| { |
| "best_metric": 0.7326732673267325, |
| "best_model_checkpoint": "final_models/structroberta_sx2_final/finetune/qnli/checkpoint-2000", |
| "epoch": 9.836065573770492, |
| "global_step": 3600, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 1.09, |
| "eval_accuracy": 0.6736658215522766, |
| "eval_f1": 0.681468830059778, |
| "eval_loss": 0.6080768704414368, |
| "eval_mcc": 0.34894047189705113, |
| "eval_runtime": 4.8444, |
| "eval_samples_per_second": 471.881, |
| "eval_steps_per_second": 59.037, |
| "step": 400 |
| }, |
| { |
| "epoch": 1.37, |
| "learning_rate": 4.316939890710383e-05, |
| "loss": 0.6434, |
| "step": 500 |
| }, |
| { |
| "epoch": 2.19, |
| "eval_accuracy": 0.7060367465019226, |
| "eval_f1": 0.6759884281581485, |
| "eval_loss": 0.5899884700775146, |
| "eval_mcc": 0.4424515611165257, |
| "eval_runtime": 4.7984, |
| "eval_samples_per_second": 476.406, |
| "eval_steps_per_second": 59.603, |
| "step": 800 |
| }, |
| { |
| "epoch": 2.73, |
| "learning_rate": 3.633879781420765e-05, |
| "loss": 0.5162, |
| "step": 1000 |
| }, |
| { |
| "epoch": 3.28, |
| "eval_accuracy": 0.7314085960388184, |
| "eval_f1": 0.718348623853211, |
| "eval_loss": 0.5835453271865845, |
| "eval_mcc": 0.47941652405255153, |
| "eval_runtime": 4.8413, |
| "eval_samples_per_second": 472.185, |
| "eval_steps_per_second": 59.075, |
| "step": 1200 |
| }, |
| { |
| "epoch": 4.1, |
| "learning_rate": 2.9508196721311478e-05, |
| "loss": 0.3707, |
| "step": 1500 |
| }, |
| { |
| "epoch": 4.37, |
| "eval_accuracy": 0.7397200465202332, |
| "eval_f1": 0.7318611987381702, |
| "eval_loss": 0.6178868412971497, |
| "eval_mcc": 0.4917193731545861, |
| "eval_runtime": 4.802, |
| "eval_samples_per_second": 476.055, |
| "eval_steps_per_second": 59.559, |
| "step": 1600 |
| }, |
| { |
| "epoch": 5.46, |
| "learning_rate": 2.2677595628415303e-05, |
| "loss": 0.255, |
| "step": 2000 |
| }, |
| { |
| "epoch": 5.46, |
| "eval_accuracy": 0.7401574850082397, |
| "eval_f1": 0.7326732673267325, |
| "eval_loss": 0.7336315512657166, |
| "eval_mcc": 0.4922779179551397, |
| "eval_runtime": 4.8165, |
| "eval_samples_per_second": 474.623, |
| "eval_steps_per_second": 59.38, |
| "step": 2000 |
| }, |
| { |
| "epoch": 6.56, |
| "eval_accuracy": 0.721784770488739, |
| "eval_f1": 0.6956937799043061, |
| "eval_loss": 0.8675853610038757, |
| "eval_mcc": 0.4724210146794968, |
| "eval_runtime": 4.798, |
| "eval_samples_per_second": 476.453, |
| "eval_steps_per_second": 59.609, |
| "step": 2400 |
| }, |
| { |
| "epoch": 6.83, |
| "learning_rate": 1.5846994535519128e-05, |
| "loss": 0.1778, |
| "step": 2500 |
| }, |
| { |
| "epoch": 7.65, |
| "eval_accuracy": 0.7335957884788513, |
| "eval_f1": 0.7255520504731862, |
| "eval_loss": 0.9163379669189453, |
| "eval_mcc": 0.4793547603035812, |
| "eval_runtime": 4.795, |
| "eval_samples_per_second": 476.742, |
| "eval_steps_per_second": 59.645, |
| "step": 2800 |
| }, |
| { |
| "epoch": 8.2, |
| "learning_rate": 9.016393442622952e-06, |
| "loss": 0.1145, |
| "step": 3000 |
| }, |
| { |
| "epoch": 8.74, |
| "eval_accuracy": 0.7327209115028381, |
| "eval_f1": 0.7206218564243255, |
| "eval_loss": 0.9837310314178467, |
| "eval_mcc": 0.48122668219972387, |
| "eval_runtime": 4.8087, |
| "eval_samples_per_second": 475.385, |
| "eval_steps_per_second": 59.475, |
| "step": 3200 |
| }, |
| { |
| "epoch": 9.56, |
| "learning_rate": 2.185792349726776e-06, |
| "loss": 0.0804, |
| "step": 3500 |
| }, |
| { |
| "epoch": 9.84, |
| "eval_accuracy": 0.7314085960388184, |
| "eval_f1": 0.7188644688644688, |
| "eval_loss": 1.1153168678283691, |
| "eval_mcc": 0.47892849292999046, |
| "eval_runtime": 4.8458, |
| "eval_samples_per_second": 471.753, |
| "eval_steps_per_second": 59.021, |
| "step": 3600 |
| }, |
| { |
| "epoch": 9.84, |
| "step": 3600, |
| "total_flos": 3.957043224322714e+16, |
| "train_loss": 0.30156181520885894, |
| "train_runtime": 2024.4492, |
| "train_samples_per_second": 216.933, |
| "train_steps_per_second": 1.808 |
| } |
| ], |
| "max_steps": 3660, |
| "num_train_epochs": 10, |
| "total_flos": 3.957043224322714e+16, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
