SpencerULee/ThingForMeToFindLaterLaterLater
07
1{2 "best_metric": 0.6401928067207336,3 "best_model_checkpoint": "ThingForMeToFindLaterLaterLater/checkpoint-3",4 "epoch": 3.0,5 "eval_steps": 500,6 "global_step": 3,7 "is_hyper_param_search": false,8 "is_local_process_zero": true,9 "is_world_process_zero": true,10 "log_history": [11 {12 "epoch": 1.0,13 "grad_norm": 8.194887161254883,14 "learning_rate": 5e-05,15 "loss": 0.737,16 "step": 117 },18 {19 "epoch": 1.0,20 "eval_accuracy": 0.6666666666666666,21 "eval_auc": 1.0,22 "eval_f1": 0.0,23 "eval_loss": 0.6558270454406738,24 "eval_precision": 0.0,25 "eval_recall": 0.0,26 "eval_runtime": 1.998,27 "eval_samples_per_second": 1.501,28 "eval_steps_per_second": 0.5,29 "step": 130 },31 {32 "epoch": 2.0,33 "grad_norm": 3.131235122680664,34 "learning_rate": 2.5e-05,35 "loss": 0.6864,36 "step": 237 },38 {39 "epoch": 2.0,40 "eval_accuracy": 0.6666666666666666,41 "eval_auc": 0.5,42 "eval_f1": 0.0,43 "eval_loss": 0.6422247290611267,44 "eval_precision": 0.0,45 "eval_recall": 0.0,46 "eval_runtime": 2.1454,47 "eval_samples_per_second": 1.398,48 "eval_steps_per_second": 0.466,49 "step": 250 },51 {52 "epoch": 3.0,53 "grad_norm": 9.23309326171875,54 "learning_rate": 0.0,55 "loss": 0.7973,56 "step": 357 },58 {59 "epoch": 3.0,60 "eval_accuracy": 0.6666666666666666,61 "eval_auc": 0.5,62 "eval_f1": 0.0,63 "eval_loss": 0.6401928067207336,64 "eval_precision": 0.0,65 "eval_recall": 0.0,66 "eval_runtime": 2.0449,67 "eval_samples_per_second": 1.467,68 "eval_steps_per_second": 0.489,69 "step": 370 }71 ],72 "logging_steps": 1,73 "max_steps": 3,74 "num_input_tokens_seen": 0,75 "num_train_epochs": 3,76 "save_steps": 500,77 "stateful_callbacks": {78 "EarlyStoppingCallback": {79 "args": {80 "early_stopping_patience": 5,81 "early_stopping_threshold": 0.0182 },83 "attributes": {84 "early_stopping_patience_counter": 085 }86 },87 "TrainerControl": {88 "args": {89 "should_epoch_stop": false,90 "should_evaluate": false,91 "should_log": false,92 "should_save": true,93 "should_training_stop": true94 },95 "attributes": {}96 }97 },98 "total_flos": 5591588179968.0,99 "train_batch_size": 8,100 "trial_name": null,101 "trial_params": null102}103 