| { | |
| "best_global_step": 46, | |
| "best_metric": 2.2721331119537354, | |
| "best_model_checkpoint": "bert_base_rand_100_v1_stsb/checkpoint-46", | |
| "epoch": 7.0, | |
| "eval_steps": 500, | |
| "global_step": 161, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 5.593254566192627, | |
| "learning_rate": 4.9e-05, | |
| "loss": 2.928, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_combined_score": 0.09373118085152038, | |
| "eval_loss": 2.4646432399749756, | |
| "eval_pearson": 0.09729602347430324, | |
| "eval_runtime": 1.0234, | |
| "eval_samples_per_second": 1465.672, | |
| "eval_spearmanr": 0.09016633822873751, | |
| "eval_steps_per_second": 5.863, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 9.706378936767578, | |
| "learning_rate": 4.8e-05, | |
| "loss": 1.9606, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_combined_score": 0.15897164929602992, | |
| "eval_loss": 2.2721331119537354, | |
| "eval_pearson": 0.1714444402412414, | |
| "eval_runtime": 0.9742, | |
| "eval_samples_per_second": 1539.784, | |
| "eval_spearmanr": 0.1464988583508184, | |
| "eval_steps_per_second": 6.159, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "grad_norm": 7.275521278381348, | |
| "learning_rate": 4.7e-05, | |
| "loss": 1.6768, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_combined_score": 0.22735907820663778, | |
| "eval_loss": 2.3988537788391113, | |
| "eval_pearson": 0.22905661681057854, | |
| "eval_runtime": 0.9526, | |
| "eval_samples_per_second": 1574.702, | |
| "eval_spearmanr": 0.22566153960269703, | |
| "eval_steps_per_second": 6.299, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "grad_norm": 15.667533874511719, | |
| "learning_rate": 4.600000000000001e-05, | |
| "loss": 1.3265, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_combined_score": 0.27087084601106837, | |
| "eval_loss": 2.473294496536255, | |
| "eval_pearson": 0.26928292870373316, | |
| "eval_runtime": 0.9818, | |
| "eval_samples_per_second": 1527.807, | |
| "eval_spearmanr": 0.27245876331840363, | |
| "eval_steps_per_second": 6.111, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "grad_norm": 12.131975173950195, | |
| "learning_rate": 4.5e-05, | |
| "loss": 1.0076, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_combined_score": 0.24535384308215563, | |
| "eval_loss": 2.8601512908935547, | |
| "eval_pearson": 0.24502840732804015, | |
| "eval_runtime": 0.9804, | |
| "eval_samples_per_second": 1530.057, | |
| "eval_spearmanr": 0.2456792788362711, | |
| "eval_steps_per_second": 6.12, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "grad_norm": 9.518048286437988, | |
| "learning_rate": 4.4000000000000006e-05, | |
| "loss": 0.7949, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_combined_score": 0.2624430834646961, | |
| "eval_loss": 2.748502254486084, | |
| "eval_pearson": 0.2605689404882653, | |
| "eval_runtime": 1.0011, | |
| "eval_samples_per_second": 1498.382, | |
| "eval_spearmanr": 0.2643172264411269, | |
| "eval_steps_per_second": 5.994, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "grad_norm": 10.415647506713867, | |
| "learning_rate": 4.3e-05, | |
| "loss": 0.626, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_combined_score": 0.2801360597793442, | |
| "eval_loss": 2.5590381622314453, | |
| "eval_pearson": 0.27921912329238546, | |
| "eval_runtime": 0.9848, | |
| "eval_samples_per_second": 1523.146, | |
| "eval_spearmanr": 0.28105299626630287, | |
| "eval_steps_per_second": 6.093, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "step": 161, | |
| "total_flos": 5294141566038528.0, | |
| "train_loss": 1.4743308843292804, | |
| "train_runtime": 87.3516, | |
| "train_samples_per_second": 3290.724, | |
| "train_steps_per_second": 13.165 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 1150, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 50, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 5, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 5 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 5294141566038528.0, | |
| "train_batch_size": 256, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |