{ "best_metric": null, "best_model_checkpoint": null, "epoch": 0.0027845164012234996, "eval_steps": 5, "global_step": 33, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 8.437928488556059e-05, "eval_loss": NaN, "eval_runtime": 159.3732, "eval_samples_per_second": 31.31, "eval_steps_per_second": 15.655, "step": 1 }, { "epoch": 0.0002531378546566818, "grad_norm": NaN, "learning_rate": 3.529411764705883e-05, "loss": 0.0, "step": 3 }, { "epoch": 0.00042189642442780296, "eval_loss": NaN, "eval_runtime": 158.2987, "eval_samples_per_second": 31.523, "eval_steps_per_second": 15.761, "step": 5 }, { "epoch": 0.0005062757093133636, "grad_norm": NaN, "learning_rate": 7.058823529411765e-05, "loss": 0.0, "step": 6 }, { "epoch": 0.0007594135639700454, "grad_norm": NaN, "learning_rate": 0.00010588235294117647, "loss": 0.0, "step": 9 }, { "epoch": 0.0008437928488556059, "eval_loss": NaN, "eval_runtime": 159.6621, "eval_samples_per_second": 31.254, "eval_steps_per_second": 15.627, "step": 10 }, { "epoch": 0.0010125514186267272, "grad_norm": NaN, "learning_rate": 0.0001411764705882353, "loss": 0.0, "step": 12 }, { "epoch": 0.0012656892732834089, "grad_norm": NaN, "learning_rate": 0.00017647058823529413, "loss": 0.0, "step": 15 }, { "epoch": 0.0012656892732834089, "eval_loss": NaN, "eval_runtime": 160.3255, "eval_samples_per_second": 31.124, "eval_steps_per_second": 15.562, "step": 15 }, { "epoch": 0.0015188271279400908, "grad_norm": NaN, "learning_rate": 0.00019807852804032305, "loss": 0.0, "step": 18 }, { "epoch": 0.0016875856977112118, "eval_loss": NaN, "eval_runtime": 160.5119, "eval_samples_per_second": 31.088, "eval_steps_per_second": 15.544, "step": 20 }, { "epoch": 0.0017719649825967724, "grad_norm": NaN, "learning_rate": 0.00017071067811865476, "loss": 0.0, "step": 21 }, { "epoch": 0.0020251028372534545, "grad_norm": NaN, "learning_rate": 0.00011950903220161285, "loss": 0.0, "step": 24 }, { "epoch": 0.002109482122139015, "eval_loss": NaN, "eval_runtime": 160.8054, "eval_samples_per_second": 31.031, "eval_steps_per_second": 15.516, "step": 25 }, { "epoch": 0.002278240691910136, "grad_norm": NaN, "learning_rate": 6.173165676349103e-05, "loss": 0.0, "step": 27 }, { "epoch": 0.0025313785465668178, "grad_norm": NaN, "learning_rate": 1.6853038769745467e-05, "loss": 0.0, "step": 30 }, { "epoch": 0.0025313785465668178, "eval_loss": NaN, "eval_runtime": 159.5505, "eval_samples_per_second": 31.275, "eval_steps_per_second": 15.638, "step": 30 }, { "epoch": 0.0027845164012234996, "grad_norm": NaN, "learning_rate": 0.0, "loss": 0.0, "step": 33 } ], "logging_steps": 3, "max_steps": 33, "num_input_tokens_seen": 0, "num_train_epochs": 1, "save_steps": 17, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 594787298181120.0, "train_batch_size": 2, "trial_name": null, "trial_params": null }