| { |
| "schema_version": 1, |
| "status": "complete", |
| "config_fingerprint": "0667d4ea7a9eeb08101909c8f4b52860faadd491ce6d4ceb8094405e04a5f75e", |
| "dataset_fingerprint": "6cf9eb3d777a338eb007dcc5606b2aac6605bf899a191ea399ccd77d5527acfb", |
| "cursor": { |
| "global_step": 3010, |
| "micro_step": 24154, |
| "dataset_epoch": 1, |
| "sample_offset": 16960, |
| "accumulation_step": 0 |
| }, |
| "extra": { |
| "last_metrics": { |
| "global_step": 3010, |
| "behavior_loss": 0.006600444903597236, |
| "behavior_microbatches": 8, |
| "behavior_singleton_microbatches": 0, |
| "oom_replayed_as_singletons": false, |
| "gradient_norm": 0.007156455423682928, |
| "learning_rate_used": 1e-06, |
| "step_seconds": 14.206, |
| "peak_reserved_gib": 24.904, |
| "optimizer_state_device": "cpu_between_updates", |
| "weighted_behavior_objective": 0.006600444903597236, |
| "validation_behavior_loss": 0.01693202504888177, |
| "lr_window_phase": "minimum_lr_plateau", |
| "lr_window_start_step": 2940, |
| "lr_window_end_step": 3010, |
| "lr_window_point_count": 8, |
| "lr_window_log_slope_per_step": 1.7980951354151945e-05, |
| "lr_window_fitted_log_descent": -0.001258666594790636, |
| "lr_window_relative_descent": -0.0012594590480323742, |
| "lr_window_residual_mad_scale": 0.002189475291935303, |
| "lr_window_descent_to_noise": -0.5748713399172849, |
| "lr_window_accepted": false, |
| "lr_window_elbow_step": 3000, |
| "lr_training_window_start_step": 2950, |
| "lr_training_window_end_step": 3010, |
| "lr_training_window_point_count": 7, |
| "lr_training_window_log_slope_per_step": -0.00010182409525911022, |
| "lr_training_window_fitted_log_descent": 0.006109445715546613, |
| "lr_training_window_relative_descent": 0.006090821000269142, |
| "lr_training_window_residual_mad_scale": 0.018786415112912798, |
| "lr_training_window_descent_to_noise": 0.3252055104088113, |
| "lr_training_window_accepted": false, |
| "lr_training_window_elbow_step": 3000, |
| "lr_window_any_signal_accepted": false, |
| "validation_interval": 10, |
| "next_validation_step": 3020, |
| "learning_rate": 1e-06 |
| }, |
| "reason": "periodic", |
| "parent": null, |
| "lr_control": { |
| "version": 3, |
| "config": { |
| "factor": 0.5, |
| "minimum_learning_rate": 1e-06, |
| "window_steps": 50, |
| "probe_every_steps": 10, |
| "confirmation_steps": 20, |
| "minimum_descent_to_noise": 1.0, |
| "minimum_relative_descent": 0.0 |
| }, |
| "state": { |
| "learning_rate": 1e-06, |
| "points": [ |
| { |
| "step": 3010, |
| "loss": 0.01693202504888177 |
| } |
| ], |
| "training_points": [], |
| "training_loss_sum": 0.0, |
| "training_loss_count": 0, |
| "last_training_step": 3010, |
| "window_start_step": 3010, |
| "next_validation_step": 3020, |
| "last_analysis": { |
| "start_step": 2940, |
| "end_step": 3010, |
| "point_count": 8, |
| "log_slope_per_step": 1.7980951354151945e-05, |
| "fitted_log_descent": -0.001258666594790636, |
| "relative_descent": -0.0012594590480323742, |
| "residual_mad_scale": 0.002189475291935303, |
| "descent_to_noise": -0.5748713399172849, |
| "accepted": false, |
| "elbow_step": 3000 |
| }, |
| "last_training_analysis": { |
| "start_step": 2950, |
| "end_step": 3010, |
| "point_count": 7, |
| "log_slope_per_step": -0.00010182409525911022, |
| "fitted_log_descent": 0.006109445715546613, |
| "relative_descent": 0.006090821000269142, |
| "residual_mad_scale": 0.018786415112912798, |
| "descent_to_noise": 0.3252055104088113, |
| "accepted": false, |
| "elbow_step": 3000 |
| }, |
| "confirming": false, |
| "pending_rollback": null, |
| "reductions": 9 |
| } |
| }, |
| "rollback_replay": null |
| }, |
| "files": { |
| "adapters.safetensors": "605e25f84b688ad82da52f79c11c042499db66cc68aa5ecc8f17bc4798546d45", |
| "training_state.pt": "f8e888df22abf4325ff4bf080c6440602e51ad02fe556d8336cca0259f88d8eb" |
| } |
| } |
|
|
