| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 5.0, | |
| "global_step": 265, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9811320754716985e-05, | |
| "loss": 18.4602, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.962264150943397e-05, | |
| "loss": 16.3339, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.943396226415095e-05, | |
| "loss": 14.7128, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.9245283018867924e-05, | |
| "loss": 13.1405, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.9056603773584906e-05, | |
| "loss": 12.334, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.886792452830189e-05, | |
| "loss": 11.5351, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.867924528301887e-05, | |
| "loss": 11.2585, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.849056603773585e-05, | |
| "loss": 10.6678, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.8301886792452835e-05, | |
| "loss": 10.3116, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.811320754716982e-05, | |
| "loss": 10.1386, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.792452830188679e-05, | |
| "loss": 9.7791, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.7735849056603775e-05, | |
| "loss": 9.4858, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.754716981132076e-05, | |
| "loss": 9.4354, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.735849056603774e-05, | |
| "loss": 9.1703, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.716981132075472e-05, | |
| "loss": 9.0127, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.6981132075471704e-05, | |
| "loss": 8.978, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.679245283018868e-05, | |
| "loss": 8.8674, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.660377358490566e-05, | |
| "loss": 8.7663, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.641509433962264e-05, | |
| "loss": 8.8133, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.6226415094339625e-05, | |
| "loss": 8.5456, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.603773584905661e-05, | |
| "loss": 8.4904, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.584905660377359e-05, | |
| "loss": 8.4489, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.566037735849057e-05, | |
| "loss": 8.4346, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.547169811320755e-05, | |
| "loss": 8.1324, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.528301886792453e-05, | |
| "loss": 8.1567, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.509433962264151e-05, | |
| "loss": 8.1832, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.4905660377358494e-05, | |
| "loss": 8.1429, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.4716981132075476e-05, | |
| "loss": 7.9504, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.452830188679246e-05, | |
| "loss": 7.9522, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.433962264150944e-05, | |
| "loss": 7.7705, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.4150943396226416e-05, | |
| "loss": 7.7457, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.39622641509434e-05, | |
| "loss": 7.8053, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.377358490566038e-05, | |
| "loss": 7.7519, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 4.358490566037736e-05, | |
| "loss": 7.8011, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 4.3396226415094345e-05, | |
| "loss": 7.7119, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.320754716981133e-05, | |
| "loss": 7.4472, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.301886792452831e-05, | |
| "loss": 7.5688, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.2830188679245284e-05, | |
| "loss": 7.3592, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 4.2641509433962266e-05, | |
| "loss": 7.4869, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 4.245283018867925e-05, | |
| "loss": 7.3489, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 4.226415094339623e-05, | |
| "loss": 7.412, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 4.207547169811321e-05, | |
| "loss": 7.2706, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 4.1886792452830195e-05, | |
| "loss": 7.2465, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.169811320754717e-05, | |
| "loss": 7.337, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 4.150943396226415e-05, | |
| "loss": 7.2097, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 4.1320754716981135e-05, | |
| "loss": 7.2128, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 4.113207547169812e-05, | |
| "loss": 7.2417, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.09433962264151e-05, | |
| "loss": 7.2044, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 4.075471698113208e-05, | |
| "loss": 7.1435, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.0566037735849064e-05, | |
| "loss": 7.075, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.037735849056604e-05, | |
| "loss": 6.9571, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 4.018867924528302e-05, | |
| "loss": 6.7984, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4e-05, | |
| "loss": 6.7736, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.9811320754716985e-05, | |
| "loss": 6.8625, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 3.962264150943397e-05, | |
| "loss": 7.0221, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 3.943396226415095e-05, | |
| "loss": 6.969, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 3.924528301886793e-05, | |
| "loss": 6.9754, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 3.905660377358491e-05, | |
| "loss": 6.9933, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 3.886792452830189e-05, | |
| "loss": 6.9458, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 3.867924528301887e-05, | |
| "loss": 6.9104, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 3.8490566037735854e-05, | |
| "loss": 6.809, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 3.8301886792452836e-05, | |
| "loss": 6.7712, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 3.811320754716982e-05, | |
| "loss": 6.671, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 3.7924528301886794e-05, | |
| "loss": 6.7017, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 3.7735849056603776e-05, | |
| "loss": 6.7152, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 3.754716981132076e-05, | |
| "loss": 6.7993, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 3.735849056603773e-05, | |
| "loss": 6.6646, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 3.7169811320754716e-05, | |
| "loss": 6.6046, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 3.69811320754717e-05, | |
| "loss": 6.6799, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 3.679245283018868e-05, | |
| "loss": 6.5709, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 3.660377358490566e-05, | |
| "loss": 6.5575, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 3.641509433962264e-05, | |
| "loss": 6.8022, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 3.622641509433962e-05, | |
| "loss": 6.6274, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 3.60377358490566e-05, | |
| "loss": 6.519, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 3.5849056603773584e-05, | |
| "loss": 6.4789, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 3.5660377358490566e-05, | |
| "loss": 6.5077, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 3.547169811320755e-05, | |
| "loss": 6.4625, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 3.528301886792453e-05, | |
| "loss": 6.6545, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 3.5094339622641506e-05, | |
| "loss": 6.58, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 3.490566037735849e-05, | |
| "loss": 6.7074, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 3.471698113207547e-05, | |
| "loss": 6.5293, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 3.452830188679245e-05, | |
| "loss": 6.4572, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 3.4339622641509435e-05, | |
| "loss": 6.4919, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 3.415094339622642e-05, | |
| "loss": 6.4721, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 3.39622641509434e-05, | |
| "loss": 6.4396, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 3.3773584905660374e-05, | |
| "loss": 6.4422, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 3.3584905660377356e-05, | |
| "loss": 6.6386, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 3.339622641509434e-05, | |
| "loss": 6.3216, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 3.320754716981132e-05, | |
| "loss": 6.5511, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 3.30188679245283e-05, | |
| "loss": 6.3281, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 3.2830188679245285e-05, | |
| "loss": 6.3153, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 3.264150943396227e-05, | |
| "loss": 6.3444, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 3.245283018867924e-05, | |
| "loss": 6.4761, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 3.2264150943396225e-05, | |
| "loss": 6.4824, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 3.207547169811321e-05, | |
| "loss": 6.4817, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 3.188679245283019e-05, | |
| "loss": 6.4784, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 3.169811320754717e-05, | |
| "loss": 6.3177, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 3.1509433962264154e-05, | |
| "loss": 6.304, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 3.132075471698113e-05, | |
| "loss": 6.176, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 3.113207547169811e-05, | |
| "loss": 6.3501, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 3.094339622641509e-05, | |
| "loss": 6.1981, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 3.0754716981132075e-05, | |
| "loss": 6.2568, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 3.056603773584906e-05, | |
| "loss": 6.3171, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 3.0377358490566036e-05, | |
| "loss": 6.2815, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 3.018867924528302e-05, | |
| "loss": 6.2513, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3e-05, | |
| "loss": 6.0677, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 2.9811320754716983e-05, | |
| "loss": 6.3859, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 2.9622641509433962e-05, | |
| "loss": 6.2803, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 2.9433962264150944e-05, | |
| "loss": 6.4167, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 2.9245283018867926e-05, | |
| "loss": 6.234, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 2.9056603773584905e-05, | |
| "loss": 6.3275, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 2.8867924528301887e-05, | |
| "loss": 6.2766, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 2.867924528301887e-05, | |
| "loss": 6.3146, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 2.8490566037735848e-05, | |
| "loss": 6.1047, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 2.830188679245283e-05, | |
| "loss": 6.1752, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 2.8113207547169812e-05, | |
| "loss": 6.2028, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 2.7924528301886794e-05, | |
| "loss": 6.0047, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 2.7735849056603773e-05, | |
| "loss": 6.0241, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 2.7547169811320755e-05, | |
| "loss": 6.25, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 2.7358490566037738e-05, | |
| "loss": 6.2054, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 2.7169811320754716e-05, | |
| "loss": 6.1963, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 2.69811320754717e-05, | |
| "loss": 6.1233, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 2.679245283018868e-05, | |
| "loss": 6.1879, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 2.6603773584905663e-05, | |
| "loss": 6.1532, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 2.641509433962264e-05, | |
| "loss": 5.8332, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 2.6226415094339624e-05, | |
| "loss": 5.9921, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 2.6037735849056606e-05, | |
| "loss": 6.08, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 2.5849056603773585e-05, | |
| "loss": 5.9691, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 2.5660377358490567e-05, | |
| "loss": 5.8631, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 2.547169811320755e-05, | |
| "loss": 5.9265, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 2.5283018867924528e-05, | |
| "loss": 6.0456, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 2.509433962264151e-05, | |
| "loss": 6.0374, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 2.4905660377358492e-05, | |
| "loss": 6.0419, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 2.4716981132075474e-05, | |
| "loss": 5.9543, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 2.4528301886792453e-05, | |
| "loss": 6.0268, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 2.4339622641509435e-05, | |
| "loss": 5.9925, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 2.4150943396226418e-05, | |
| "loss": 6.16, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 2.3962264150943396e-05, | |
| "loss": 5.9924, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 2.377358490566038e-05, | |
| "loss": 5.8427, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 2.358490566037736e-05, | |
| "loss": 5.9679, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 2.339622641509434e-05, | |
| "loss": 6.1985, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 2.320754716981132e-05, | |
| "loss": 5.9773, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 2.3018867924528304e-05, | |
| "loss": 5.9543, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 2.2830188679245286e-05, | |
| "loss": 6.0227, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 2.2641509433962265e-05, | |
| "loss": 5.9294, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 2.2452830188679247e-05, | |
| "loss": 6.1774, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 2.226415094339623e-05, | |
| "loss": 6.0009, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 2.2075471698113208e-05, | |
| "loss": 5.9207, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 2.188679245283019e-05, | |
| "loss": 5.9735, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 2.1698113207547172e-05, | |
| "loss": 5.9751, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 2.1509433962264154e-05, | |
| "loss": 5.8208, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 2.1320754716981133e-05, | |
| "loss": 5.8645, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 2.1132075471698115e-05, | |
| "loss": 5.9207, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 2.0943396226415098e-05, | |
| "loss": 5.972, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 2.0754716981132076e-05, | |
| "loss": 5.8784, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 2.056603773584906e-05, | |
| "loss": 6.0364, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 2.037735849056604e-05, | |
| "loss": 6.0571, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 2.018867924528302e-05, | |
| "loss": 5.8623, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 2e-05, | |
| "loss": 5.6061, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 1.9811320754716984e-05, | |
| "loss": 5.8306, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 1.9622641509433966e-05, | |
| "loss": 6.0213, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 1.9433962264150945e-05, | |
| "loss": 6.022, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 1.9245283018867927e-05, | |
| "loss": 5.9523, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 1.905660377358491e-05, | |
| "loss": 5.8842, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 1.8867924528301888e-05, | |
| "loss": 5.9051, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 1.8679245283018867e-05, | |
| "loss": 5.91, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 1.849056603773585e-05, | |
| "loss": 5.7862, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 1.830188679245283e-05, | |
| "loss": 5.8941, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 1.811320754716981e-05, | |
| "loss": 6.0136, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 1.7924528301886792e-05, | |
| "loss": 5.7909, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 1.7735849056603774e-05, | |
| "loss": 5.8893, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 1.7547169811320753e-05, | |
| "loss": 5.8309, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 3.26, | |
| "learning_rate": 1.7358490566037735e-05, | |
| "loss": 5.6179, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 1.7169811320754717e-05, | |
| "loss": 5.8054, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 1.69811320754717e-05, | |
| "loss": 5.8038, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 1.6792452830188678e-05, | |
| "loss": 5.8717, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 1.660377358490566e-05, | |
| "loss": 5.7068, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "learning_rate": 1.6415094339622643e-05, | |
| "loss": 5.8516, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 1.622641509433962e-05, | |
| "loss": 5.7989, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 1.6037735849056604e-05, | |
| "loss": 5.9828, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 1.5849056603773586e-05, | |
| "loss": 5.8526, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 1.5660377358490564e-05, | |
| "loss": 5.9012, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 1.5471698113207547e-05, | |
| "loss": 5.9608, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 1.528301886792453e-05, | |
| "loss": 5.8062, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 1.509433962264151e-05, | |
| "loss": 5.7624, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 1.4905660377358491e-05, | |
| "loss": 5.8388, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 1.4716981132075472e-05, | |
| "loss": 5.8675, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 1.4528301886792452e-05, | |
| "loss": 5.971, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 1.4339622641509435e-05, | |
| "loss": 5.8333, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "learning_rate": 1.4150943396226415e-05, | |
| "loss": 5.9516, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 1.3962264150943397e-05, | |
| "loss": 5.8128, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 1.3773584905660378e-05, | |
| "loss": 5.6509, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "learning_rate": 1.3584905660377358e-05, | |
| "loss": 5.7938, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 1.339622641509434e-05, | |
| "loss": 5.736, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "learning_rate": 1.320754716981132e-05, | |
| "loss": 5.6844, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 1.3018867924528303e-05, | |
| "loss": 5.7359, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 1.2830188679245283e-05, | |
| "loss": 5.8941, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 1.2641509433962264e-05, | |
| "loss": 5.8553, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 1.2452830188679246e-05, | |
| "loss": 5.6521, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 1.2264150943396227e-05, | |
| "loss": 5.921, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 1.2075471698113209e-05, | |
| "loss": 5.7235, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 1.188679245283019e-05, | |
| "loss": 5.8019, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 1.169811320754717e-05, | |
| "loss": 5.7486, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 1.1509433962264152e-05, | |
| "loss": 5.7429, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 1.1320754716981132e-05, | |
| "loss": 5.8486, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 1.1132075471698115e-05, | |
| "loss": 5.7921, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 1.0943396226415095e-05, | |
| "loss": 5.8232, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 1.0754716981132077e-05, | |
| "loss": 6.083, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "learning_rate": 1.0566037735849058e-05, | |
| "loss": 5.7263, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 1.0377358490566038e-05, | |
| "loss": 5.6783, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 1.018867924528302e-05, | |
| "loss": 5.7754, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 1e-05, | |
| "loss": 6.1204, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 9.811320754716983e-06, | |
| "loss": 5.6473, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 9.622641509433963e-06, | |
| "loss": 5.8676, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 9.433962264150944e-06, | |
| "loss": 5.7221, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 9.245283018867924e-06, | |
| "loss": 5.8749, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 9.056603773584905e-06, | |
| "loss": 5.7737, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 8.867924528301887e-06, | |
| "loss": 5.7663, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 8.679245283018868e-06, | |
| "loss": 5.6642, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 8.49056603773585e-06, | |
| "loss": 5.815, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 8.30188679245283e-06, | |
| "loss": 5.6594, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 4.19, | |
| "learning_rate": 8.11320754716981e-06, | |
| "loss": 6.0083, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 4.21, | |
| "learning_rate": 7.924528301886793e-06, | |
| "loss": 5.7564, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 4.23, | |
| "learning_rate": 7.735849056603773e-06, | |
| "loss": 5.677, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 4.25, | |
| "learning_rate": 7.547169811320755e-06, | |
| "loss": 5.8026, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 4.26, | |
| "learning_rate": 7.358490566037736e-06, | |
| "loss": 5.9558, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 4.28, | |
| "learning_rate": 7.169811320754717e-06, | |
| "loss": 5.6896, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 4.3, | |
| "learning_rate": 6.981132075471699e-06, | |
| "loss": 5.8383, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 4.32, | |
| "learning_rate": 6.792452830188679e-06, | |
| "loss": 5.8355, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 4.34, | |
| "learning_rate": 6.60377358490566e-06, | |
| "loss": 5.6567, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 4.36, | |
| "learning_rate": 6.415094339622642e-06, | |
| "loss": 5.7913, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 4.38, | |
| "learning_rate": 6.226415094339623e-06, | |
| "loss": 5.9586, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "learning_rate": 6.037735849056604e-06, | |
| "loss": 5.7501, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 4.42, | |
| "learning_rate": 5.849056603773585e-06, | |
| "loss": 5.6516, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 4.43, | |
| "learning_rate": 5.660377358490566e-06, | |
| "loss": 5.7582, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 4.45, | |
| "learning_rate": 5.4716981132075475e-06, | |
| "loss": 5.6714, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 4.47, | |
| "learning_rate": 5.283018867924529e-06, | |
| "loss": 5.7091, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 4.49, | |
| "learning_rate": 5.09433962264151e-06, | |
| "loss": 5.6222, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 4.51, | |
| "learning_rate": 4.9056603773584915e-06, | |
| "loss": 5.7866, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 4.53, | |
| "learning_rate": 4.716981132075472e-06, | |
| "loss": 5.6759, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 4.55, | |
| "learning_rate": 4.5283018867924524e-06, | |
| "loss": 5.679, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 4.57, | |
| "learning_rate": 4.339622641509434e-06, | |
| "loss": 5.7489, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 4.58, | |
| "learning_rate": 4.150943396226415e-06, | |
| "loss": 5.6387, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 4.6, | |
| "learning_rate": 3.962264150943396e-06, | |
| "loss": 5.6617, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 4.62, | |
| "learning_rate": 3.7735849056603773e-06, | |
| "loss": 5.6338, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 4.64, | |
| "learning_rate": 3.5849056603773586e-06, | |
| "loss": 5.6264, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 4.66, | |
| "learning_rate": 3.3962264150943395e-06, | |
| "loss": 5.8117, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 3.207547169811321e-06, | |
| "loss": 5.8644, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 4.7, | |
| "learning_rate": 3.018867924528302e-06, | |
| "loss": 5.7927, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 4.72, | |
| "learning_rate": 2.830188679245283e-06, | |
| "loss": 5.7877, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "learning_rate": 2.6415094339622644e-06, | |
| "loss": 5.7415, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 4.75, | |
| "learning_rate": 2.4528301886792457e-06, | |
| "loss": 5.5958, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 4.77, | |
| "learning_rate": 2.2641509433962262e-06, | |
| "loss": 5.4249, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 4.79, | |
| "learning_rate": 2.0754716981132075e-06, | |
| "loss": 5.7735, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 4.81, | |
| "learning_rate": 1.8867924528301887e-06, | |
| "loss": 5.672, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 1.6981132075471698e-06, | |
| "loss": 5.5705, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 4.85, | |
| "learning_rate": 1.509433962264151e-06, | |
| "loss": 5.8487, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 4.87, | |
| "learning_rate": 1.3207547169811322e-06, | |
| "loss": 5.8151, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 4.89, | |
| "learning_rate": 1.1320754716981131e-06, | |
| "loss": 5.6904, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 4.91, | |
| "learning_rate": 9.433962264150943e-07, | |
| "loss": 5.7021, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 4.92, | |
| "learning_rate": 7.547169811320755e-07, | |
| "loss": 5.7196, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 4.94, | |
| "learning_rate": 5.660377358490566e-07, | |
| "loss": 5.8667, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 4.96, | |
| "learning_rate": 3.773584905660378e-07, | |
| "loss": 5.8458, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "learning_rate": 1.886792452830189e-07, | |
| "loss": 5.7733, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 0.0, | |
| "loss": 5.8065, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "step": 265, | |
| "total_flos": 209285213159424.0, | |
| "train_loss": 6.612753483034529, | |
| "train_runtime": 33.2807, | |
| "train_samples_per_second": 503.444, | |
| "train_steps_per_second": 7.963 | |
| } | |
| ], | |
| "max_steps": 265, | |
| "num_train_epochs": 5, | |
| "total_flos": 209285213159424.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |