| { | |
| "best_metric": 0.4375007450580597, | |
| "best_model_checkpoint": "./distilroberta-base-mrcp-glue-alcss/checkpoint-500", | |
| "epoch": 5.0, | |
| "eval_steps": 500, | |
| "global_step": 2295, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0893246187363834, | |
| "grad_norm": 6.300823211669922, | |
| "learning_rate": 3.910675381263617e-05, | |
| "loss": 0.5528, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.0893246187363834, | |
| "eval_accuracy": 0.8406862745098039, | |
| "eval_f1": 0.8849557522123894, | |
| "eval_loss": 0.4375007450580597, | |
| "eval_runtime": 1.4834, | |
| "eval_samples_per_second": 275.048, | |
| "eval_steps_per_second": 34.381, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.178649237472767, | |
| "grad_norm": 0.10454433411359787, | |
| "learning_rate": 2.8213507625272335e-05, | |
| "loss": 0.384, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.178649237472767, | |
| "eval_accuracy": 0.8161764705882353, | |
| "eval_f1": 0.8713550600343053, | |
| "eval_loss": 0.8148161768913269, | |
| "eval_runtime": 1.3452, | |
| "eval_samples_per_second": 303.291, | |
| "eval_steps_per_second": 37.911, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 3.2679738562091503, | |
| "grad_norm": 0.07294944673776627, | |
| "learning_rate": 1.7320261437908496e-05, | |
| "loss": 0.2497, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 3.2679738562091503, | |
| "eval_accuracy": 0.8308823529411765, | |
| "eval_f1": 0.880415944540728, | |
| "eval_loss": 0.8264561295509338, | |
| "eval_runtime": 1.6257, | |
| "eval_samples_per_second": 250.972, | |
| "eval_steps_per_second": 31.371, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 4.357298474945534, | |
| "grad_norm": 188.12437438964844, | |
| "learning_rate": 6.427015250544663e-06, | |
| "loss": 0.1549, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 4.357298474945534, | |
| "eval_accuracy": 0.8284313725490197, | |
| "eval_f1": 0.8745519713261649, | |
| "eval_loss": 0.8781077861785889, | |
| "eval_runtime": 1.5145, | |
| "eval_samples_per_second": 269.391, | |
| "eval_steps_per_second": 33.674, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "step": 2295, | |
| "total_flos": 342853570063632.0, | |
| "train_loss": 0.30509075671759045, | |
| "train_runtime": 357.8137, | |
| "train_samples_per_second": 51.256, | |
| "train_steps_per_second": 6.414 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 2295, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 5, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 342853570063632.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |