| { | |
| "best_global_step": 588, | |
| "best_metric": 0.608646188850967, | |
| "best_model_checkpoint": "distilbert_emotion_sentiment_finetuned_cpu/checkpoint-588", | |
| "epoch": 4.0, | |
| "eval_steps": 500, | |
| "global_step": 588, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.3401360544217687, | |
| "grad_norm": 0.9348878860473633, | |
| "learning_rate": 2.75e-05, | |
| "loss": 0.489, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.6802721088435374, | |
| "grad_norm": 1.040361762046814, | |
| "learning_rate": 2.4948979591836734e-05, | |
| "loss": 0.3948, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy_subset": 0.30716723549488056, | |
| "eval_f1_samples": 0.43230944254835035, | |
| "eval_loss": 0.36293554306030273, | |
| "eval_runtime": 32.8378, | |
| "eval_samples_per_second": 8.923, | |
| "eval_steps_per_second": 0.579, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 1.0204081632653061, | |
| "grad_norm": 0.7896512746810913, | |
| "learning_rate": 2.239795918367347e-05, | |
| "loss": 0.3604, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.3605442176870748, | |
| "grad_norm": 0.8474708199501038, | |
| "learning_rate": 1.9846938775510205e-05, | |
| "loss": 0.3515, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.7006802721088436, | |
| "grad_norm": 1.1555696725845337, | |
| "learning_rate": 1.729591836734694e-05, | |
| "loss": 0.3323, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy_subset": 0.3003412969283277, | |
| "eval_f1_samples": 0.5386803185437997, | |
| "eval_loss": 0.33012497425079346, | |
| "eval_runtime": 32.6293, | |
| "eval_samples_per_second": 8.98, | |
| "eval_steps_per_second": 0.582, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 2.0408163265306123, | |
| "grad_norm": 0.9709883332252502, | |
| "learning_rate": 1.4744897959183674e-05, | |
| "loss": 0.3121, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 2.380952380952381, | |
| "grad_norm": 2.1932430267333984, | |
| "learning_rate": 1.2193877551020408e-05, | |
| "loss": 0.2793, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 2.7210884353741496, | |
| "grad_norm": 2.272308826446533, | |
| "learning_rate": 9.642857142857144e-06, | |
| "loss": 0.2628, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy_subset": 0.378839590443686, | |
| "eval_f1_samples": 0.58646188850967, | |
| "eval_loss": 0.3093891739845276, | |
| "eval_runtime": 31.9778, | |
| "eval_samples_per_second": 9.163, | |
| "eval_steps_per_second": 0.594, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 3.061224489795918, | |
| "grad_norm": 0.6931778788566589, | |
| "learning_rate": 7.0918367346938785e-06, | |
| "loss": 0.2488, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 3.4013605442176873, | |
| "grad_norm": 1.2945401668548584, | |
| "learning_rate": 4.540816326530613e-06, | |
| "loss": 0.2215, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 3.741496598639456, | |
| "grad_norm": 1.3985503911972046, | |
| "learning_rate": 1.989795918367347e-06, | |
| "loss": 0.236, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy_subset": 0.39590443686006827, | |
| "eval_f1_samples": 0.608646188850967, | |
| "eval_loss": 0.306503027677536, | |
| "eval_runtime": 33.5148, | |
| "eval_samples_per_second": 8.742, | |
| "eval_steps_per_second": 0.567, | |
| "step": 588 | |
| } | |
| ], | |
| "logging_steps": 50, | |
| "max_steps": 588, | |
| "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": 154873720286208.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |