| { | |
| "best_metric": 3.5801477432250977, | |
| "best_model_checkpoint": "/users/zyong2/data/zyong2/bigscience/data/processed/024/bloom-350m_az_bitfit_100000samples_-1vocab_original-frozen/checkpoint-25000", | |
| "epoch": 2.1258503401360542, | |
| "global_step": 25000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9e-05, | |
| "loss": 4.2702, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 8e-05, | |
| "loss": 3.9326, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "eval_loss": 3.8811988830566406, | |
| "eval_runtime": 485.7533, | |
| "eval_samples_per_second": 9.486, | |
| "eval_steps_per_second": 4.743, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 7e-05, | |
| "loss": 3.7946, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 6e-05, | |
| "loss": 3.7099, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "eval_loss": 3.7074594497680664, | |
| "eval_runtime": 487.18, | |
| "eval_samples_per_second": 9.459, | |
| "eval_steps_per_second": 4.729, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 5e-05, | |
| "loss": 3.657, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 4e-05, | |
| "loss": 3.6085, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "eval_loss": 3.6284306049346924, | |
| "eval_runtime": 484.984, | |
| "eval_samples_per_second": 9.501, | |
| "eval_steps_per_second": 4.751, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 3e-05, | |
| "loss": 3.5817, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 2e-05, | |
| "loss": 3.5696, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "eval_loss": 3.5915658473968506, | |
| "eval_runtime": 485.8978, | |
| "eval_samples_per_second": 9.483, | |
| "eval_steps_per_second": 4.742, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 1e-05, | |
| "loss": 3.5612, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 0.0, | |
| "loss": 3.5533, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "eval_loss": 3.5801477432250977, | |
| "eval_runtime": 485.0169, | |
| "eval_samples_per_second": 9.501, | |
| "eval_steps_per_second": 4.75, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "step": 25000, | |
| "total_flos": 3.714827943936e+17, | |
| "train_loss": 3.7238579296875, | |
| "train_runtime": 52437.7197, | |
| "train_samples_per_second": 3.814, | |
| "train_steps_per_second": 0.477 | |
| } | |
| ], | |
| "max_steps": 25000, | |
| "num_train_epochs": 3, | |
| "total_flos": 3.714827943936e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |