| { | |
| "best_metric": 8.475200957036577e-06, | |
| "best_model_checkpoint": "./t5_sql_model/checkpoint-1000", | |
| "epoch": 10.0, | |
| "eval_steps": 100, | |
| "global_step": 1000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 0.19069062173366547, | |
| "learning_rate": 4.5050000000000004e-05, | |
| "loss": 0.0952, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 0.01930948905646801, | |
| "eval_runtime": 4.2969, | |
| "eval_samples_per_second": 46.545, | |
| "eval_steps_per_second": 23.273, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 0.11524572968482971, | |
| "learning_rate": 4.0050000000000004e-05, | |
| "loss": 0.0134, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 0.00038512772880494595, | |
| "eval_runtime": 4.2896, | |
| "eval_samples_per_second": 46.624, | |
| "eval_steps_per_second": 23.312, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "grad_norm": 0.08386600762605667, | |
| "learning_rate": 3.505e-05, | |
| "loss": 0.0035, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_loss": 7.71833147155121e-05, | |
| "eval_runtime": 4.2776, | |
| "eval_samples_per_second": 46.755, | |
| "eval_steps_per_second": 23.377, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "grad_norm": 0.04387281462550163, | |
| "learning_rate": 3.0050000000000002e-05, | |
| "loss": 0.0018, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_loss": 3.606174868764356e-05, | |
| "eval_runtime": 4.288, | |
| "eval_samples_per_second": 46.642, | |
| "eval_steps_per_second": 23.321, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "grad_norm": 0.02179218828678131, | |
| "learning_rate": 2.5050000000000002e-05, | |
| "loss": 0.0011, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_loss": 1.9488097677822225e-05, | |
| "eval_runtime": 4.2812, | |
| "eval_samples_per_second": 46.715, | |
| "eval_steps_per_second": 23.358, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "grad_norm": 0.02126687951385975, | |
| "learning_rate": 2.0050000000000003e-05, | |
| "loss": 0.001, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_loss": 1.42730787047185e-05, | |
| "eval_runtime": 4.3012, | |
| "eval_samples_per_second": 46.498, | |
| "eval_steps_per_second": 23.249, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "grad_norm": 0.03087877482175827, | |
| "learning_rate": 1.505e-05, | |
| "loss": 0.0007, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_loss": 1.0926911272690631e-05, | |
| "eval_runtime": 4.2954, | |
| "eval_samples_per_second": 46.561, | |
| "eval_steps_per_second": 23.281, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "grad_norm": 0.015643499791622162, | |
| "learning_rate": 1.005e-05, | |
| "loss": 0.0006, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_loss": 9.499091902398504e-06, | |
| "eval_runtime": 4.2684, | |
| "eval_samples_per_second": 46.857, | |
| "eval_steps_per_second": 23.428, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "grad_norm": 0.020900582894682884, | |
| "learning_rate": 5.050000000000001e-06, | |
| "loss": 0.0005, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_loss": 8.78215996635845e-06, | |
| "eval_runtime": 4.2696, | |
| "eval_samples_per_second": 46.843, | |
| "eval_steps_per_second": 23.422, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "grad_norm": 0.00630880007520318, | |
| "learning_rate": 5.0000000000000004e-08, | |
| "loss": 0.0005, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_loss": 8.475200957036577e-06, | |
| "eval_runtime": 4.2698, | |
| "eval_samples_per_second": 46.841, | |
| "eval_steps_per_second": 23.42, | |
| "step": 1000 | |
| } | |
| ], | |
| "logging_steps": 100, | |
| "max_steps": 1000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 10, | |
| "save_steps": 100, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1082734411776000.0, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |