| { | |
| "best_metric": 2.574218511581421, | |
| "best_model_checkpoint": "/mnt/bn/qingyi-bn-lq/llama/saved_chatglmwild_cn/checkpoint-200", | |
| "epoch": 1.0342598577892697, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 3.2000000000000003e-06, | |
| "loss": 3.3366, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 7e-06, | |
| "loss": 3.2662, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.1000000000000001e-05, | |
| "loss": 3.0126, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.5000000000000002e-05, | |
| "loss": 2.3476, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.9e-05, | |
| "loss": 1.2085, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.8951048951048952e-05, | |
| "loss": 0.3935, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.7552447552447553e-05, | |
| "loss": 0.1678, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.6153846153846154e-05, | |
| "loss": 0.0775, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.4755244755244758e-05, | |
| "loss": 0.0617, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 1.3356643356643357e-05, | |
| "loss": 0.0479, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "eval_loss": 2.574218511581421, | |
| "eval_runtime": 11.8812, | |
| "eval_samples_per_second": 168.332, | |
| "eval_steps_per_second": 2.693, | |
| "step": 200 | |
| } | |
| ], | |
| "max_steps": 386, | |
| "num_train_epochs": 2, | |
| "total_flos": 2.313659325706404e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |