{ "best_metric": null, "best_model_checkpoint": null, "epoch": 2.9311187103077674, "global_step": 6000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.05, "learning_rate": 1e-05, "loss": 0.2271, "step": 100 }, { "epoch": 0.1, "learning_rate": 2e-05, "loss": 0.049, "step": 200 }, { "epoch": 0.15, "learning_rate": 3e-05, "loss": 0.0196, "step": 300 }, { "epoch": 0.2, "learning_rate": 4e-05, "loss": 0.0283, "step": 400 }, { "epoch": 0.24, "learning_rate": 5e-05, "loss": 0.0107, "step": 500 }, { "epoch": 0.29, "learning_rate": 4.9113632334692436e-05, "loss": 0.0128, "step": 600 }, { "epoch": 0.34, "learning_rate": 4.822726466938486e-05, "loss": 0.0086, "step": 700 }, { "epoch": 0.39, "learning_rate": 4.7340897004077296e-05, "loss": 0.011, "step": 800 }, { "epoch": 0.44, "learning_rate": 4.645452933876972e-05, "loss": 0.0023, "step": 900 }, { "epoch": 0.49, "learning_rate": 4.5568161673462156e-05, "loss": 0.0064, "step": 1000 }, { "epoch": 0.54, "learning_rate": 4.468179400815459e-05, "loss": 0.0233, "step": 1100 }, { "epoch": 0.59, "learning_rate": 4.379542634284701e-05, "loss": 0.0033, "step": 1200 }, { "epoch": 0.64, "learning_rate": 4.290905867753944e-05, "loss": 0.0217, "step": 1300 }, { "epoch": 0.68, "learning_rate": 4.2022691012231876e-05, "loss": 0.0014, "step": 1400 }, { "epoch": 0.73, "learning_rate": 4.11363233469243e-05, "loss": 0.0177, "step": 1500 }, { "epoch": 0.78, "learning_rate": 4.0249955681616736e-05, "loss": 0.0085, "step": 1600 }, { "epoch": 0.83, "learning_rate": 3.936358801630917e-05, "loss": 0.0172, "step": 1700 }, { "epoch": 0.88, "learning_rate": 3.8477220351001596e-05, "loss": 0.0001, "step": 1800 }, { "epoch": 0.93, "learning_rate": 3.759085268569403e-05, "loss": 0.0003, "step": 1900 }, { "epoch": 0.98, "learning_rate": 3.6704485020386456e-05, "loss": 0.0107, "step": 2000 }, { "epoch": 1.03, "learning_rate": 3.581811735507889e-05, "loss": 0.0001, "step": 2100 }, { "epoch": 1.07, "learning_rate": 3.493174968977132e-05, "loss": 0.0, "step": 2200 }, { "epoch": 1.12, "learning_rate": 3.404538202446375e-05, "loss": 0.0051, "step": 2300 }, { "epoch": 1.17, "learning_rate": 3.315901435915618e-05, "loss": 0.0, "step": 2400 }, { "epoch": 1.22, "learning_rate": 3.227264669384861e-05, "loss": 0.0, "step": 2500 }, { "epoch": 1.27, "learning_rate": 3.138627902854104e-05, "loss": 0.0, "step": 2600 }, { "epoch": 1.32, "learning_rate": 3.049991136323347e-05, "loss": 0.0095, "step": 2700 }, { "epoch": 1.37, "learning_rate": 2.96135436979259e-05, "loss": 0.0062, "step": 2800 }, { "epoch": 1.42, "learning_rate": 2.872717603261833e-05, "loss": 0.0001, "step": 2900 }, { "epoch": 1.47, "learning_rate": 2.784080836731076e-05, "loss": 0.009, "step": 3000 }, { "epoch": 1.51, "learning_rate": 2.6954440702003194e-05, "loss": 0.0054, "step": 3100 }, { "epoch": 1.56, "learning_rate": 2.6068073036695624e-05, "loss": 0.0001, "step": 3200 }, { "epoch": 1.61, "learning_rate": 2.5181705371388054e-05, "loss": 0.0123, "step": 3300 }, { "epoch": 1.66, "learning_rate": 2.4295337706080484e-05, "loss": 0.0, "step": 3400 }, { "epoch": 1.71, "learning_rate": 2.3408970040772914e-05, "loss": 0.0052, "step": 3500 }, { "epoch": 1.76, "learning_rate": 2.2522602375465344e-05, "loss": 0.0, "step": 3600 }, { "epoch": 1.81, "learning_rate": 2.1636234710157774e-05, "loss": 0.0001, "step": 3700 }, { "epoch": 1.86, "learning_rate": 2.0749867044850204e-05, "loss": 0.0, "step": 3800 }, { "epoch": 1.91, "learning_rate": 1.9863499379542634e-05, "loss": 0.006, "step": 3900 }, { "epoch": 1.95, "learning_rate": 1.8977131714235067e-05, "loss": 0.0, "step": 4000 }, { "epoch": 2.0, "learning_rate": 1.8090764048927497e-05, "loss": 0.002, "step": 4100 }, { "epoch": 2.05, "learning_rate": 1.7204396383619924e-05, "loss": 0.0022, "step": 4200 }, { "epoch": 2.1, "learning_rate": 1.6318028718312357e-05, "loss": 0.0004, "step": 4300 }, { "epoch": 2.15, "learning_rate": 1.5431661053004787e-05, "loss": 0.0, "step": 4400 }, { "epoch": 2.2, "learning_rate": 1.4545293387697218e-05, "loss": 0.004, "step": 4500 }, { "epoch": 2.25, "learning_rate": 1.3658925722389648e-05, "loss": 0.0001, "step": 4600 }, { "epoch": 2.3, "learning_rate": 1.277255805708208e-05, "loss": 0.0, "step": 4700 }, { "epoch": 2.34, "learning_rate": 1.1886190391774508e-05, "loss": 0.0001, "step": 4800 }, { "epoch": 2.39, "learning_rate": 1.099982272646694e-05, "loss": 0.0, "step": 4900 }, { "epoch": 2.44, "learning_rate": 1.011345506115937e-05, "loss": 0.0, "step": 5000 }, { "epoch": 2.49, "learning_rate": 9.2270873958518e-06, "loss": 0.0042, "step": 5100 }, { "epoch": 2.54, "learning_rate": 8.34071973054423e-06, "loss": 0.0, "step": 5200 }, { "epoch": 2.59, "learning_rate": 7.454352065236661e-06, "loss": 0.0, "step": 5300 }, { "epoch": 2.64, "learning_rate": 6.5679843999290904e-06, "loss": 0.0, "step": 5400 }, { "epoch": 2.69, "learning_rate": 5.681616734621521e-06, "loss": 0.0035, "step": 5500 }, { "epoch": 2.74, "learning_rate": 4.795249069313951e-06, "loss": 0.0025, "step": 5600 }, { "epoch": 2.78, "learning_rate": 3.908881404006382e-06, "loss": 0.0025, "step": 5700 }, { "epoch": 2.83, "learning_rate": 3.0225137386988123e-06, "loss": 0.0, "step": 5800 }, { "epoch": 2.88, "learning_rate": 2.1361460733912427e-06, "loss": 0.0015, "step": 5900 }, { "epoch": 2.93, "learning_rate": 1.2497784080836732e-06, "loss": 0.0016, "step": 6000 } ], "max_steps": 6141, "num_train_epochs": 3, "total_flos": 1948422670478280.0, "trial_name": null, "trial_params": null }