| { | |
| "best_metric": 0.5776590704917908, | |
| "best_model_checkpoint": "BERT-WMM/run-13/checkpoint-534", | |
| "epoch": 5.0, | |
| "eval_steps": 500, | |
| "global_step": 1335, | |
| "is_hyper_param_search": true, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.7577464788732394, | |
| "eval_loss": 0.6081846952438354, | |
| "eval_runtime": 2.4174, | |
| "eval_samples_per_second": 881.101, | |
| "eval_steps_per_second": 55.431, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "grad_norm": 9.532360076904297, | |
| "learning_rate": 9.65168237740807e-06, | |
| "loss": 0.6057, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.7657276995305164, | |
| "eval_loss": 0.5776590704917908, | |
| "eval_runtime": 2.4511, | |
| "eval_samples_per_second": 868.982, | |
| "eval_steps_per_second": 54.668, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.7690140845070422, | |
| "eval_loss": 0.6647635102272034, | |
| "eval_runtime": 2.4588, | |
| "eval_samples_per_second": 866.264, | |
| "eval_steps_per_second": 54.497, | |
| "step": 801 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "grad_norm": 26.327552795410156, | |
| "learning_rate": 3.872231852014017e-06, | |
| "loss": 0.2807, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.7615023474178404, | |
| "eval_loss": 0.758091151714325, | |
| "eval_runtime": 2.483, | |
| "eval_samples_per_second": 857.84, | |
| "eval_steps_per_second": 53.967, | |
| "step": 1068 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.7643192488262911, | |
| "eval_loss": 0.8205846548080444, | |
| "eval_runtime": 2.4655, | |
| "eval_samples_per_second": 863.914, | |
| "eval_steps_per_second": 54.35, | |
| "step": 1335 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 1335, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 5, | |
| "save_steps": 500, | |
| "total_flos": 697511614823352.0, | |
| "train_batch_size": 32, | |
| "trial_name": null, | |
| "trial_params": { | |
| "num_train_epochs": 5 | |
| } | |
| } | |