| { |
| "schema_version": 1, |
| "status": "complete", |
| "config_fingerprint": "0667d4ea7a9eeb08101909c8f4b52860faadd491ce6d4ceb8094405e04a5f75e", |
| "dataset_fingerprint": "6cf9eb3d777a338eb007dcc5606b2aac6605bf899a191ea399ccd77d5527acfb", |
| "cursor": { |
| "global_step": 2540, |
| "micro_step": 20380, |
| "dataset_epoch": 1, |
| "sample_offset": 9440, |
| "accumulation_step": 0 |
| }, |
| "extra": { |
| "last_metrics": { |
| "global_step": 2540, |
| "behavior_loss": 0.014207982196239755, |
| "behavior_microbatches": 8, |
| "behavior_singleton_microbatches": 0, |
| "oom_replayed_as_singletons": false, |
| "gradient_norm": 0.01899070478975773, |
| "learning_rate_used": 1e-06, |
| "step_seconds": 12.963, |
| "peak_reserved_gib": 24.904, |
| "optimizer_state_device": "cpu_between_updates", |
| "weighted_behavior_objective": 0.014207982196239755, |
| "validation_behavior_loss": 0.01691136188339442, |
| "lr_window_phase": "continue", |
| "lr_window_start_step": 2490, |
| "lr_window_end_step": 2540, |
| "lr_window_point_count": 6, |
| "lr_window_log_slope_per_step": 0.00010029188921891751, |
| "lr_window_fitted_log_descent": -0.005014594460945876, |
| "lr_window_relative_descent": -0.005027188582420461, |
| "lr_window_residual_mad_scale": 0.0013824822312143686, |
| "lr_window_descent_to_noise": -3.6272397197764126, |
| "lr_window_accepted": false, |
| "lr_window_elbow_step": 2490, |
| "lr_training_window_start_step": 2500, |
| "lr_training_window_end_step": 2540, |
| "lr_training_window_point_count": 5, |
| "lr_training_window_log_slope_per_step": -0.003805868978774614, |
| "lr_training_window_fitted_log_descent": 0.15223475915098456, |
| "lr_training_window_relative_descent": 0.14121335095000065, |
| "lr_training_window_residual_mad_scale": 0.05720054641447164, |
| "lr_training_window_descent_to_noise": 2.6614214145421067, |
| "lr_training_window_accepted": true, |
| "lr_training_window_elbow_step": 2540, |
| "lr_window_any_signal_accepted": true, |
| "validation_interval": 10, |
| "next_validation_step": 2550, |
| "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": 2540, |
| "loss": 0.01691136188339442 |
| } |
| ], |
| "training_points": [], |
| "training_loss_sum": 0.0, |
| "training_loss_count": 0, |
| "last_training_step": 2540, |
| "window_start_step": 2540, |
| "next_validation_step": 2550, |
| "last_analysis": { |
| "start_step": 2490, |
| "end_step": 2540, |
| "point_count": 6, |
| "log_slope_per_step": 0.00010029188921891751, |
| "fitted_log_descent": -0.005014594460945876, |
| "relative_descent": -0.005027188582420461, |
| "residual_mad_scale": 0.0013824822312143686, |
| "descent_to_noise": -3.6272397197764126, |
| "accepted": false, |
| "elbow_step": 2490 |
| }, |
| "last_training_analysis": { |
| "start_step": 2500, |
| "end_step": 2540, |
| "point_count": 5, |
| "log_slope_per_step": -0.003805868978774614, |
| "fitted_log_descent": 0.15223475915098456, |
| "relative_descent": 0.14121335095000065, |
| "residual_mad_scale": 0.05720054641447164, |
| "descent_to_noise": 2.6614214145421067, |
| "accepted": true, |
| "elbow_step": 2540 |
| }, |
| "confirming": false, |
| "pending_rollback": null, |
| "reductions": 9 |
| } |
| }, |
| "rollback_replay": null |
| }, |
| "files": { |
| "adapters.safetensors": "f8fe52d0b625e933521e5d9346da158bd5462738c1af4ea3a046397027f25569", |
| "training_state.pt": "119a07a21135833df96d7ebab90824a25fcae0df32fcb8b8563bf34e6b5c1616" |
| } |
| } |
|
|
