| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 19.2, | |
| "global_step": 240, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0001, | |
| "loss": 2.3459, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 9.787234042553192e-05, | |
| "loss": 2.1597, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 9.574468085106384e-05, | |
| "loss": 1.7456, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 9.361702127659576e-05, | |
| "loss": 1.5638, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 9.148936170212766e-05, | |
| "loss": 1.4497, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 8.936170212765958e-05, | |
| "loss": 1.3674, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 8.723404255319149e-05, | |
| "loss": 1.3003, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 8.510638297872341e-05, | |
| "loss": 1.2443, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 8.297872340425533e-05, | |
| "loss": 1.2049, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 8.085106382978723e-05, | |
| "loss": 1.2089, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "learning_rate": 7.872340425531916e-05, | |
| "loss": 1.1278, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 4.8, | |
| "learning_rate": 7.659574468085106e-05, | |
| "loss": 1.1186, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 5.2, | |
| "learning_rate": 7.446808510638298e-05, | |
| "loss": 1.0933, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 5.6, | |
| "learning_rate": 7.23404255319149e-05, | |
| "loss": 1.0794, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 7.021276595744681e-05, | |
| "loss": 1.0174, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 6.4, | |
| "learning_rate": 6.808510638297873e-05, | |
| "loss": 0.9832, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 6.8, | |
| "learning_rate": 6.595744680851063e-05, | |
| "loss": 0.9853, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 7.2, | |
| "learning_rate": 6.382978723404256e-05, | |
| "loss": 0.9455, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 7.6, | |
| "learning_rate": 6.170212765957447e-05, | |
| "loss": 0.9044, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "learning_rate": 5.9574468085106384e-05, | |
| "loss": 0.9453, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 8.4, | |
| "learning_rate": 5.744680851063831e-05, | |
| "loss": 0.8726, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 8.8, | |
| "learning_rate": 5.531914893617022e-05, | |
| "loss": 0.8506, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 9.2, | |
| "learning_rate": 5.319148936170213e-05, | |
| "loss": 0.8197, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 9.6, | |
| "learning_rate": 5.1063829787234044e-05, | |
| "loss": 0.794, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "learning_rate": 4.893617021276596e-05, | |
| "loss": 0.806, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 10.4, | |
| "learning_rate": 4.680851063829788e-05, | |
| "loss": 0.7553, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 10.8, | |
| "learning_rate": 4.468085106382979e-05, | |
| "loss": 0.7085, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 11.2, | |
| "learning_rate": 4.2553191489361704e-05, | |
| "loss": 0.7396, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 11.6, | |
| "learning_rate": 4.0425531914893614e-05, | |
| "loss": 0.6653, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "learning_rate": 3.829787234042553e-05, | |
| "loss": 0.6648, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 12.4, | |
| "learning_rate": 3.617021276595745e-05, | |
| "loss": 0.6164, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 12.8, | |
| "learning_rate": 3.4042553191489365e-05, | |
| "loss": 0.6197, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 13.2, | |
| "learning_rate": 3.191489361702128e-05, | |
| "loss": 0.5731, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 13.6, | |
| "learning_rate": 2.9787234042553192e-05, | |
| "loss": 0.5713, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "learning_rate": 2.765957446808511e-05, | |
| "loss": 0.5612, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 14.4, | |
| "learning_rate": 2.5531914893617022e-05, | |
| "loss": 0.5372, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 14.8, | |
| "learning_rate": 2.340425531914894e-05, | |
| "loss": 0.5098, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 15.2, | |
| "learning_rate": 2.1276595744680852e-05, | |
| "loss": 0.4798, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 15.6, | |
| "learning_rate": 1.9148936170212766e-05, | |
| "loss": 0.4931, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "learning_rate": 1.7021276595744682e-05, | |
| "loss": 0.4652, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 16.4, | |
| "learning_rate": 1.4893617021276596e-05, | |
| "loss": 0.4244, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 16.8, | |
| "learning_rate": 1.2765957446808511e-05, | |
| "loss": 0.4701, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 17.2, | |
| "learning_rate": 1.0638297872340426e-05, | |
| "loss": 0.4204, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 17.6, | |
| "learning_rate": 8.510638297872341e-06, | |
| "loss": 0.4216, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "learning_rate": 6.3829787234042555e-06, | |
| "loss": 0.4294, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 18.4, | |
| "learning_rate": 4.255319148936171e-06, | |
| "loss": 0.4151, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 18.8, | |
| "learning_rate": 2.1276595744680853e-06, | |
| "loss": 0.4077, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 19.2, | |
| "learning_rate": 0.0, | |
| "loss": 0.4046, | |
| "step": 240 | |
| } | |
| ], | |
| "max_steps": 240, | |
| "num_train_epochs": 20, | |
| "total_flos": 6.017479499317248e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |