| { |
| "best_global_step": null, |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 5.0, |
| "eval_steps": 500, |
| "global_step": 80, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.32, |
| "grad_norm": 3.0088891059134757, |
| "learning_rate": 2e-05, |
| "loss": 0.7923, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.04150344803929329, |
| "step": 5, |
| "valid_targets_mean": 4567.2, |
| "valid_targets_min": 2907 |
| }, |
| { |
| "epoch": 0.64, |
| "grad_norm": 0.9748835611922116, |
| "learning_rate": 3.998096443163716e-05, |
| "loss": 0.6915, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.037823863327503204, |
| "step": 10, |
| "valid_targets_mean": 4694.8, |
| "valid_targets_min": 1740 |
| }, |
| { |
| "epoch": 0.96, |
| "grad_norm": 0.5292027668233105, |
| "learning_rate": 3.931851652578137e-05, |
| "loss": 0.6339, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.03384627401828766, |
| "step": 15, |
| "valid_targets_mean": 6716.5, |
| "valid_targets_min": 3779 |
| }, |
| { |
| "epoch": 1.256, |
| "grad_norm": 0.381874909525484, |
| "learning_rate": 3.774021666356444e-05, |
| "loss": 0.5937, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.0336514413356781, |
| "step": 20, |
| "valid_targets_mean": 4429.5, |
| "valid_targets_min": 1506 |
| }, |
| { |
| "epoch": 1.576, |
| "grad_norm": 0.3191036829485198, |
| "learning_rate": 3.532088886237956e-05, |
| "loss": 0.5791, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.03018685057759285, |
| "step": 25, |
| "valid_targets_mean": 6942.8, |
| "valid_targets_min": 5230 |
| }, |
| { |
| "epoch": 1.896, |
| "grad_norm": 0.2936828471213373, |
| "learning_rate": 3.217522858017442e-05, |
| "loss": 0.5804, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.02674776315689087, |
| "step": 30, |
| "valid_targets_mean": 3804.5, |
| "valid_targets_min": 1421 |
| }, |
| { |
| "epoch": 2.192, |
| "grad_norm": 0.25770700064193613, |
| "learning_rate": 2.8452365234813992e-05, |
| "loss": 0.569, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.03195125609636307, |
| "step": 35, |
| "valid_targets_mean": 4945.2, |
| "valid_targets_min": 929 |
| }, |
| { |
| "epoch": 2.512, |
| "grad_norm": 0.2547784828064005, |
| "learning_rate": 2.4328792278762058e-05, |
| "loss": 0.5312, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.030446790158748627, |
| "step": 40, |
| "valid_targets_mean": 5575.0, |
| "valid_targets_min": 3849 |
| }, |
| { |
| "epoch": 2.832, |
| "grad_norm": 0.25899697668472715, |
| "learning_rate": 2e-05, |
| "loss": 0.5487, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.03602331131696701, |
| "step": 45, |
| "valid_targets_mean": 5869.8, |
| "valid_targets_min": 3374 |
| }, |
| { |
| "epoch": 3.128, |
| "grad_norm": 0.22099844272575836, |
| "learning_rate": 1.5671207721237945e-05, |
| "loss": 0.5284, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.04063715040683746, |
| "step": 50, |
| "valid_targets_mean": 8575.2, |
| "valid_targets_min": 5225 |
| }, |
| { |
| "epoch": 3.448, |
| "grad_norm": 0.2275701442675398, |
| "learning_rate": 1.1547634765186016e-05, |
| "loss": 0.5323, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.02942601963877678, |
| "step": 55, |
| "valid_targets_mean": 6379.2, |
| "valid_targets_min": 1888 |
| }, |
| { |
| "epoch": 3.768, |
| "grad_norm": 0.2236212833364483, |
| "learning_rate": 7.824771419825588e-06, |
| "loss": 0.5299, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.034093089401721954, |
| "step": 60, |
| "valid_targets_mean": 5950.8, |
| "valid_targets_min": 4866 |
| }, |
| { |
| "epoch": 4.064, |
| "grad_norm": 0.2179261363674421, |
| "learning_rate": 4.679111137620442e-06, |
| "loss": 0.5101, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.027605395764112473, |
| "step": 65, |
| "valid_targets_mean": 5062.5, |
| "valid_targets_min": 4995 |
| }, |
| { |
| "epoch": 4.384, |
| "grad_norm": 0.23062712401292512, |
| "learning_rate": 2.259783336435566e-06, |
| "loss": 0.5329, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.0404040664434433, |
| "step": 70, |
| "valid_targets_mean": 6195.2, |
| "valid_targets_min": 5413 |
| }, |
| { |
| "epoch": 4.704, |
| "grad_norm": 0.2295496459564114, |
| "learning_rate": 6.814834742186361e-07, |
| "loss": 0.5152, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.02953474223613739, |
| "step": 75, |
| "valid_targets_mean": 5352.2, |
| "valid_targets_min": 4831 |
| }, |
| { |
| "epoch": 5.0, |
| "grad_norm": 0.2811848755096405, |
| "learning_rate": 1.9035568362844037e-08, |
| "loss": 0.509, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.06422175467014313, |
| "step": 80, |
| "valid_targets_mean": 8112.2, |
| "valid_targets_min": 6354 |
| }, |
| { |
| "epoch": 5.0, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.06422175467014313, |
| "step": 80, |
| "total_flos": 2.548584738193408e+17, |
| "train_loss": 0.5735973387956619, |
| "train_runtime": 10563.2448, |
| "train_samples_per_second": 0.473, |
| "train_steps_per_second": 0.008, |
| "valid_targets_mean": 8112.2, |
| "valid_targets_min": 6354 |
| } |
| ], |
| "logging_steps": 5, |
| "max_steps": 80, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 5, |
| "save_steps": 500, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": false, |
| "should_training_stop": false |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 2.548584738193408e+17, |
| "train_batch_size": 1, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
