p3ngdump/kcbert-base-hate-withunsmile
111
1{2 "best_metric": null,3 "best_model_checkpoint": null,4 "epoch": 4.888268156424581,5 "eval_steps": 500,6 "global_step": 3500,7 "is_hyper_param_search": false,8 "is_local_process_zero": true,9 "is_world_process_zero": true,10 "log_history": [11 {12 "epoch": 0.7,13 "grad_norm": 6.08462381362915,14 "learning_rate": 1.7206703910614527e-05,15 "loss": 0.3865,16 "step": 50017 },18 {19 "epoch": 1.4,20 "grad_norm": 5.282430648803711,21 "learning_rate": 1.4413407821229052e-05,22 "loss": 0.2899,23 "step": 100024 },25 {26 "epoch": 2.09,27 "grad_norm": 4.065394878387451,28 "learning_rate": 1.1620111731843577e-05,29 "loss": 0.2214,30 "step": 150031 },32 {33 "epoch": 2.79,34 "grad_norm": 8.418371200561523,35 "learning_rate": 8.826815642458101e-06,36 "loss": 0.1093,37 "step": 200038 },39 {40 "epoch": 3.49,41 "grad_norm": 20.9011173248291,42 "learning_rate": 6.033519553072626e-06,43 "loss": 0.0653,44 "step": 250045 },46 {47 "epoch": 4.19,48 "grad_norm": 4.803398132324219,49 "learning_rate": 3.240223463687151e-06,50 "loss": 0.0406,51 "step": 300052 },53 {54 "epoch": 4.89,55 "grad_norm": 0.0067610470578074455,56 "learning_rate": 4.46927374301676e-07,57 "loss": 0.0164,58 "step": 350059 }60 ],61 "logging_steps": 500,62 "max_steps": 3580,63 "num_input_tokens_seen": 0,64 "num_train_epochs": 5,65 "save_steps": 500,66 "total_flos": 3179834312329860.0,67 "train_batch_size": 32,68 "trial_name": null,69 "trial_params": null70}71 