| { | |
| "best_global_step": null, | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.911764705882353, | |
| "eval_steps": 500, | |
| "global_step": 99, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.14705882352941177, | |
| "grad_norm": 33.5, | |
| "learning_rate": 6.3462972571428566e-06, | |
| "loss": 2.992, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.29411764705882354, | |
| "grad_norm": 11.6875, | |
| "learning_rate": 1.4279168828571427e-05, | |
| "loss": 1.9335, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.4411764705882353, | |
| "grad_norm": 5.25, | |
| "learning_rate": 2.22120404e-05, | |
| "loss": 0.864, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.5882352941176471, | |
| "grad_norm": 2.875, | |
| "learning_rate": 3.0144911971428565e-05, | |
| "loss": 0.4883, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.7352941176470589, | |
| "grad_norm": 3.140625, | |
| "learning_rate": 3.807778354285714e-05, | |
| "loss": 0.3903, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.8823529411764706, | |
| "grad_norm": 2.265625, | |
| "learning_rate": 4.601065511428571e-05, | |
| "loss": 0.3232, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.9705882352941176, | |
| "eval_loss": 0.239887997508049, | |
| "eval_runtime": 7.6236, | |
| "eval_samples_per_second": 26.234, | |
| "eval_steps_per_second": 13.117, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 1.0294117647058822, | |
| "grad_norm": 1.9921875, | |
| "learning_rate": 5.394352668571428e-05, | |
| "loss": 0.3028, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 1.1764705882352942, | |
| "grad_norm": 2.265625, | |
| "learning_rate": 5.516490471297409e-05, | |
| "loss": 0.2643, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 1.3235294117647058, | |
| "grad_norm": 1.953125, | |
| "learning_rate": 5.3703220774514946e-05, | |
| "loss": 0.2536, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 1.4705882352941178, | |
| "grad_norm": 2.0, | |
| "learning_rate": 5.120213643384868e-05, | |
| "loss": 0.2485, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 1.6176470588235294, | |
| "grad_norm": 1.6484375, | |
| "learning_rate": 4.779849649317176e-05, | |
| "loss": 0.216, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 1.7647058823529411, | |
| "grad_norm": 1.5390625, | |
| "learning_rate": 4.3678528352678866e-05, | |
| "loss": 0.2206, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 1.9117647058823528, | |
| "grad_norm": 1.4609375, | |
| "learning_rate": 3.906765272932645e-05, | |
| "loss": 0.2196, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 1.9411764705882353, | |
| "eval_loss": 0.21876347064971924, | |
| "eval_runtime": 7.5337, | |
| "eval_samples_per_second": 26.547, | |
| "eval_steps_per_second": 13.274, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 2.0588235294117645, | |
| "grad_norm": 1.0703125, | |
| "learning_rate": 3.421814994504162e-05, | |
| "loss": 0.1871, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 2.2058823529411766, | |
| "grad_norm": 1.3046875, | |
| "learning_rate": 2.939535661341893e-05, | |
| "loss": 0.1307, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 2.3529411764705883, | |
| "grad_norm": 1.53125, | |
| "learning_rate": 2.4863147962002475e-05, | |
| "loss": 0.1182, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "grad_norm": 1.671875, | |
| "learning_rate": 2.086950011312128e-05, | |
| "loss": 0.1194, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 2.6470588235294117, | |
| "grad_norm": 1.140625, | |
| "learning_rate": 1.7632922271377727e-05, | |
| "loss": 0.1088, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 2.7941176470588234, | |
| "grad_norm": 1.0859375, | |
| "learning_rate": 1.5330501169652206e-05, | |
| "loss": 0.0998, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 2.911764705882353, | |
| "eval_loss": 0.19598107039928436, | |
| "eval_runtime": 7.443, | |
| "eval_samples_per_second": 26.871, | |
| "eval_steps_per_second": 13.435, | |
| "step": 99 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 102, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 3.536168761247662e+17, | |
| "train_batch_size": 40, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |