| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.999941965062968, | |
| "global_step": 25845, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.903269491197524e-05, | |
| "loss": 2.3352, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.806538982395047e-05, | |
| "loss": 1.6234, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.709808473592571e-05, | |
| "loss": 1.4139, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.6130779647900954e-05, | |
| "loss": 1.3156, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.5163474559876186e-05, | |
| "loss": 1.2138, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.4196169471851423e-05, | |
| "loss": 1.1438, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.322886438382666e-05, | |
| "loss": 1.0925, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.22615592958019e-05, | |
| "loss": 1.0486, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.129425420777713e-05, | |
| "loss": 1.0017, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.032694911975237e-05, | |
| "loss": 0.9716, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.935964403172761e-05, | |
| "loss": 0.9455, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.8392338943702845e-05, | |
| "loss": 0.9172, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 3.742503385567808e-05, | |
| "loss": 0.8947, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 3.645772876765332e-05, | |
| "loss": 0.8728, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 3.549042367962855e-05, | |
| "loss": 0.8525, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 3.452311859160379e-05, | |
| "loss": 0.8435, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 3.3555813503579034e-05, | |
| "loss": 0.8124, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 3.258850841555427e-05, | |
| "loss": 0.7954, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 3.1621203327529504e-05, | |
| "loss": 0.7863, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 3.065389823950474e-05, | |
| "loss": 0.7622, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 2.968659315147998e-05, | |
| "loss": 0.7563, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 2.8719288063455214e-05, | |
| "loss": 0.7407, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 2.7751982975430452e-05, | |
| "loss": 0.7332, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 2.678467788740569e-05, | |
| "loss": 0.7212, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 2.5817372799380925e-05, | |
| "loss": 0.7079, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 2.4850067711356163e-05, | |
| "loss": 0.6991, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 2.38827626233314e-05, | |
| "loss": 0.6898, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 2.2915457535306635e-05, | |
| "loss": 0.6913, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 2.1948152447281877e-05, | |
| "loss": 0.6734, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 2.098084735925711e-05, | |
| "loss": 0.6651, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 2.0013542271232346e-05, | |
| "loss": 0.6523, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 1.9046237183207584e-05, | |
| "loss": 0.6501, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 1.8078932095182822e-05, | |
| "loss": 0.6342, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 1.711162700715806e-05, | |
| "loss": 0.6408, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 1.6144321919133294e-05, | |
| "loss": 0.6268, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 1.517701683110853e-05, | |
| "loss": 0.6075, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 1.420971174308377e-05, | |
| "loss": 0.6085, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 1.3242406655059007e-05, | |
| "loss": 0.6056, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 1.2275101567034243e-05, | |
| "loss": 0.5969, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 1.1307796479009481e-05, | |
| "loss": 0.5935, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 1.0340491390984717e-05, | |
| "loss": 0.5892, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 9.373186302959953e-06, | |
| "loss": 0.5869, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 8.405881214935191e-06, | |
| "loss": 0.579, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 7.438576126910428e-06, | |
| "loss": 0.5754, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 6.471271038885665e-06, | |
| "loss": 0.57, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 5.503965950860902e-06, | |
| "loss": 0.5651, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 4.536660862836139e-06, | |
| "loss": 0.5601, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 3.5693557748113754e-06, | |
| "loss": 0.5589, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 2.6020506867866125e-06, | |
| "loss": 0.5586, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 1.6347455987618494e-06, | |
| "loss": 0.5576, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 6.674405107370864e-07, | |
| "loss": 0.5455, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 25845, | |
| "total_flos": 2.4105988823764173e+17, | |
| "train_loss": 0.8042618184561056, | |
| "train_runtime": 26760.2018, | |
| "train_samples_per_second": 9.658, | |
| "train_steps_per_second": 0.966 | |
| } | |
| ], | |
| "max_steps": 25845, | |
| "num_train_epochs": 3, | |
| "total_flos": 2.4105988823764173e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |