| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.998109640831758, | |
| "global_step": 396, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5e-06, | |
| "loss": 1.6488, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1e-05, | |
| "loss": 1.6585, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.5e-05, | |
| "loss": 1.6373, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 2e-05, | |
| "loss": 1.5933, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 2.5e-05, | |
| "loss": 1.4982, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 3e-05, | |
| "loss": 1.3297, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 3.5e-05, | |
| "loss": 1.1515, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4e-05, | |
| "loss": 0.9944, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.5e-05, | |
| "loss": 0.8887, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 5e-05, | |
| "loss": 0.8275, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.954044117647059e-05, | |
| "loss": 0.8152, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.908088235294118e-05, | |
| "loss": 0.7991, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.8621323529411765e-05, | |
| "loss": 0.7932, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.816176470588236e-05, | |
| "loss": 0.7928, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.7702205882352946e-05, | |
| "loss": 0.7968, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.7242647058823534e-05, | |
| "loss": 0.7744, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.678308823529412e-05, | |
| "loss": 0.7791, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.632352941176471e-05, | |
| "loss": 0.7736, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.5863970588235296e-05, | |
| "loss": 0.766, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.5404411764705883e-05, | |
| "loss": 0.7717, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.494485294117647e-05, | |
| "loss": 0.7638, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.448529411764706e-05, | |
| "loss": 0.768, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.4025735294117646e-05, | |
| "loss": 0.7573, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.356617647058824e-05, | |
| "loss": 0.7633, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4.310661764705883e-05, | |
| "loss": 0.7562, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 4.2647058823529415e-05, | |
| "loss": 0.7576, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.21875e-05, | |
| "loss": 0.7516, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.172794117647059e-05, | |
| "loss": 0.7501, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 4.126838235294118e-05, | |
| "loss": 0.7601, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 4.0808823529411765e-05, | |
| "loss": 0.7502, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 4.034926470588236e-05, | |
| "loss": 0.7583, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 3.9889705882352946e-05, | |
| "loss": 0.756, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 3.943014705882353e-05, | |
| "loss": 0.7511, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 3.897058823529412e-05, | |
| "loss": 0.7434, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 3.851102941176471e-05, | |
| "loss": 0.7586, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 3.8051470588235296e-05, | |
| "loss": 0.7486, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 3.759191176470588e-05, | |
| "loss": 0.7499, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 3.713235294117647e-05, | |
| "loss": 0.7516, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 3.667279411764706e-05, | |
| "loss": 0.7451, | |
| "step": 390 | |
| } | |
| ], | |
| "max_steps": 1188, | |
| "num_train_epochs": 3, | |
| "total_flos": 2.029497948313328e+19, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |