| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.005371434710211097, | |
| "eval_steps": 10, | |
| "global_step": 50, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.00010742869420422194, | |
| "eval_loss": 3.1173388957977295, | |
| "eval_runtime": 479.8029, | |
| "eval_samples_per_second": 8.17, | |
| "eval_steps_per_second": 4.085, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0005371434710211097, | |
| "grad_norm": 3.4544060230255127, | |
| "learning_rate": 5e-05, | |
| "loss": 2.8156, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0010742869420422194, | |
| "grad_norm": 3.225125551223755, | |
| "learning_rate": 0.0001, | |
| "loss": 2.5232, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0010742869420422194, | |
| "eval_loss": 2.313772439956665, | |
| "eval_runtime": 482.0841, | |
| "eval_samples_per_second": 8.131, | |
| "eval_steps_per_second": 4.066, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0016114304130633293, | |
| "grad_norm": 4.3110456466674805, | |
| "learning_rate": 9.619397662556435e-05, | |
| "loss": 1.6044, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.002148573884084439, | |
| "grad_norm": 4.142456531524658, | |
| "learning_rate": 8.535533905932738e-05, | |
| "loss": 1.1602, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.002148573884084439, | |
| "eval_loss": 1.0064030885696411, | |
| "eval_runtime": 480.91, | |
| "eval_samples_per_second": 8.151, | |
| "eval_steps_per_second": 4.076, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0026857173551055485, | |
| "grad_norm": 4.778378963470459, | |
| "learning_rate": 6.91341716182545e-05, | |
| "loss": 0.8835, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.0032228608261266585, | |
| "grad_norm": 5.472487449645996, | |
| "learning_rate": 5e-05, | |
| "loss": 0.8802, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0032228608261266585, | |
| "eval_loss": 0.7272688150405884, | |
| "eval_runtime": 481.7044, | |
| "eval_samples_per_second": 8.138, | |
| "eval_steps_per_second": 4.069, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.003760004297147768, | |
| "grad_norm": 4.021572113037109, | |
| "learning_rate": 3.086582838174551e-05, | |
| "loss": 0.6523, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.004297147768168878, | |
| "grad_norm": 5.594470977783203, | |
| "learning_rate": 1.4644660940672627e-05, | |
| "loss": 0.706, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.004297147768168878, | |
| "eval_loss": 0.6382240056991577, | |
| "eval_runtime": 481.9991, | |
| "eval_samples_per_second": 8.133, | |
| "eval_steps_per_second": 4.066, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.004834291239189987, | |
| "grad_norm": 3.5907773971557617, | |
| "learning_rate": 3.8060233744356633e-06, | |
| "loss": 0.4785, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.005371434710211097, | |
| "grad_norm": 3.8063602447509766, | |
| "learning_rate": 0.0, | |
| "loss": 0.6601, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.005371434710211097, | |
| "eval_loss": 0.6105340123176575, | |
| "eval_runtime": 482.4457, | |
| "eval_samples_per_second": 8.125, | |
| "eval_steps_per_second": 4.063, | |
| "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": 1.84959646040064e+16, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |