| { | |
| "best_metric": 0.8860787153244019, | |
| "best_model_checkpoint": "sequence_classification\\checkpoint-1238", | |
| "epoch": 4.0, | |
| "eval_steps": 500, | |
| "global_step": 2476, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.8077544426494345, | |
| "grad_norm": 15.20165729522705, | |
| "learning_rate": 7.980613893376414e-05, | |
| "loss": 1.3148, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 0.9956253170967102, | |
| "eval_pearson": 0.7673949737715109, | |
| "eval_runtime": 593.2409, | |
| "eval_samples_per_second": 8.337, | |
| "eval_steps_per_second": 0.261, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 1.615508885298869, | |
| "grad_norm": 19.32335662841797, | |
| "learning_rate": 5.9612277867528274e-05, | |
| "loss": 0.9627, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 0.8860787153244019, | |
| "eval_pearson": 0.7987432834261069, | |
| "eval_runtime": 588.6019, | |
| "eval_samples_per_second": 8.403, | |
| "eval_steps_per_second": 0.263, | |
| "step": 1238 | |
| }, | |
| { | |
| "epoch": 2.4232633279483036, | |
| "grad_norm": 10.007705688476562, | |
| "learning_rate": 3.941841680129241e-05, | |
| "loss": 0.8641, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_loss": 0.9137663245201111, | |
| "eval_pearson": 0.7964932034547004, | |
| "eval_runtime": 589.478, | |
| "eval_samples_per_second": 8.39, | |
| "eval_steps_per_second": 0.263, | |
| "step": 1857 | |
| }, | |
| { | |
| "epoch": 3.231017770597738, | |
| "grad_norm": 4.174856662750244, | |
| "learning_rate": 1.9224555735056543e-05, | |
| "loss": 0.7651, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_loss": 0.8885385990142822, | |
| "eval_pearson": 0.8035657020343545, | |
| "eval_runtime": 588.4581, | |
| "eval_samples_per_second": 8.405, | |
| "eval_steps_per_second": 0.263, | |
| "step": 2476 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "step": 2476, | |
| "total_flos": 2.3738772810018816e+17, | |
| "train_loss": 0.9284627911347372, | |
| "train_runtime": 31889.2207, | |
| "train_samples_per_second": 2.481, | |
| "train_steps_per_second": 0.078 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 2476, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 4, | |
| "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": 2.3738772810018816e+17, | |
| "train_batch_size": 32, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |