| { | |
| "best_metric": 0.22905202209949493, | |
| "best_model_checkpoint": "mobilebert_sa_GLUE_Experiment_logit_kd_pretrain_mrpc/checkpoint-435", | |
| "epoch": 20.0, | |
| "global_step": 580, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.9e-05, | |
| "loss": 0.536, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.7279411764705882, | |
| "eval_combined_score": 0.7781900627329756, | |
| "eval_f1": 0.8284389489953632, | |
| "eval_loss": 0.413361519575119, | |
| "eval_runtime": 0.8016, | |
| "eval_samples_per_second": 508.993, | |
| "eval_steps_per_second": 4.99, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.8e-05, | |
| "loss": 0.3419, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.8284313725490197, | |
| "eval_combined_score": 0.8542841794251947, | |
| "eval_f1": 0.8801369863013697, | |
| "eval_loss": 0.30050206184387207, | |
| "eval_runtime": 0.7827, | |
| "eval_samples_per_second": 521.241, | |
| "eval_steps_per_second": 5.11, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 4.7e-05, | |
| "loss": 0.2413, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.8235294117647058, | |
| "eval_combined_score": 0.8507477567298105, | |
| "eval_f1": 0.8779661016949152, | |
| "eval_loss": 0.2707367241382599, | |
| "eval_runtime": 0.8474, | |
| "eval_samples_per_second": 481.468, | |
| "eval_steps_per_second": 4.72, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 4.600000000000001e-05, | |
| "loss": 0.1852, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.8284313725490197, | |
| "eval_combined_score": 0.8560761513907889, | |
| "eval_f1": 0.883720930232558, | |
| "eval_loss": 0.3246958255767822, | |
| "eval_runtime": 0.7815, | |
| "eval_samples_per_second": 522.064, | |
| "eval_steps_per_second": 5.118, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 4.5e-05, | |
| "loss": 0.1524, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.8431372549019608, | |
| "eval_combined_score": 0.8665858095815646, | |
| "eval_f1": 0.8900343642611684, | |
| "eval_loss": 0.2855992019176483, | |
| "eval_runtime": 0.7813, | |
| "eval_samples_per_second": 522.186, | |
| "eval_steps_per_second": 5.119, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 4.4000000000000006e-05, | |
| "loss": 0.1297, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.8455882352941176, | |
| "eval_combined_score": 0.8702064715702642, | |
| "eval_f1": 0.8948247078464107, | |
| "eval_loss": 0.29985857009887695, | |
| "eval_runtime": 0.7783, | |
| "eval_samples_per_second": 524.229, | |
| "eval_steps_per_second": 5.139, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 4.3e-05, | |
| "loss": 0.1219, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.8529411764705882, | |
| "eval_combined_score": 0.8757949125596185, | |
| "eval_f1": 0.8986486486486487, | |
| "eval_loss": 0.27967169880867004, | |
| "eval_runtime": 0.7807, | |
| "eval_samples_per_second": 522.641, | |
| "eval_steps_per_second": 5.124, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "learning_rate": 4.2e-05, | |
| "loss": 0.1141, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.8602941176470589, | |
| "eval_combined_score": 0.8804088389282414, | |
| "eval_f1": 0.900523560209424, | |
| "eval_loss": 0.2462492138147354, | |
| "eval_runtime": 0.7782, | |
| "eval_samples_per_second": 524.315, | |
| "eval_steps_per_second": 5.14, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "learning_rate": 4.1e-05, | |
| "loss": 0.1127, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.8578431372549019, | |
| "eval_combined_score": 0.8780443756449948, | |
| "eval_f1": 0.8982456140350877, | |
| "eval_loss": 0.25573617219924927, | |
| "eval_runtime": 0.7773, | |
| "eval_samples_per_second": 524.89, | |
| "eval_steps_per_second": 5.146, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "learning_rate": 4e-05, | |
| "loss": 0.1091, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.8480392156862745, | |
| "eval_combined_score": 0.8723529411764706, | |
| "eval_f1": 0.8966666666666667, | |
| "eval_loss": 0.2853253185749054, | |
| "eval_runtime": 0.7789, | |
| "eval_samples_per_second": 523.847, | |
| "eval_steps_per_second": 5.136, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "learning_rate": 3.9000000000000006e-05, | |
| "loss": 0.1007, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_accuracy": 0.8553921568627451, | |
| "eval_combined_score": 0.8767461647871584, | |
| "eval_f1": 0.8981001727115717, | |
| "eval_loss": 0.2471570074558258, | |
| "eval_runtime": 0.7834, | |
| "eval_samples_per_second": 520.839, | |
| "eval_steps_per_second": 5.106, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "learning_rate": 3.8e-05, | |
| "loss": 0.0979, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_accuracy": 0.8504901960784313, | |
| "eval_combined_score": 0.8727494009652054, | |
| "eval_f1": 0.8950086058519794, | |
| "eval_loss": 0.24307508766651154, | |
| "eval_runtime": 0.7814, | |
| "eval_samples_per_second": 522.12, | |
| "eval_steps_per_second": 5.119, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "learning_rate": 3.7e-05, | |
| "loss": 0.0954, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_accuracy": 0.8578431372549019, | |
| "eval_combined_score": 0.8792640343808756, | |
| "eval_f1": 0.9006849315068494, | |
| "eval_loss": 0.24562659859657288, | |
| "eval_runtime": 0.7831, | |
| "eval_samples_per_second": 521.003, | |
| "eval_steps_per_second": 5.108, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "learning_rate": 3.6e-05, | |
| "loss": 0.0946, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_accuracy": 0.8578431372549019, | |
| "eval_combined_score": 0.8797690262545697, | |
| "eval_f1": 0.9016949152542373, | |
| "eval_loss": 0.25261467695236206, | |
| "eval_runtime": 0.7806, | |
| "eval_samples_per_second": 522.682, | |
| "eval_steps_per_second": 5.124, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "learning_rate": 3.5e-05, | |
| "loss": 0.0946, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_accuracy": 0.8578431372549019, | |
| "eval_combined_score": 0.8785743464052287, | |
| "eval_f1": 0.8993055555555555, | |
| "eval_loss": 0.22905202209949493, | |
| "eval_runtime": 0.78, | |
| "eval_samples_per_second": 523.081, | |
| "eval_steps_per_second": 5.128, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "learning_rate": 3.4000000000000007e-05, | |
| "loss": 0.0938, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_accuracy": 0.8602941176470589, | |
| "eval_combined_score": 0.8815950997093898, | |
| "eval_f1": 0.9028960817717206, | |
| "eval_loss": 0.2451842576265335, | |
| "eval_runtime": 0.7798, | |
| "eval_samples_per_second": 523.204, | |
| "eval_steps_per_second": 5.129, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "learning_rate": 3.3e-05, | |
| "loss": 0.0919, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "eval_accuracy": 0.8651960784313726, | |
| "eval_combined_score": 0.8851023570049782, | |
| "eval_f1": 0.9050086355785838, | |
| "eval_loss": 0.23652492463588715, | |
| "eval_runtime": 0.7816, | |
| "eval_samples_per_second": 522.014, | |
| "eval_steps_per_second": 5.118, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "learning_rate": 3.2000000000000005e-05, | |
| "loss": 0.0916, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "eval_accuracy": 0.8651960784313726, | |
| "eval_combined_score": 0.8855894922071392, | |
| "eval_f1": 0.905982905982906, | |
| "eval_loss": 0.23631499707698822, | |
| "eval_runtime": 0.781, | |
| "eval_samples_per_second": 522.414, | |
| "eval_steps_per_second": 5.122, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "learning_rate": 3.1e-05, | |
| "loss": 0.0915, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "eval_accuracy": 0.8651960784313726, | |
| "eval_combined_score": 0.8857496576143234, | |
| "eval_f1": 0.9063032367972743, | |
| "eval_loss": 0.24315035343170166, | |
| "eval_runtime": 0.78, | |
| "eval_samples_per_second": 523.058, | |
| "eval_steps_per_second": 5.128, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "learning_rate": 3e-05, | |
| "loss": 0.0905, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "eval_accuracy": 0.8651960784313726, | |
| "eval_combined_score": 0.8854282278949316, | |
| "eval_f1": 0.9056603773584906, | |
| "eval_loss": 0.2296983003616333, | |
| "eval_runtime": 0.7873, | |
| "eval_samples_per_second": 518.229, | |
| "eval_steps_per_second": 5.081, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "step": 580, | |
| "total_flos": 2329744947281920.0, | |
| "train_loss": 0.14934097281817732, | |
| "train_runtime": 450.6297, | |
| "train_samples_per_second": 406.986, | |
| "train_steps_per_second": 3.218 | |
| } | |
| ], | |
| "max_steps": 1450, | |
| "num_train_epochs": 50, | |
| "total_flos": 2329744947281920.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |