Invalid JSON: Unexpected token 'N', ..."ad_norm": NaN,
"... is not valid JSON
| { | |
| "best_global_step": 11, | |
| "best_metric": 3.189903974533081, | |
| "best_model_checkpoint": "model/model-KeyPhrases-Filtered-Valid-CodeT5plus-30062025/checkpoint-11", | |
| "epoch": 2.6, | |
| "eval_steps": 1, | |
| "global_step": 13, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": NaN, | |
| "learning_rate": 5e-05, | |
| "loss": 5.3862, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "eval_loss": 6.960543155670166, | |
| "eval_runtime": 0.2213, | |
| "eval_samples_per_second": 45.177, | |
| "eval_steps_per_second": 22.589, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 48.27574157714844, | |
| "learning_rate": 5e-05, | |
| "loss": 5.157, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "eval_loss": 6.043793201446533, | |
| "eval_runtime": 0.1902, | |
| "eval_samples_per_second": 52.568, | |
| "eval_steps_per_second": 26.284, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 29.348539352416992, | |
| "learning_rate": 4.875e-05, | |
| "loss": 4.5692, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "eval_loss": 5.366965293884277, | |
| "eval_runtime": 0.2582, | |
| "eval_samples_per_second": 38.73, | |
| "eval_steps_per_second": 19.365, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": NaN, | |
| "learning_rate": 4.75e-05, | |
| "loss": 4.1059, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "eval_loss": 5.366965293884277, | |
| "eval_runtime": 0.1956, | |
| "eval_samples_per_second": 51.136, | |
| "eval_steps_per_second": 25.568, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 23.603010177612305, | |
| "learning_rate": 4.75e-05, | |
| "loss": 4.3436, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 4.698573589324951, | |
| "eval_runtime": 0.1934, | |
| "eval_samples_per_second": 51.7, | |
| "eval_steps_per_second": 25.85, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "grad_norm": 33.01340866088867, | |
| "learning_rate": 4.6250000000000006e-05, | |
| "loss": 2.76, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "eval_loss": 4.0736799240112305, | |
| "eval_runtime": 0.1874, | |
| "eval_samples_per_second": 53.373, | |
| "eval_steps_per_second": 26.686, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "grad_norm": 39.976985931396484, | |
| "learning_rate": 4.5e-05, | |
| "loss": 3.339, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "eval_loss": 3.5937087535858154, | |
| "eval_runtime": 0.1926, | |
| "eval_samples_per_second": 51.908, | |
| "eval_steps_per_second": 25.954, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "grad_norm": Infinity, | |
| "learning_rate": 4.375e-05, | |
| "loss": 1.9873, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "eval_loss": 3.5937087535858154, | |
| "eval_runtime": 0.1925, | |
| "eval_samples_per_second": 51.939, | |
| "eval_steps_per_second": 25.969, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "grad_norm": 25.079561233520508, | |
| "learning_rate": 4.375e-05, | |
| "loss": 1.5978, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "eval_loss": 3.3703582286834717, | |
| "eval_runtime": 0.1963, | |
| "eval_samples_per_second": 50.937, | |
| "eval_steps_per_second": 25.469, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 16.453521728515625, | |
| "learning_rate": 4.25e-05, | |
| "loss": 3.1761, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 3.2319865226745605, | |
| "eval_runtime": 0.1975, | |
| "eval_samples_per_second": 50.629, | |
| "eval_steps_per_second": 25.314, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "grad_norm": 14.565013885498047, | |
| "learning_rate": 4.125e-05, | |
| "loss": 2.0577, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "eval_loss": 3.189903974533081, | |
| "eval_runtime": 0.1998, | |
| "eval_samples_per_second": 50.048, | |
| "eval_steps_per_second": 25.024, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "grad_norm": 14.65517520904541, | |
| "learning_rate": 4e-05, | |
| "loss": 1.5143, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "eval_loss": 3.2138240337371826, | |
| "eval_runtime": 0.2563, | |
| "eval_samples_per_second": 39.023, | |
| "eval_steps_per_second": 19.512, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "grad_norm": 15.011869430541992, | |
| "learning_rate": 3.875e-05, | |
| "loss": 0.9242, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "eval_loss": 3.242772340774536, | |
| "eval_runtime": 0.196, | |
| "eval_samples_per_second": 51.011, | |
| "eval_steps_per_second": 25.506, | |
| "step": 13 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 40, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 8, | |
| "save_steps": 1, | |
| "stateful_callbacks": { | |
| "LossLoggerCallback": { | |
| "args": { | |
| "early_stopping_patience": 2, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 2 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 15832905154560.0, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |