| { | |
| "best_global_step": 171, | |
| "best_metric": 1.0352519750595093, | |
| "best_model_checkpoint": "./gemma-patent-assistant/checkpoint-171", | |
| "epoch": 3.0, | |
| "eval_steps": 500, | |
| "global_step": 171, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.17777777777777778, | |
| "grad_norm": 12.329893112182617, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 12.9427, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.35555555555555557, | |
| "grad_norm": 8.226090431213379, | |
| "learning_rate": 1.7777777777777777e-05, | |
| "loss": 12.1671, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.5333333333333333, | |
| "grad_norm": 2.704000473022461, | |
| "learning_rate": 1.9865386046236597e-05, | |
| "loss": 10.8122, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.7111111111111111, | |
| "grad_norm": 1.6485083103179932, | |
| "learning_rate": 1.932472229404356e-05, | |
| "loss": 9.8163, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.8888888888888888, | |
| "grad_norm": 1.5979063510894775, | |
| "learning_rate": 1.839229287286327e-05, | |
| "loss": 9.4013, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 1.1528561115264893, | |
| "eval_runtime": 680.397, | |
| "eval_samples_per_second": 0.147, | |
| "eval_steps_per_second": 0.073, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 1.0533333333333332, | |
| "grad_norm": 1.6079875230789185, | |
| "learning_rate": 1.710727255106447e-05, | |
| "loss": 8.2906, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 1.231111111111111, | |
| "grad_norm": 2.0471384525299072, | |
| "learning_rate": 1.5693639270213138e-05, | |
| "loss": 8.8615, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 1.4088888888888889, | |
| "grad_norm": 1.443812370300293, | |
| "learning_rate": 1.3897858732926794e-05, | |
| "loss": 8.4692, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 1.5866666666666667, | |
| "grad_norm": 1.348875880241394, | |
| "learning_rate": 1.1938314902110701e-05, | |
| "loss": 8.3483, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 1.7644444444444445, | |
| "grad_norm": 1.4459165334701538, | |
| "learning_rate": 9.897335376977104e-06, | |
| "loss": 8.1664, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.942222222222222, | |
| "grad_norm": 1.4485529661178589, | |
| "learning_rate": 7.860669167935028e-06, | |
| "loss": 8.2337, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 1.046842098236084, | |
| "eval_runtime": 457.2398, | |
| "eval_samples_per_second": 0.219, | |
| "eval_steps_per_second": 0.109, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 2.1066666666666665, | |
| "grad_norm": 1.3392361402511597, | |
| "learning_rate": 5.913884067217686e-06, | |
| "loss": 7.5023, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 2.2844444444444445, | |
| "grad_norm": 1.6693141460418701, | |
| "learning_rate": 4.138771633147856e-06, | |
| "loss": 8.154, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 2.462222222222222, | |
| "grad_norm": 1.43593168258667, | |
| "learning_rate": 2.6099108277934105e-06, | |
| "loss": 8.0723, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "grad_norm": 1.5561100244522095, | |
| "learning_rate": 1.3915346821563235e-06, | |
| "loss": 7.9664, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 2.8177777777777777, | |
| "grad_norm": 1.4849770069122314, | |
| "learning_rate": 5.348316317440549e-07, | |
| "loss": 7.8758, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 2.9955555555555557, | |
| "grad_norm": 1.4685490131378174, | |
| "learning_rate": 7.579490328064265e-08, | |
| "loss": 8.2081, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_loss": 1.0352519750595093, | |
| "eval_runtime": 449.063, | |
| "eval_samples_per_second": 0.223, | |
| "eval_steps_per_second": 0.111, | |
| "step": 171 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 171, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 7.438388691861504e+17, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |