| { | |
| "best_metric": 0.12014812231063843, | |
| "best_model_checkpoint": "./model/checkpoints/checkpoint-100", | |
| "epoch": 1.1594202898550725, | |
| "eval_steps": 50, | |
| "global_step": 100, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.1855072463768116, | |
| "grad_norm": 0.9348331689834595, | |
| "learning_rate": 9.62962962962963e-05, | |
| "loss": 1.7376, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.3710144927536232, | |
| "grad_norm": 0.2084011733531952, | |
| "learning_rate": 8.148148148148148e-05, | |
| "loss": 0.2418, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.5565217391304348, | |
| "grad_norm": 0.2130679488182068, | |
| "learning_rate": 6.666666666666667e-05, | |
| "loss": 0.1875, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.5797101449275363, | |
| "eval_loss": 0.16450659930706024, | |
| "eval_model_preparation_time": 0.0017, | |
| "eval_runtime": 107.7969, | |
| "eval_samples_per_second": 1.707, | |
| "eval_steps_per_second": 0.213, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.7420289855072464, | |
| "grad_norm": 0.2075987011194229, | |
| "learning_rate": 5.185185185185185e-05, | |
| "loss": 0.1725, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.927536231884058, | |
| "grad_norm": 0.19163866341114044, | |
| "learning_rate": 3.7037037037037037e-05, | |
| "loss": 0.1437, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 1.1130434782608696, | |
| "grad_norm": 0.19873079657554626, | |
| "learning_rate": 2.2222222222222223e-05, | |
| "loss": 0.138, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 1.1594202898550725, | |
| "eval_loss": 0.12014812231063843, | |
| "eval_model_preparation_time": 0.0017, | |
| "eval_runtime": 15.2427, | |
| "eval_samples_per_second": 12.071, | |
| "eval_steps_per_second": 1.509, | |
| "step": 100 | |
| } | |
| ], | |
| "logging_steps": 16, | |
| "max_steps": 120, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "save_steps": 50, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.2500271180693504e+16, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |