| { | |
| "best_metric": 1.4365631341934204, | |
| "best_model_checkpoint": "test_/checkpoint-1128", | |
| "epoch": 6.0, | |
| "global_step": 1128, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.001, | |
| "loss": 1.6043, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 1.5687600374221802, | |
| "eval_micro_f1": 0.24933333333333332, | |
| "eval_runtime": 51.1135, | |
| "eval_samples_per_second": 58.693, | |
| "eval_steps_per_second": 0.47, | |
| "eval_unavg_f1_rating1": 0.33407572383073497, | |
| "eval_unavg_f1_rating2": 0.0, | |
| "eval_unavg_f1_rating3": 0.0, | |
| "eval_unavg_f1_rating4": 0.36225728155339804, | |
| "eval_unavg_f1_rating5": 0.0034305317324185253, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 0.001, | |
| "loss": 1.5576, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 1.5380204916000366, | |
| "eval_micro_f1": 0.29033333333333333, | |
| "eval_runtime": 51.1315, | |
| "eval_samples_per_second": 58.672, | |
| "eval_steps_per_second": 0.469, | |
| "eval_unavg_f1_rating1": 0.2691823899371069, | |
| "eval_unavg_f1_rating2": 0.20084566596194503, | |
| "eval_unavg_f1_rating3": 0.3557312252964427, | |
| "eval_unavg_f1_rating4": 0.0, | |
| "eval_unavg_f1_rating5": 0.4060676779463244, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 0.001, | |
| "loss": 1.5317, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_loss": 1.4792137145996094, | |
| "eval_micro_f1": 0.362, | |
| "eval_runtime": 51.3091, | |
| "eval_samples_per_second": 58.469, | |
| "eval_steps_per_second": 0.468, | |
| "eval_unavg_f1_rating1": 0.4780600461893765, | |
| "eval_unavg_f1_rating2": 0.26759339704604695, | |
| "eval_unavg_f1_rating3": 0.25631431244153413, | |
| "eval_unavg_f1_rating4": 0.07207207207207207, | |
| "eval_unavg_f1_rating5": 0.5166425470332849, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 0.001, | |
| "loss": 1.5124, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_loss": 1.491315245628357, | |
| "eval_micro_f1": 0.3546666666666667, | |
| "eval_runtime": 51.1691, | |
| "eval_samples_per_second": 58.629, | |
| "eval_steps_per_second": 0.469, | |
| "eval_unavg_f1_rating1": 0.4857908847184987, | |
| "eval_unavg_f1_rating2": 0.0, | |
| "eval_unavg_f1_rating3": 0.3111111111111111, | |
| "eval_unavg_f1_rating4": 0.016, | |
| "eval_unavg_f1_rating5": 0.5133757961783438, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 0.001, | |
| "loss": 1.4943, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_loss": 1.453855276107788, | |
| "eval_micro_f1": 0.349, | |
| "eval_runtime": 51.2727, | |
| "eval_samples_per_second": 58.511, | |
| "eval_steps_per_second": 0.468, | |
| "eval_unavg_f1_rating1": 0.3043478260869565, | |
| "eval_unavg_f1_rating2": 0.2603773584905661, | |
| "eval_unavg_f1_rating3": 0.35506094329623744, | |
| "eval_unavg_f1_rating4": 0.2099056603773585, | |
| "eval_unavg_f1_rating5": 0.5144061841180605, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 0.001, | |
| "loss": 1.4977, | |
| "step": 1128 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_loss": 1.4365631341934204, | |
| "eval_micro_f1": 0.36499999999999994, | |
| "eval_runtime": 51.4954, | |
| "eval_samples_per_second": 58.258, | |
| "eval_steps_per_second": 0.466, | |
| "eval_unavg_f1_rating1": 0.46589018302828616, | |
| "eval_unavg_f1_rating2": 0.027439024390243903, | |
| "eval_unavg_f1_rating3": 0.36269963936115407, | |
| "eval_unavg_f1_rating4": 0.1606425702811245, | |
| "eval_unavg_f1_rating5": 0.5419532324621733, | |
| "step": 1128 | |
| } | |
| ], | |
| "max_steps": 1880, | |
| "num_train_epochs": 10, | |
| "total_flos": 1.7679088885879104e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } |