{ "best_metric": 2.354668378829956, "best_model_checkpoint": "bert_tiny_lda_100_v1_stsb/checkpoint-207", "epoch": 14.0, "eval_steps": 500, "global_step": 322, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 1.0, "grad_norm": 23.892215728759766, "learning_rate": 0.00098, "loss": 4.3798, "step": 23 }, { "epoch": 1.0, "eval_combined_score": NaN, "eval_loss": 3.4025566577911377, "eval_pearson": NaN, "eval_runtime": 0.4491, "eval_samples_per_second": 3339.749, "eval_spearmanr": NaN, "eval_steps_per_second": 13.359, "step": 23 }, { "epoch": 2.0, "grad_norm": 5.764803409576416, "learning_rate": 0.00096, "loss": 2.2683, "step": 46 }, { "epoch": 2.0, "eval_combined_score": NaN, "eval_loss": 2.737987518310547, "eval_pearson": NaN, "eval_runtime": 0.4468, "eval_samples_per_second": 3357.434, "eval_spearmanr": NaN, "eval_steps_per_second": 13.43, "step": 46 }, { "epoch": 3.0, "grad_norm": 16.960420608520508, "learning_rate": 0.00094, "loss": 2.2687, "step": 69 }, { "epoch": 3.0, "eval_combined_score": NaN, "eval_loss": 2.7826263904571533, "eval_pearson": NaN, "eval_runtime": 0.4471, "eval_samples_per_second": 3354.985, "eval_spearmanr": NaN, "eval_steps_per_second": 13.42, "step": 69 }, { "epoch": 4.0, "grad_norm": 4.696272373199463, "learning_rate": 0.00092, "loss": 2.1923, "step": 92 }, { "epoch": 4.0, "eval_combined_score": NaN, "eval_loss": 2.6349358558654785, "eval_pearson": NaN, "eval_runtime": 0.4533, "eval_samples_per_second": 3309.076, "eval_spearmanr": NaN, "eval_steps_per_second": 13.236, "step": 92 }, { "epoch": 5.0, "grad_norm": 15.793731689453125, "learning_rate": 0.0009000000000000001, "loss": 2.2028, "step": 115 }, { "epoch": 5.0, "eval_combined_score": -0.011056950325972904, "eval_loss": 2.632760524749756, "eval_pearson": -0.011258424775794945, "eval_runtime": 0.4508, "eval_samples_per_second": 3327.54, "eval_spearmanr": -0.010855475876150862, "eval_steps_per_second": 13.31, "step": 115 }, { "epoch": 6.0, "grad_norm": 3.9431076049804688, "learning_rate": 0.00088, "loss": 2.1848, "step": 138 }, { "epoch": 6.0, "eval_combined_score": -0.023699091607435406, "eval_loss": 2.380463123321533, "eval_pearson": -0.02456100685625376, "eval_runtime": 0.447, "eval_samples_per_second": 3355.397, "eval_spearmanr": -0.02283717635861705, "eval_steps_per_second": 13.422, "step": 138 }, { "epoch": 7.0, "grad_norm": 18.49576187133789, "learning_rate": 0.00086, "loss": 2.194, "step": 161 }, { "epoch": 7.0, "eval_combined_score": NaN, "eval_loss": 2.8532462120056152, "eval_pearson": NaN, "eval_runtime": 0.4543, "eval_samples_per_second": 3301.537, "eval_spearmanr": NaN, "eval_steps_per_second": 13.206, "step": 161 }, { "epoch": 8.0, "grad_norm": 5.551238536834717, "learning_rate": 0.00084, "loss": 2.1835, "step": 184 }, { "epoch": 8.0, "eval_combined_score": NaN, "eval_loss": 2.561283588409424, "eval_pearson": NaN, "eval_runtime": 0.4562, "eval_samples_per_second": 3288.233, "eval_spearmanr": NaN, "eval_steps_per_second": 13.153, "step": 184 }, { "epoch": 9.0, "grad_norm": 13.460580825805664, "learning_rate": 0.00082, "loss": 2.1788, "step": 207 }, { "epoch": 9.0, "eval_combined_score": NaN, "eval_loss": 2.354668378829956, "eval_pearson": NaN, "eval_runtime": 0.4458, "eval_samples_per_second": 3364.865, "eval_spearmanr": NaN, "eval_steps_per_second": 13.459, "step": 207 }, { "epoch": 10.0, "grad_norm": 7.859861850738525, "learning_rate": 0.0008, "loss": 2.1688, "step": 230 }, { "epoch": 10.0, "eval_combined_score": NaN, "eval_loss": 2.424130916595459, "eval_pearson": NaN, "eval_runtime": 0.4484, "eval_samples_per_second": 3345.222, "eval_spearmanr": NaN, "eval_steps_per_second": 13.381, "step": 230 }, { "epoch": 11.0, "grad_norm": 11.691555976867676, "learning_rate": 0.0007800000000000001, "loss": 2.1963, "step": 253 }, { "epoch": 11.0, "eval_combined_score": NaN, "eval_loss": 2.674691915512085, "eval_pearson": NaN, "eval_runtime": 0.4529, "eval_samples_per_second": 3311.671, "eval_spearmanr": NaN, "eval_steps_per_second": 13.247, "step": 253 }, { "epoch": 12.0, "grad_norm": 4.0701680183410645, "learning_rate": 0.00076, "loss": 2.186, "step": 276 }, { "epoch": 12.0, "eval_combined_score": NaN, "eval_loss": 2.4511916637420654, "eval_pearson": NaN, "eval_runtime": 0.4443, "eval_samples_per_second": 3376.464, "eval_spearmanr": NaN, "eval_steps_per_second": 13.506, "step": 276 }, { "epoch": 13.0, "grad_norm": 3.6423840522766113, "learning_rate": 0.00074, "loss": 2.186, "step": 299 }, { "epoch": 13.0, "eval_combined_score": NaN, "eval_loss": 2.465453863143921, "eval_pearson": NaN, "eval_runtime": 0.4517, "eval_samples_per_second": 3320.427, "eval_spearmanr": NaN, "eval_steps_per_second": 13.282, "step": 299 }, { "epoch": 14.0, "grad_norm": 5.533896446228027, "learning_rate": 0.0007199999999999999, "loss": 2.2089, "step": 322 }, { "epoch": 14.0, "eval_combined_score": NaN, "eval_loss": 2.737987518310547, "eval_pearson": NaN, "eval_runtime": 0.4475, "eval_samples_per_second": 3352.138, "eval_spearmanr": NaN, "eval_steps_per_second": 13.409, "step": 322 }, { "epoch": 14.0, "step": 322, "total_flos": 2110567487529984.0, "train_loss": 2.3570685771681505, "train_runtime": 58.2449, "train_samples_per_second": 4935.193, "train_steps_per_second": 19.744 } ], "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": 2110567487529984.0, "train_batch_size": 256, "trial_name": null, "trial_params": null }