| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 5.0, | |
| "eval_steps": 500, | |
| "global_step": 29290, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.9800000000000004e-05, | |
| "loss": 0.0133, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.996339134609366e-05, | |
| "loss": 0.0228, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.985249132519566e-05, | |
| "loss": 0.0218, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.966762760626733e-05, | |
| "loss": 0.0226, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.9409939710148224e-05, | |
| "loss": 0.0242, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.9079161983362835e-05, | |
| "loss": 0.0263, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.867672207018106e-05, | |
| "loss": 0.029, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.8203817675244414e-05, | |
| "loss": 0.0297, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 4.76618562131017e-05, | |
| "loss": 0.029, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 4.7055020252691094e-05, | |
| "loss": 0.0306, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.6380242832979805e-05, | |
| "loss": 0.0313, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.5641835501722764e-05, | |
| "loss": 0.0259, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 4.484199583887917e-05, | |
| "loss": 0.0195, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 4.398310425366118e-05, | |
| "loss": 0.0225, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 4.306771690017899e-05, | |
| "loss": 0.0187, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 4.2100548160331335e-05, | |
| "loss": 0.0244, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 4.1080600968516616e-05, | |
| "loss": 0.0216, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 4.001279617165866e-05, | |
| "loss": 0.0247, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 3.890257905202517e-05, | |
| "loss": 0.025, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 3.774880507376146e-05, | |
| "loss": 0.0245, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 3.655708927484144e-05, | |
| "loss": 0.0312, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 3.5330978330177316e-05, | |
| "loss": 0.0252, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 3.407412127834942e-05, | |
| "loss": 0.0255, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 3.2790258661673314e-05, | |
| "loss": 0.0213, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 3.148321139394187e-05, | |
| "loss": 0.0235, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 3.0156869388973275e-05, | |
| "loss": 0.0227, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 2.8815179983807373e-05, | |
| "loss": 0.0201, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 2.746213619100438e-05, | |
| "loss": 0.0177, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 2.6101764815008288e-05, | |
| "loss": 0.021, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 2.4740842343999416e-05, | |
| "loss": 0.0199, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 2.3377965787034856e-05, | |
| "loss": 0.0193, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 2.2019916579053264e-05, | |
| "loss": 0.0194, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 2.0670736421154722e-05, | |
| "loss": 0.0217, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 1.9337097701276256e-05, | |
| "loss": 0.0195, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 1.8017625555614108e-05, | |
| "loss": 0.0192, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 1.6718933710332394e-05, | |
| "loss": 0.0167, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 1.544488721260973e-05, | |
| "loss": 0.0137, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 1.420173812196476e-05, | |
| "loss": 0.0139, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 1.2990597437888902e-05, | |
| "loss": 0.0139, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 1.1812737544610991e-05, | |
| "loss": 0.0127, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "learning_rate": 1.0674124368506597e-05, | |
| "loss": 0.0138, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "learning_rate": 9.578146545347833e-06, | |
| "loss": 0.0137, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 8.528065823492207e-06, | |
| "loss": 0.0139, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "learning_rate": 7.527007356567281e-06, | |
| "loss": 0.0142, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "learning_rate": 6.577950402675978e-06, | |
| "loss": 0.0113, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 5.683719457802727e-06, | |
| "loss": 0.0129, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 4.846975849808403e-06, | |
| "loss": 0.012, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 4.070209818031235e-06, | |
| "loss": 0.0091, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 4.18, | |
| "learning_rate": 3.355733102065542e-06, | |
| "loss": 0.0091, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 4.27, | |
| "learning_rate": 2.7056720617751495e-06, | |
| "loss": 0.0107, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "learning_rate": 2.121961349016713e-06, | |
| "loss": 0.0125, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "learning_rate": 1.6063381499069525e-06, | |
| "loss": 0.011, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 4.52, | |
| "learning_rate": 1.160337014769089e-06, | |
| "loss": 0.0093, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "learning_rate": 7.852852911454378e-07, | |
| "loss": 0.0127, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 4.69, | |
| "learning_rate": 4.822991734677207e-07, | |
| "loss": 0.0101, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "learning_rate": 2.52667105948437e-07, | |
| "loss": 0.0083, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 4.87, | |
| "learning_rate": 9.615235819493418e-08, | |
| "loss": 0.0098, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 4.95, | |
| "learning_rate": 1.3754150416017686e-08, | |
| "loss": 0.0105, | |
| "step": 29000 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 29290, | |
| "num_train_epochs": 5, | |
| "save_steps": 500, | |
| "total_flos": 2.4830219073675264e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |