| { | |
| "best_metric": 2.2437143325805664, | |
| "best_model_checkpoint": "/users/zyong2/data/zyong2/bigscience/data/processed/024/bloom-350m_az_fish_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": 2.9924, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 8e-05, | |
| "loss": 2.5627, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "eval_loss": 2.5240488052368164, | |
| "eval_runtime": 484.0815, | |
| "eval_samples_per_second": 9.519, | |
| "eval_steps_per_second": 4.76, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 7e-05, | |
| "loss": 2.4428, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 6e-05, | |
| "loss": 2.3709, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "eval_loss": 2.3797976970672607, | |
| "eval_runtime": 484.2796, | |
| "eval_samples_per_second": 9.515, | |
| "eval_steps_per_second": 4.758, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 5e-05, | |
| "loss": 2.3225, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 4e-05, | |
| "loss": 2.2726, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "eval_loss": 2.3043975830078125, | |
| "eval_runtime": 483.9345, | |
| "eval_samples_per_second": 9.522, | |
| "eval_steps_per_second": 4.761, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 3e-05, | |
| "loss": 2.2479, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 2e-05, | |
| "loss": 2.2341, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "eval_loss": 2.264699935913086, | |
| "eval_runtime": 483.6102, | |
| "eval_samples_per_second": 9.528, | |
| "eval_steps_per_second": 4.764, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 1e-05, | |
| "loss": 2.2163, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 0.0, | |
| "loss": 2.2019, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "eval_loss": 2.2437143325805664, | |
| "eval_runtime": 482.513, | |
| "eval_samples_per_second": 9.55, | |
| "eval_steps_per_second": 4.775, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "step": 25000, | |
| "total_flos": 3.714827943936e+17, | |
| "train_loss": 2.3864154296875, | |
| "train_runtime": 60699.5161, | |
| "train_samples_per_second": 3.295, | |
| "train_steps_per_second": 0.412 | |
| } | |
| ], | |
| "max_steps": 25000, | |
| "num_train_epochs": 3, | |
| "total_flos": 3.714827943936e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |