| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.2097819681791395, | |
| "global_step": 15000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 2.5e-05, | |
| "loss": 2.4058, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7915, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.981448501038884e-05, | |
| "loss": 1.7151, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.9628970020777684e-05, | |
| "loss": 1.6582, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.944345503116652e-05, | |
| "loss": 1.634, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.925794004155536e-05, | |
| "loss": 1.6209, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.90724250519442e-05, | |
| "loss": 1.6075, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.888691006233304e-05, | |
| "loss": 1.5856, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.870139507272188e-05, | |
| "loss": 1.5721, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.8515880083110716e-05, | |
| "loss": 1.5656, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.833036509349956e-05, | |
| "loss": 1.5553, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.814485010388839e-05, | |
| "loss": 1.5447, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.7959335114277236e-05, | |
| "loss": 1.5448, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.777382012466607e-05, | |
| "loss": 1.5326, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.758830513505492e-05, | |
| "loss": 1.5182, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.7402790145443755e-05, | |
| "loss": 1.5269, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4.721727515583259e-05, | |
| "loss": 1.5015, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 4.703176016622143e-05, | |
| "loss": 1.5099, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.684624517661027e-05, | |
| "loss": 1.5057, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 4.666073018699911e-05, | |
| "loss": 1.5067, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 4.647521519738795e-05, | |
| "loss": 1.4833, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 4.6289700207776794e-05, | |
| "loss": 1.4926, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 4.610418521816563e-05, | |
| "loss": 1.4981, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 4.591867022855447e-05, | |
| "loss": 1.4826, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 4.573315523894331e-05, | |
| "loss": 1.4849, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.5547640249332144e-05, | |
| "loss": 1.4865, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 4.536212525972099e-05, | |
| "loss": 1.467, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 4.5176610270109826e-05, | |
| "loss": 1.3277, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 4.499109528049867e-05, | |
| "loss": 1.2816, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 4.480558029088751e-05, | |
| "loss": 1.2934, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 4.4620065301276345e-05, | |
| "loss": 1.2981, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 4.443455031166518e-05, | |
| "loss": 1.2945, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 4.424903532205402e-05, | |
| "loss": 1.2848, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 4.4063520332442865e-05, | |
| "loss": 1.2948, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 4.38780053428317e-05, | |
| "loss": 1.2911, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 4.369249035322055e-05, | |
| "loss": 1.3008, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 4.3506975363609384e-05, | |
| "loss": 1.3118, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 4.332146037399822e-05, | |
| "loss": 1.3036, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 4.313594538438706e-05, | |
| "loss": 1.2999, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 4.29504303947759e-05, | |
| "loss": 1.2994, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 4.276491540516474e-05, | |
| "loss": 1.2935, | |
| "step": 10250 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 4.257940041555358e-05, | |
| "loss": 1.3001, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 4.239388542594242e-05, | |
| "loss": 1.3013, | |
| "step": 10750 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 4.2208370436331254e-05, | |
| "loss": 1.3073, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 4.20228554467201e-05, | |
| "loss": 1.3056, | |
| "step": 11250 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 4.1837340457108936e-05, | |
| "loss": 1.3076, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 4.165182546749777e-05, | |
| "loss": 1.2909, | |
| "step": 11750 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 4.146631047788662e-05, | |
| "loss": 1.3034, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 4.1280795488275455e-05, | |
| "loss": 1.2981, | |
| "step": 12250 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 4.109528049866429e-05, | |
| "loss": 1.2989, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 4.090976550905313e-05, | |
| "loss": 1.3018, | |
| "step": 12750 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 4.0724250519441975e-05, | |
| "loss": 1.2971, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 4.053873552983081e-05, | |
| "loss": 1.2975, | |
| "step": 13250 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 4.035322054021965e-05, | |
| "loss": 1.2909, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 4.0167705550608494e-05, | |
| "loss": 1.1078, | |
| "step": 13750 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 3.998219056099733e-05, | |
| "loss": 0.9997, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 3.979667557138617e-05, | |
| "loss": 1.0043, | |
| "step": 14250 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 3.961116058177501e-05, | |
| "loss": 0.9951, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 3.942564559216385e-05, | |
| "loss": 1.0098, | |
| "step": 14750 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 3.924013060255269e-05, | |
| "loss": 1.0087, | |
| "step": 15000 | |
| } | |
| ], | |
| "max_steps": 67880, | |
| "num_train_epochs": 10, | |
| "total_flos": 2.462482633434071e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |