| { |
| "best_metric": 0.49181562662124634, |
| "best_model_checkpoint": "output_pipe/4/origin/checkpoint-400", |
| "epoch": 4.0, |
| "eval_steps": 200, |
| "global_step": 944, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.423728813559322, |
| "grad_norm": 1.2542049884796143, |
| "learning_rate": 2.8389261744966442e-05, |
| "loss": 0.6887, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.847457627118644, |
| "grad_norm": 3.1040308475494385, |
| "learning_rate": 2.503355704697987e-05, |
| "loss": 0.5755, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.847457627118644, |
| "eval_accuracy": 0.740308019118428, |
| "eval_f1": 0.7395134139189725, |
| "eval_loss": 0.5309526324272156, |
| "eval_matthews_correlation": 0.4922795965623092, |
| "eval_precision": 0.7487400222919133, |
| "eval_recall": 0.7435667559231265, |
| "eval_runtime": 0.2771, |
| "eval_samples_per_second": 6795.975, |
| "eval_steps_per_second": 108.274, |
| "step": 200 |
| }, |
| { |
| "epoch": 1.271186440677966, |
| "grad_norm": 2.9270331859588623, |
| "learning_rate": 2.167785234899329e-05, |
| "loss": 0.512, |
| "step": 300 |
| }, |
| { |
| "epoch": 1.694915254237288, |
| "grad_norm": 7.294169902801514, |
| "learning_rate": 1.832214765100671e-05, |
| "loss": 0.4662, |
| "step": 400 |
| }, |
| { |
| "epoch": 1.694915254237288, |
| "eval_accuracy": 0.7631439192777483, |
| "eval_f1": 0.7619919694558183, |
| "eval_loss": 0.49181562662124634, |
| "eval_matthews_correlation": 0.5254431311127685, |
| "eval_precision": 0.7638565365838093, |
| "eval_recall": 0.7615914765730039, |
| "eval_runtime": 0.2685, |
| "eval_samples_per_second": 7014.179, |
| "eval_steps_per_second": 111.75, |
| "step": 400 |
| }, |
| { |
| "epoch": 2.1186440677966103, |
| "grad_norm": 5.567413330078125, |
| "learning_rate": 1.4966442953020135e-05, |
| "loss": 0.4446, |
| "step": 500 |
| }, |
| { |
| "epoch": 2.542372881355932, |
| "grad_norm": 4.504443645477295, |
| "learning_rate": 1.1610738255033558e-05, |
| "loss": 0.3611, |
| "step": 600 |
| }, |
| { |
| "epoch": 2.542372881355932, |
| "eval_accuracy": 0.758364312267658, |
| "eval_f1": 0.7559833932054777, |
| "eval_loss": 0.5178119540214539, |
| "eval_matthews_correlation": 0.5175228785946929, |
| "eval_precision": 0.7618896341971502, |
| "eval_recall": 0.7556706097684982, |
| "eval_runtime": 0.274, |
| "eval_samples_per_second": 6872.743, |
| "eval_steps_per_second": 109.497, |
| "step": 600 |
| }, |
| { |
| "epoch": 2.9661016949152543, |
| "grad_norm": 5.900634765625, |
| "learning_rate": 8.25503355704698e-06, |
| "loss": 0.3646, |
| "step": 700 |
| }, |
| { |
| "epoch": 3.389830508474576, |
| "grad_norm": 6.066167831420898, |
| "learning_rate": 4.899328859060402e-06, |
| "loss": 0.2931, |
| "step": 800 |
| }, |
| { |
| "epoch": 3.389830508474576, |
| "eval_accuracy": 0.7695167286245354, |
| "eval_f1": 0.7686836950737632, |
| "eval_loss": 0.5691679120063782, |
| "eval_matthews_correlation": 0.5380657164892454, |
| "eval_precision": 0.7697224525023564, |
| "eval_recall": 0.768345027059395, |
| "eval_runtime": 0.2648, |
| "eval_samples_per_second": 7111.768, |
| "eval_steps_per_second": 113.305, |
| "step": 800 |
| }, |
| { |
| "epoch": 3.8135593220338984, |
| "grad_norm": 5.857461452484131, |
| "learning_rate": 1.5436241610738255e-06, |
| "loss": 0.2891, |
| "step": 900 |
| }, |
| { |
| "epoch": 4.0, |
| "step": 944, |
| "total_flos": 1235615997853440.0, |
| "train_loss": 0.43632790597818666, |
| "train_runtime": 41.6247, |
| "train_samples_per_second": 1447.603, |
| "train_steps_per_second": 22.679 |
| } |
| ], |
| "logging_steps": 100, |
| "max_steps": 944, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 4, |
| "save_steps": 200, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": true |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 1235615997853440.0, |
| "train_batch_size": 64, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|