.9062, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.614e-05, | |
| "loss": 6.8069, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.62e-05, | |
| "loss": 6.7746, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.626e-05, | |
| "loss": 7.0022, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.6320000000000003e-05, | |
| "loss": 6.8936, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.6380000000000002e-05, | |
| "loss": 6.8545, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.6440000000000002e-05, | |
| "loss": 6.9683, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.65e-05, | |
| "loss": 7.0178, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.656e-05, | |
| "loss": 7.0096, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.6620000000000004e-05, | |
| "loss": 6.9511, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.6680000000000003e-05, | |
| "loss": 6.9008, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.6740000000000002e-05, | |
| "loss": 6.8291, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.6800000000000002e-05, | |
| "loss": 6.9077, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.686e-05, | |
| "loss": 6.9171, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.6919999999999997e-05, | |
| "loss": 6.9774, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.698e-05, | |
| "loss": 6.9438, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.704e-05, | |
| "loss": 6.9102, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.71e-05, | |
| "loss": 6.8897, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.716e-05, | |
| "loss": 6.9512, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.7219999999999998e-05, | |
| "loss": 6.9808, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.728e-05, | |
| "loss": 6.9348, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.734e-05, | |
| "loss": 6.8064, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.74e-05, | |
| "loss": 6.7494, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.746e-05, | |
| "loss": 6.735, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.7519999999999998e-05, | |
| "loss": 6.7118, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.758e-05, | |
| "loss": 6.7464, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.764e-05, | |
| "loss": 6.7746, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.77e-05, | |
| "loss": 6.3516, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.776e-05, | |
| "loss": 6.3981, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.782e-05, | |
| "loss": 7.3073, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.7879999999999998e-05, | |
| "loss": 7.2329, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.794e-05, | |
| "loss": 7.1772, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.8e-05, | |
| "loss": 7.0102, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.806e-05, | |
| "loss": 7.0688, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.812e-05, | |
| "loss": 6.8743, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.818e-05, | |
| "loss": 6.8592, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.824e-05, | |
| "loss": 7.0024, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.83e-05, | |
| "loss": 6.8502, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.836e-05, | |
| "loss": 6.9345, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.842e-05, | |
| "loss": 6.8982, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.848e-05, | |
| "loss": 6.9162, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.854e-05, | |
| "loss": 6.8384, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.86e-05, | |
| "loss": 6.8191, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.866e-05, | |
| "loss": 6.9335, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.872e-05, | |
| "loss": 6.8588, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.878e-05, | |
| "loss": 6.8222, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.884e-05, | |
| "loss": 6.7389, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.8900000000000002e-05, | |
| "loss": 6.8438, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.896e-05, | |
| "loss": 6.8296, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.902e-05, | |
| "loss": 6.8216, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.908e-05, | |
| "loss": 6.7952, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.914e-05, | |
| "loss": 6.8608, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.9200000000000003e-05, | |
| "loss": 6.7743, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.9260000000000002e-05, | |
| "loss": 6.8434, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.932e-05, | |
| "loss": 6.8534, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.938e-05, | |
| "loss": 6.8232, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.944e-05, | |
| "loss": 6.7599, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.95e-05, | |
| "loss": 6.8421, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.9560000000000002e-05, | |
| "loss": 6.8585, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.9620000000000002e-05, | |
| "loss": 6.8067, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.968e-05, | |
| "loss": 6.7269, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.974e-05, | |
| "loss": 6.8948, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.98e-05, | |
| "loss": 6.7869, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.9860000000000003e-05, | |
| "loss": 6.8859, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.9920000000000002e-05, | |
| "loss": 6.7629, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.9980000000000002e-05, | |
| "loss": 6.6553, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.004e-05, | |
| "loss": 6.7713, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.01e-05, | |
| "loss": 6.7695, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.016e-05, | |
| "loss": 6.753, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.0220000000000003e-05, | |
| "loss": 6.6573, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.0280000000000002e-05, | |
| "loss": 6.7765, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.0340000000000002e-05, | |
| "loss": 6.5781, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.04e-05, | |
| "loss": 6.6206, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.046e-05, | |
| "loss": 6.5597, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.0520000000000003e-05, | |
| "loss": 6.6449, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.0580000000000003e-05, | |
| "loss": 6.4841, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.064e-05, | |
| "loss": 6.3366, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.07e-05, | |
| "loss": 6.4165, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.0759999999999998e-05, | |
| "loss": 6.1109, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.082e-05, | |
| "loss": 7.1829, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.088e-05, | |
| "loss": 7.0255, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.094e-05, | |
| "loss": 6.7856, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.1e-05, | |
| "loss": 6.6913, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.1059999999999998e-05, | |
| "loss": 6.7244, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.1119999999999998e-05, | |
| "loss": 6.8081, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.118e-05, | |
| "loss": 6.6442, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.124e-05, | |
| "loss": 6.6805, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.13e-05, | |
| "loss": 6.5563, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.136e-05, | |
| "loss": 6.5178, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.1419999999999998e-05, | |
| "loss": 6.5198, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.148e-05, | |
| "loss": 6.4337, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.154e-05, | |
| "loss": 6.3207, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 2.16e-05, | |
| "loss": 6.3906, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 2.166e-05, | |
| "loss": 6.4776, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 2.172e-05, | |
| "loss": 6.3665, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 2.178e-05, | |
| "loss": 6.4642, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 2.184e-05, | |
| "loss": 6.3793, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 2.19e-05, | |
| "loss": 6.2186, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 2.196e-05, | |
| "loss": 6.2923, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 2.202e-05, | |
| "loss": 6.2903, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 2.208e-05, | |
| "loss": 6.2592, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 2.214e-05, | |
| "loss": 6.173, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 2.22e-05, | |
| "loss": 6.0022, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 2.226e-05, | |
| "loss": 6.1701, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 2.232e-05, | |
| "loss": 6.0781, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 2.238e-05, | |
| "loss": 6.1111, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 2.2440000000000002e-05, | |
| "loss": 6.0296, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 2.25e-05, | |
| "loss": 6.0029, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 2.256e-05, | |
| "loss": 6.0384, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 2.262e-05, | |
| "loss": 6.1197, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 2.268e-05, | |
| "loss": 6.0461, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 2.274e-05, | |
| "loss": 5.9918, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 2.2800000000000002e-05, | |
| "loss": 5.9277, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 2.286e-05, | |
| "loss": 5.9975, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 2.292e-05, | |
| "loss": 5.9381, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 2.298e-05, | |
| "loss": 5.9793, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 2.304e-05, | |
| "loss": 5.9764, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 2.3100000000000002e-05, | |
| "loss": 5.955, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 2.3160000000000002e-05, | |
| "loss": 5.8213, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 2.322e-05, | |
| "loss": 5.9625, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 2.328e-05, | |
| "loss": 5.951, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 2.334e-05, | |
| "loss": 5.7552, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 2.3400000000000003e-05, | |
| "loss": 5.8177, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 2.3460000000000002e-05, | |
| "loss": 5.8311, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 2.3520000000000002e-05, | |
| "loss": 5.8945, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 2.358e-05, | |
| "loss": 5.8766, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 2.364e-05, | |
| "loss": 5.8002, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 2.37e-05, | |
| "loss": 5.9109, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 2.3760000000000003e-05, | |
| "loss": 5.6506, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 2.3820000000000002e-05, | |
| "loss": 6.0025, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 2.3880000000000002e-05, | |
| "loss": 5.8234, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 2.394e-05, | |
| "loss": 5.9229, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 2.4e-05, | |
| "loss": 5.7209, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 2.4060000000000003e-05, | |
| "loss": 5.6651, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 2.4120000000000003e-05, | |
| "loss": 5.824, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 2.4180000000000002e-05, | |
| "loss": 5.7593, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 2.4240000000000002e-05, | |
| "loss": 5.6796, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 2.43e-05, | |
| "loss": 5.6031, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 2.4360000000000004e-05, | |
| "loss": 5.6286, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 2.442e-05, | |
| "loss": 5.7918, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 2.448e-05, | |
| "loss": 5.6609, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 2.454e-05, | |
| "loss": 5.6679, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 2.4599999999999998e-05, | |
| "loss": 5.7014, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 2.4659999999999998e-05, | |
| "loss": 5.7206, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 2.472e-05, | |
| "loss": 5.6453, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 2.478e-05, | |
| "loss": 5.6255, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 2.484e-05, | |
| "loss": 5.5376, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 2.49e-05, | |
| "loss": 5.5823, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 2.4959999999999998e-05, | |
| "loss": 5.5957, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 2.502e-05, | |
| "loss": 5.6762, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 2.508e-05, | |
| "loss": 5.4908, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 2.514e-05, | |
| "loss": 5.5229, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 2.52e-05, | |
| "loss": 5.6172, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 2.526e-05, | |
| "loss": 5.4369, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 2.5319999999999998e-05, | |
| "loss": 5.5638, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 2.538e-05, | |
| "loss": 5.4714, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 2.544e-05, | |
| "loss": 5.441, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 2.55e-05, | |
| "loss": 5.3765, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 2.556e-05, | |
| "loss": 5.4495, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.562e-05, | |
| "loss": 5.4951, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.568e-05, | |
| "loss": 5.6691, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.574e-05, | |
| "loss": 5.4317, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.58e-05, | |
| "loss": 5.49, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 2.586e-05, | |
| "loss": 5.4139, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 2.592e-05, | |
| "loss": 5.4088, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 2.5980000000000002e-05, | |
| "loss": 5.3708, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 2.604e-05, | |
| "loss": 5.4885, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 2.61e-05, | |
| "loss": 5.4998, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 2.616e-05, | |
| "loss": 5.5172, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 2.622e-05, | |
| "loss": 5.3714, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 2.628e-05, | |
| "loss": 5.6947, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 2.6340000000000002e-05, | |
| "loss": 5.3448, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 2.64e-05, | |
| "loss": 5.5714, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 2.646e-05, | |
| "loss": 5.484, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 2.652e-05, | |
| "loss": 5.22, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 446, | |
| "total_flos": 0.0, | |
| "train_loss": 6.890713336756411, | |
| "train_runtime": 5024.7099, | |
| "train_samples_per_second": 5.68, | |
| "train_steps_per_second": 0.089 | |
| } | |
| ], | |
| "max_steps": 446, | |
| "num_train_epochs": 1, | |
| "total_flos": 0.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |