| { |
| "schema_version": 1, |
| "status": "complete", |
| "config_fingerprint": "0667d4ea7a9eeb08101909c8f4b52860faadd491ce6d4ceb8094405e04a5f75e", |
| "dataset_fingerprint": "6cf9eb3d777a338eb007dcc5606b2aac6605bf899a191ea399ccd77d5527acfb", |
| "cursor": { |
| "global_step": 2000, |
| "micro_step": 16049, |
| "dataset_epoch": 1, |
| "sample_offset": 800, |
| "accumulation_step": 0 |
| }, |
| "extra": { |
| "last_metrics": { |
| "global_step": 2000, |
| "behavior_loss": 0.008517224341630936, |
| "behavior_microbatches": 8, |
| "behavior_singleton_microbatches": 0, |
| "oom_replayed_as_singletons": false, |
| "gradient_norm": 0.003370017744600773, |
| "learning_rate_used": 1.25e-06, |
| "step_seconds": 11.96, |
| "peak_reserved_gib": 24.904, |
| "optimizer_state_device": "cpu_between_updates", |
| "weighted_behavior_objective": 0.008517224341630936, |
| "validation_behavior_loss": 0.016927855852991343, |
| "lr_window_phase": "continue", |
| "lr_window_start_step": 1950, |
| "lr_window_end_step": 2000, |
| "lr_window_point_count": 6, |
| "lr_window_log_slope_per_step": -5.001375998792135e-05, |
| "lr_window_fitted_log_descent": 0.0025006879993960673, |
| "lr_window_relative_descent": 0.002497563883849585, |
| "lr_window_residual_mad_scale": 0.002629185529678503, |
| "lr_window_descent_to_noise": 0.9511264880960499, |
| "lr_window_accepted": false, |
| "lr_window_elbow_step": 1960, |
| "lr_training_window_start_step": 1960, |
| "lr_training_window_end_step": 2000, |
| "lr_training_window_point_count": 5, |
| "lr_training_window_log_slope_per_step": -0.002990695905858872, |
| "lr_training_window_fitted_log_descent": 0.11962783623435488, |
| "lr_training_window_relative_descent": 0.11274942220382322, |
| "lr_training_window_residual_mad_scale": 0.020574252168350727, |
| "lr_training_window_descent_to_noise": 5.814443959152878, |
| "lr_training_window_accepted": true, |
| "lr_training_window_elbow_step": 2000, |
| "lr_window_any_signal_accepted": true, |
| "validation_interval": 10, |
| "next_validation_step": 2010, |
| "learning_rate": 1.25e-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": 1.25e-06, |
| "points": [ |
| { |
| "step": 2000, |
| "loss": 0.016927855852991343 |
| } |
| ], |
| "training_points": [], |
| "training_loss_sum": 0.0, |
| "training_loss_count": 0, |
| "last_training_step": 2000, |
| "window_start_step": 2000, |
| "next_validation_step": 2010, |
| "last_analysis": { |
| "start_step": 1950, |
| "end_step": 2000, |
| "point_count": 6, |
| "log_slope_per_step": -5.001375998792135e-05, |
| "fitted_log_descent": 0.0025006879993960673, |
| "relative_descent": 0.002497563883849585, |
| "residual_mad_scale": 0.002629185529678503, |
| "descent_to_noise": 0.9511264880960499, |
| "accepted": false, |
| "elbow_step": 1960 |
| }, |
| "last_training_analysis": { |
| "start_step": 1960, |
| "end_step": 2000, |
| "point_count": 5, |
| "log_slope_per_step": -0.002990695905858872, |
| "fitted_log_descent": 0.11962783623435488, |
| "relative_descent": 0.11274942220382322, |
| "residual_mad_scale": 0.020574252168350727, |
| "descent_to_noise": 5.814443959152878, |
| "accepted": true, |
| "elbow_step": 2000 |
| }, |
| "confirming": false, |
| "pending_rollback": null, |
| "reductions": 8 |
| } |
| }, |
| "rollback_replay": null |
| }, |
| "files": { |
| "adapters.safetensors": "ade88e4b6928ca2556a868b566e7bccd3d927e80a0d3847ebd4583fab10bc63d", |
| "training_state.pt": "64568575b5996863ee327362782da679e2c3c1a5187ffef62e75f7430280cdef" |
| } |
| } |
|
|
