| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.004640263566970604, | |
| "eval_steps": 10, | |
| "global_step": 50, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 9.280527133941208e-05, | |
| "eval_loss": 12.635147094726562, | |
| "eval_runtime": 511.924, | |
| "eval_samples_per_second": 8.863, | |
| "eval_steps_per_second": 4.432, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0004640263566970604, | |
| "grad_norm": 7.037264347076416, | |
| "learning_rate": 5e-05, | |
| "loss": 12.0158, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0009280527133941208, | |
| "grad_norm": 13.486483573913574, | |
| "learning_rate": 0.0001, | |
| "loss": 12.3291, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0009280527133941208, | |
| "eval_loss": 10.266976356506348, | |
| "eval_runtime": 512.5519, | |
| "eval_samples_per_second": 8.852, | |
| "eval_steps_per_second": 4.427, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0013920790700911812, | |
| "grad_norm": 14.867263793945312, | |
| "learning_rate": 9.619397662556435e-05, | |
| "loss": 8.3446, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.0018561054267882415, | |
| "grad_norm": 11.387835502624512, | |
| "learning_rate": 8.535533905932738e-05, | |
| "loss": 5.7081, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0018561054267882415, | |
| "eval_loss": 4.9822258949279785, | |
| "eval_runtime": 512.3945, | |
| "eval_samples_per_second": 8.855, | |
| "eval_steps_per_second": 4.428, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.002320131783485302, | |
| "grad_norm": 12.567839622497559, | |
| "learning_rate": 6.91341716182545e-05, | |
| "loss": 4.1337, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.0027841581401823624, | |
| "grad_norm": 11.602899551391602, | |
| "learning_rate": 5e-05, | |
| "loss": 2.1336, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0027841581401823624, | |
| "eval_loss": 1.5324822664260864, | |
| "eval_runtime": 512.2383, | |
| "eval_samples_per_second": 8.857, | |
| "eval_steps_per_second": 4.43, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.003248184496879423, | |
| "grad_norm": 8.674376487731934, | |
| "learning_rate": 3.086582838174551e-05, | |
| "loss": 1.1276, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.003712210853576483, | |
| "grad_norm": 10.511404991149902, | |
| "learning_rate": 1.4644660940672627e-05, | |
| "loss": 0.9892, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.003712210853576483, | |
| "eval_loss": 0.6702454090118408, | |
| "eval_runtime": 512.2876, | |
| "eval_samples_per_second": 8.856, | |
| "eval_steps_per_second": 4.429, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.0041762372102735435, | |
| "grad_norm": 6.408113956451416, | |
| "learning_rate": 3.8060233744356633e-06, | |
| "loss": 0.5061, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.004640263566970604, | |
| "grad_norm": 9.136789321899414, | |
| "learning_rate": 0.0, | |
| "loss": 0.4922, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.004640263566970604, | |
| "eval_loss": 0.594207227230072, | |
| "eval_runtime": 512.5019, | |
| "eval_samples_per_second": 8.853, | |
| "eval_steps_per_second": 4.427, | |
| "step": 50 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 50, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 13, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.5770856217116672e+16, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |