| { |
| "schema_version": 1, |
| "status": "complete", |
| "config_fingerprint": "0667d4ea7a9eeb08101909c8f4b52860faadd491ce6d4ceb8094405e04a5f75e", |
| "dataset_fingerprint": "6cf9eb3d777a338eb007dcc5606b2aac6605bf899a191ea399ccd77d5527acfb", |
| "cursor": { |
| "global_step": 4020, |
| "micro_step": 32261, |
| "dataset_epoch": 2, |
| "sample_offset": 1920, |
| "accumulation_step": 0 |
| }, |
| "extra": { |
| "last_metrics": { |
| "global_step": 4020, |
| "behavior_loss": 0.0124094303464517, |
| "behavior_microbatches": 8, |
| "behavior_singleton_microbatches": 0, |
| "oom_replayed_as_singletons": false, |
| "gradient_norm": 0.014340076595544815, |
| "learning_rate_used": 1e-06, |
| "step_seconds": 12.155, |
| "peak_reserved_gib": 24.904, |
| "optimizer_state_device": "cpu_between_updates", |
| "weighted_behavior_objective": 0.0124094303464517, |
| "validation_behavior_loss": 0.016728750793263317, |
| "lr_window_phase": "continue", |
| "lr_window_start_step": 3970, |
| "lr_window_end_step": 4020, |
| "lr_window_point_count": 6, |
| "lr_window_log_slope_per_step": 2.857140227696675e-06, |
| "lr_window_fitted_log_descent": -0.00014285701138483375, |
| "lr_window_relative_descent": -0.0001428672159335953, |
| "lr_window_residual_mad_scale": 0.0018129892036015775, |
| "lr_window_descent_to_noise": -0.0787963938787073, |
| "lr_window_accepted": false, |
| "lr_window_elbow_step": 4010, |
| "lr_training_window_start_step": 3980, |
| "lr_training_window_end_step": 4020, |
| "lr_training_window_point_count": 5, |
| "lr_training_window_log_slope_per_step": -0.004031772364867914, |
| "lr_training_window_fitted_log_descent": 0.16127089459471655, |
| "lr_training_window_relative_descent": 0.14893850808069897, |
| "lr_training_window_residual_mad_scale": 0.024335189799852216, |
| "lr_training_window_descent_to_noise": 6.627065411082018, |
| "lr_training_window_accepted": true, |
| "lr_training_window_elbow_step": 4020, |
| "lr_window_any_signal_accepted": true, |
| "validation_interval": 10, |
| "next_validation_step": 4030, |
| "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": 4020, |
| "loss": 0.016728750793263317 |
| } |
| ], |
| "training_points": [], |
| "training_loss_sum": 0.0, |
| "training_loss_count": 0, |
| "last_training_step": 4020, |
| "window_start_step": 4020, |
| "next_validation_step": 4030, |
| "last_analysis": { |
| "start_step": 3970, |
| "end_step": 4020, |
| "point_count": 6, |
| "log_slope_per_step": 2.857140227696675e-06, |
| "fitted_log_descent": -0.00014285701138483375, |
| "relative_descent": -0.0001428672159335953, |
| "residual_mad_scale": 0.0018129892036015775, |
| "descent_to_noise": -0.0787963938787073, |
| "accepted": false, |
| "elbow_step": 4010 |
| }, |
| "last_training_analysis": { |
| "start_step": 3980, |
| "end_step": 4020, |
| "point_count": 5, |
| "log_slope_per_step": -0.004031772364867914, |
| "fitted_log_descent": 0.16127089459471655, |
| "relative_descent": 0.14893850808069897, |
| "residual_mad_scale": 0.024335189799852216, |
| "descent_to_noise": 6.627065411082018, |
| "accepted": true, |
| "elbow_step": 4020 |
| }, |
| "confirming": false, |
| "pending_rollback": null, |
| "reductions": 9 |
| } |
| }, |
| "rollback_replay": null |
| }, |
| "files": { |
| "adapters.safetensors": "fe1e1978fafed74732035f4a372e9deca4fbee6d60b1339bb89c093dcba35ea1", |
| "training_state.pt": "b5091e7e6a9feead01cb1a310fa0440ec3936c5b7f801fc04ede701907236c51" |
| } |
| } |
|
|
