| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 4.999864773495605, | |
| "global_step": 18485, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.864755206924534e-05, | |
| "loss": 2.1578, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.729510413849067e-05, | |
| "loss": 1.8998, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.5942656207736005e-05, | |
| "loss": 1.7979, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.459020827698134e-05, | |
| "loss": 1.7267, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.3237760346226674e-05, | |
| "loss": 1.6795, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 4.188531241547201e-05, | |
| "loss": 1.637, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 4.053286448471734e-05, | |
| "loss": 1.6087, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 3.918041655396267e-05, | |
| "loss": 1.5754, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 3.782796862320801e-05, | |
| "loss": 1.5461, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 3.647552069245334e-05, | |
| "loss": 1.5271, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 3.512307276169868e-05, | |
| "loss": 1.5077, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 3.377062483094401e-05, | |
| "loss": 1.4905, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 3.241817690018935e-05, | |
| "loss": 1.4797, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 3.1065728969434676e-05, | |
| "loss": 1.4601, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 2.9713281038680014e-05, | |
| "loss": 1.4501, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 2.8360833107925345e-05, | |
| "loss": 1.4295, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 2.700838517717068e-05, | |
| "loss": 1.4207, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 2.5655937246416014e-05, | |
| "loss": 1.409, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 2.4303489315661348e-05, | |
| "loss": 1.3997, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 2.2951041384906682e-05, | |
| "loss": 1.3922, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 2.1598593454152017e-05, | |
| "loss": 1.385, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 2.024614552339735e-05, | |
| "loss": 1.3686, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 1.8893697592642685e-05, | |
| "loss": 1.3558, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 1.754124966188802e-05, | |
| "loss": 1.3533, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 1.618880173113335e-05, | |
| "loss": 1.3479, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 1.4836353800378685e-05, | |
| "loss": 1.3409, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 1.3483905869624019e-05, | |
| "loss": 1.3362, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 1.2131457938869355e-05, | |
| "loss": 1.3266, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 1.0779010008114688e-05, | |
| "loss": 1.3236, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 9.426562077360022e-06, | |
| "loss": 1.3211, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 4.19, | |
| "learning_rate": 8.074114146605356e-06, | |
| "loss": 1.3162, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 4.33, | |
| "learning_rate": 6.72166621585069e-06, | |
| "loss": 1.3068, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 4.46, | |
| "learning_rate": 5.369218285096024e-06, | |
| "loss": 1.2994, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 4.6, | |
| "learning_rate": 4.016770354341358e-06, | |
| "loss": 1.3019, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 4.73, | |
| "learning_rate": 2.664322423586692e-06, | |
| "loss": 1.2935, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 4.87, | |
| "learning_rate": 1.311874492832026e-06, | |
| "loss": 1.2931, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "step": 18485, | |
| "total_flos": 1.5583311452399616e+17, | |
| "train_loss": 1.463753378478327, | |
| "train_runtime": 53109.6475, | |
| "train_samples_per_second": 11.138, | |
| "train_steps_per_second": 0.348 | |
| } | |
| ], | |
| "max_steps": 18485, | |
| "num_train_epochs": 5, | |
| "total_flos": 1.5583311452399616e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |