| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 35.0, | |
| "global_step": 19705, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 4.8731286475513835e-05, | |
| "loss": 3.7437, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 4.746257295102766e-05, | |
| "loss": 3.203, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 4.619385942654149e-05, | |
| "loss": 2.9759, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 4.492514590205532e-05, | |
| "loss": 2.7917, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "learning_rate": 4.365643237756914e-05, | |
| "loss": 2.6427, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 5.33, | |
| "learning_rate": 4.2387718853082975e-05, | |
| "loss": 2.5541, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 6.22, | |
| "learning_rate": 4.111900532859681e-05, | |
| "loss": 2.4485, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 7.1, | |
| "learning_rate": 3.985029180411063e-05, | |
| "loss": 2.3436, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 7.99, | |
| "learning_rate": 3.8581578279624465e-05, | |
| "loss": 2.2573, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 8.88, | |
| "learning_rate": 3.73128647551383e-05, | |
| "loss": 2.1701, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 9.77, | |
| "learning_rate": 3.6044151230652115e-05, | |
| "loss": 2.0915, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 10.66, | |
| "learning_rate": 3.477543770616595e-05, | |
| "loss": 2.0424, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 11.55, | |
| "learning_rate": 3.350672418167978e-05, | |
| "loss": 1.9703, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 12.43, | |
| "learning_rate": 3.2238010657193605e-05, | |
| "loss": 1.9065, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 13.32, | |
| "learning_rate": 3.096929713270744e-05, | |
| "loss": 1.877, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 14.21, | |
| "learning_rate": 2.9700583608221266e-05, | |
| "loss": 1.8305, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 15.1, | |
| "learning_rate": 2.8431870083735095e-05, | |
| "loss": 1.7644, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 15.99, | |
| "learning_rate": 2.7163156559248927e-05, | |
| "loss": 1.7179, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 16.87, | |
| "learning_rate": 2.589444303476275e-05, | |
| "loss": 1.6708, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 17.76, | |
| "learning_rate": 2.462572951027658e-05, | |
| "loss": 1.6338, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 18.65, | |
| "learning_rate": 2.335701598579041e-05, | |
| "loss": 1.5861, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 19.54, | |
| "learning_rate": 2.208830246130424e-05, | |
| "loss": 1.553, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 20.43, | |
| "learning_rate": 2.0819588936818067e-05, | |
| "loss": 1.538, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 21.31, | |
| "learning_rate": 1.9550875412331896e-05, | |
| "loss": 1.5036, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 22.2, | |
| "learning_rate": 1.8282161887845725e-05, | |
| "loss": 1.4526, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 23.09, | |
| "learning_rate": 1.7013448363359553e-05, | |
| "loss": 1.4547, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 23.98, | |
| "learning_rate": 1.5744734838873386e-05, | |
| "loss": 1.4221, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 24.87, | |
| "learning_rate": 1.4476021314387211e-05, | |
| "loss": 1.3794, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 25.75, | |
| "learning_rate": 1.3207307789901041e-05, | |
| "loss": 1.3785, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 26.64, | |
| "learning_rate": 1.193859426541487e-05, | |
| "loss": 1.345, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 27.53, | |
| "learning_rate": 1.0669880740928699e-05, | |
| "loss": 1.326, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 28.42, | |
| "learning_rate": 9.401167216442528e-06, | |
| "loss": 1.3241, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 29.31, | |
| "learning_rate": 8.132453691956356e-06, | |
| "loss": 1.2962, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 30.2, | |
| "learning_rate": 6.863740167470185e-06, | |
| "loss": 1.2871, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 31.08, | |
| "learning_rate": 5.595026642984015e-06, | |
| "loss": 1.2761, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 31.97, | |
| "learning_rate": 4.3263131184978435e-06, | |
| "loss": 1.2549, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 32.86, | |
| "learning_rate": 3.0575995940116723e-06, | |
| "loss": 1.252, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 33.75, | |
| "learning_rate": 1.7888860695255013e-06, | |
| "loss": 1.2429, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 34.64, | |
| "learning_rate": 5.201725450393302e-07, | |
| "loss": 1.2436, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 35.0, | |
| "step": 19705, | |
| "total_flos": 9895714212962304.0, | |
| "train_loss": 1.8335362397363058, | |
| "train_runtime": 2725.9313, | |
| "train_samples_per_second": 216.554, | |
| "train_steps_per_second": 7.229 | |
| } | |
| ], | |
| "max_steps": 19705, | |
| "num_train_epochs": 35, | |
| "total_flos": 9895714212962304.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |