| { | |
| "best_metric": 0.04861336573958397, | |
| "best_model_checkpoint": "/home/manjavacasema/data1/ing-forms/results-ing-single/checkpoint-1266", | |
| "epoch": 5.0, | |
| "global_step": 3165, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 3.907582938388626e-05, | |
| "loss": 0.2536, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 0.08746412396430969, | |
| "eval_runtime": 2.5343, | |
| "eval_samples_per_second": 221.758, | |
| "eval_steps_per_second": 28.016, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 3.315165876777251e-05, | |
| "loss": 0.0895, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 0.04861336573958397, | |
| "eval_runtime": 2.5502, | |
| "eval_samples_per_second": 220.371, | |
| "eval_steps_per_second": 27.84, | |
| "step": 1266 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 2.722748815165877e-05, | |
| "loss": 0.0324, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_loss": 0.06070622429251671, | |
| "eval_runtime": 2.5516, | |
| "eval_samples_per_second": 220.251, | |
| "eval_steps_per_second": 27.825, | |
| "step": 1899 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 2.1303317535545027e-05, | |
| "loss": 0.0173, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 1.537914691943128e-05, | |
| "loss": 0.0076, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_loss": 0.09658198803663254, | |
| "eval_runtime": 2.5515, | |
| "eval_samples_per_second": 220.262, | |
| "eval_steps_per_second": 27.827, | |
| "step": 2532 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "learning_rate": 9.454976303317536e-06, | |
| "loss": 0.0044, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_loss": 0.05032559856772423, | |
| "eval_runtime": 2.5533, | |
| "eval_samples_per_second": 220.103, | |
| "eval_steps_per_second": 27.807, | |
| "step": 3165 | |
| } | |
| ], | |
| "max_steps": 3798, | |
| "num_train_epochs": 6, | |
| "total_flos": 2464245296909520.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |