PlutoG99001/Generated
04
1{2 "best_metric": null,3 "best_model_checkpoint": null,4 "epoch": 3.5555555555555554,5 "eval_steps": 500,6 "global_step": 12,7 "is_hyper_param_search": false,8 "is_local_process_zero": true,9 "is_world_process_zero": true,10 "log_history": [11 {12 "epoch": 0.5925925925925926,13 "grad_norm": 1.2005751132965088,14 "learning_rate": 0.0001666666666666667,15 "loss": 9.4962,16 "step": 217 },18 {19 "epoch": 1.1851851851851851,20 "grad_norm": 1.47324538230896,21 "learning_rate": 0.00013333333333333334,22 "loss": 9.1456,23 "step": 424 },25 {26 "epoch": 1.7777777777777777,27 "grad_norm": 2.0616180896759033,28 "learning_rate": 0.0001,29 "loss": 8.731,30 "step": 631 },32 {33 "epoch": 2.3703703703703702,34 "grad_norm": 2.349423885345459,35 "learning_rate": 6.666666666666667e-05,36 "loss": 8.2574,37 "step": 838 },39 {40 "epoch": 2.962962962962963,41 "grad_norm": 1.723225474357605,42 "learning_rate": 3.3333333333333335e-05,43 "loss": 7.9055,44 "step": 1045 },46 {47 "epoch": 3.5555555555555554,48 "grad_norm": 1.465832233428955,49 "learning_rate": 0.0,50 "loss": 7.7976,51 "step": 1252 },53 {54 "epoch": 3.5555555555555554,55 "step": 12,56 "total_flos": 69844190099040.0,57 "train_loss": 8.555556774139404,58 "train_runtime": 79.8982,59 "train_samples_per_second": 2.703,60 "train_steps_per_second": 0.1561 }62 ],63 "logging_steps": 2,64 "max_steps": 12,65 "num_input_tokens_seen": 0,66 "num_train_epochs": 4,67 "save_steps": 500,68 "stateful_callbacks": {69 "TrainerControl": {70 "args": {71 "should_epoch_stop": false,72 "should_evaluate": false,73 "should_log": false,74 "should_save": true,75 "should_training_stop": true76 },77 "attributes": {}78 }79 },80 "total_flos": 69844190099040.0,81 "train_batch_size": 2,82 "trial_name": null,83 "trial_params": null84}85