| { | |
| "best_global_step": 102, | |
| "best_metric": 0.6126338243484497, | |
| "best_model_checkpoint": "tiny_bert_rand_10_v1_cola/checkpoint-102", | |
| "epoch": 8.0, | |
| "eval_steps": 500, | |
| "global_step": 272, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 0.7733844518661499, | |
| "learning_rate": 4.9e-05, | |
| "loss": 0.6189, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.6912751793861389, | |
| "eval_loss": 0.6155545711517334, | |
| "eval_matthews_correlation": 0.0, | |
| "eval_runtime": 0.3291, | |
| "eval_samples_per_second": 3169.627, | |
| "eval_steps_per_second": 15.195, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 1.0032224655151367, | |
| "learning_rate": 4.8e-05, | |
| "loss": 0.6016, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.6912751793861389, | |
| "eval_loss": 0.6128833293914795, | |
| "eval_matthews_correlation": 0.0, | |
| "eval_runtime": 0.3939, | |
| "eval_samples_per_second": 2647.727, | |
| "eval_steps_per_second": 12.693, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "grad_norm": 1.3006558418273926, | |
| "learning_rate": 4.7e-05, | |
| "loss": 0.5821, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.6912751793861389, | |
| "eval_loss": 0.6126338243484497, | |
| "eval_matthews_correlation": 0.03640174058932784, | |
| "eval_runtime": 0.3141, | |
| "eval_samples_per_second": 3320.497, | |
| "eval_steps_per_second": 15.918, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "grad_norm": 2.361246347427368, | |
| "learning_rate": 4.600000000000001e-05, | |
| "loss": 0.5453, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.6797698736190796, | |
| "eval_loss": 0.6591368317604065, | |
| "eval_matthews_correlation": 0.08682477354041367, | |
| "eval_runtime": 0.3481, | |
| "eval_samples_per_second": 2996.167, | |
| "eval_steps_per_second": 14.363, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "grad_norm": 1.5645312070846558, | |
| "learning_rate": 4.5e-05, | |
| "loss": 0.4994, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.6701821684837341, | |
| "eval_loss": 0.6550384759902954, | |
| "eval_matthews_correlation": 0.10962781931858787, | |
| "eval_runtime": 0.338, | |
| "eval_samples_per_second": 3085.914, | |
| "eval_steps_per_second": 14.793, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "grad_norm": 2.503261089324951, | |
| "learning_rate": 4.4000000000000006e-05, | |
| "loss": 0.4658, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.6462128758430481, | |
| "eval_loss": 0.7169947624206543, | |
| "eval_matthews_correlation": 0.10178831079319654, | |
| "eval_runtime": 0.3557, | |
| "eval_samples_per_second": 2932.551, | |
| "eval_steps_per_second": 14.058, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "grad_norm": 3.4820024967193604, | |
| "learning_rate": 4.3e-05, | |
| "loss": 0.4329, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.6289549469947815, | |
| "eval_loss": 0.7630059123039246, | |
| "eval_matthews_correlation": 0.08670797170558309, | |
| "eval_runtime": 0.3292, | |
| "eval_samples_per_second": 3167.835, | |
| "eval_steps_per_second": 15.186, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "grad_norm": 3.2115254402160645, | |
| "learning_rate": 4.2e-05, | |
| "loss": 0.4002, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.6356663703918457, | |
| "eval_loss": 0.8304768204689026, | |
| "eval_matthews_correlation": 0.07336935024063443, | |
| "eval_runtime": 0.3217, | |
| "eval_samples_per_second": 3242.475, | |
| "eval_steps_per_second": 15.544, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "step": 272, | |
| "total_flos": 1793902531485696.0, | |
| "train_loss": 0.518292974023258, | |
| "train_runtime": 46.429, | |
| "train_samples_per_second": 9208.683, | |
| "train_steps_per_second": 36.615 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 1700, | |
| "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": 1793902531485696.0, | |
| "train_batch_size": 256, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |