| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.9999490445859873, | |
| "global_step": 29436, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.915069982334556e-05, | |
| "loss": 1.8335, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.830139964669113e-05, | |
| "loss": 1.3358, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.745209947003669e-05, | |
| "loss": 1.2093, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.660279929338225e-05, | |
| "loss": 1.1204, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.5753499116727814e-05, | |
| "loss": 1.0652, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.490419894007338e-05, | |
| "loss": 0.9992, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.405489876341894e-05, | |
| "loss": 0.9696, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.32055985867645e-05, | |
| "loss": 0.9374, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.235629841011007e-05, | |
| "loss": 0.9147, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.150699823345564e-05, | |
| "loss": 0.8813, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.06576980568012e-05, | |
| "loss": 0.8565, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.980839788014676e-05, | |
| "loss": 0.8328, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.895909770349233e-05, | |
| "loss": 0.8021, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.810979752683789e-05, | |
| "loss": 0.7917, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 3.726049735018345e-05, | |
| "loss": 0.7935, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 3.641119717352902e-05, | |
| "loss": 0.7608, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 3.556189699687458e-05, | |
| "loss": 0.7619, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 3.471259682022014e-05, | |
| "loss": 0.7359, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 3.386329664356571e-05, | |
| "loss": 0.7315, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.301399646691127e-05, | |
| "loss": 0.7218, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 3.216469629025683e-05, | |
| "loss": 0.7217, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 3.131539611360239e-05, | |
| "loss": 0.7214, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 3.0466095936947957e-05, | |
| "loss": 0.6976, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 2.961679576029352e-05, | |
| "loss": 0.6782, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 2.8767495583639082e-05, | |
| "loss": 0.6743, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 2.791819540698465e-05, | |
| "loss": 0.6816, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 2.706889523033021e-05, | |
| "loss": 0.6558, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 2.621959505367577e-05, | |
| "loss": 0.6545, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 2.5370294877021332e-05, | |
| "loss": 0.6515, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 2.4520994700366897e-05, | |
| "loss": 0.6412, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 2.367169452371246e-05, | |
| "loss": 0.6214, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 2.2822394347058025e-05, | |
| "loss": 0.6273, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 2.197309417040359e-05, | |
| "loss": 0.6249, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 2.1123793993749154e-05, | |
| "loss": 0.6123, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 2.0274493817094715e-05, | |
| "loss": 0.6152, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 1.942519364044028e-05, | |
| "loss": 0.6166, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 1.857589346378584e-05, | |
| "loss": 0.6075, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 1.7726593287131404e-05, | |
| "loss": 0.5962, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 1.687729311047697e-05, | |
| "loss": 0.5921, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 1.602799293382253e-05, | |
| "loss": 0.5915, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 1.5178692757168095e-05, | |
| "loss": 0.5669, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 1.4329392580513656e-05, | |
| "loss": 0.5725, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 1.348009240385922e-05, | |
| "loss": 0.5728, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 1.2630792227204785e-05, | |
| "loss": 0.5588, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 1.1781492050550347e-05, | |
| "loss": 0.5648, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 1.093219187389591e-05, | |
| "loss": 0.5581, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 1.0082891697241474e-05, | |
| "loss": 0.5553, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 9.233591520587037e-06, | |
| "loss": 0.5439, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 8.3842913439326e-06, | |
| "loss": 0.5367, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 7.534991167278163e-06, | |
| "loss": 0.5401, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 6.685690990623726e-06, | |
| "loss": 0.5302, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 5.83639081396929e-06, | |
| "loss": 0.541, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 4.987090637314853e-06, | |
| "loss": 0.5218, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 4.137790460660416e-06, | |
| "loss": 0.5149, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 3.2884902840059795e-06, | |
| "loss": 0.5358, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 2.4391901073515425e-06, | |
| "loss": 0.5353, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 1.589889930697106e-06, | |
| "loss": 0.5171, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 7.405897540426689e-07, | |
| "loss": 0.528, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 29436, | |
| "total_flos": 2.745512583065764e+17, | |
| "train_loss": 0.7166476119453937, | |
| "train_runtime": 30706.9194, | |
| "train_samples_per_second": 9.586, | |
| "train_steps_per_second": 0.959 | |
| } | |
| ], | |
| "max_steps": 29436, | |
| "num_train_epochs": 3, | |
| "total_flos": 2.745512583065764e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |