Multi-Domain-Expert-Learning/all_layers_all_domains
015
1{2 "best_metric": null,3 "best_model_checkpoint": null,4 "epoch": 0.9863013698630136,5 "global_step": 27,6 "is_hyper_param_search": false,7 "is_local_process_zero": true,8 "is_world_process_zero": true,9 "log_history": [10 {11 "epoch": 0.73,12 "learning_rate": 2.5925925925925925e-05,13 "loss": 2.2144,14 "step": 2015 },16 {17 "epoch": 0.99,18 "step": 27,19 "total_flos": 1.5437027289858048e+17,20 "train_loss": 2.189213894031666,21 "train_runtime": 168.693,22 "train_samples_per_second": 331.964,23 "train_steps_per_second": 0.1624 }25 ],26 "max_steps": 27,27 "num_train_epochs": 1,28 "total_flos": 1.5437027289858048e+17,29 "trial_name": null,30 "trial_params": null31}32 