| { |
| "schema_version": 1, |
| "status": "complete", |
| "config_fingerprint": "0667d4ea7a9eeb08101909c8f4b52860faadd491ce6d4ceb8094405e04a5f75e", |
| "dataset_fingerprint": "6cf9eb3d777a338eb007dcc5606b2aac6605bf899a191ea399ccd77d5527acfb", |
| "cursor": { |
| "global_step": 1000, |
| "micro_step": 8021, |
| "dataset_epoch": 0, |
| "sample_offset": 16000, |
| "accumulation_step": 0 |
| }, |
| "extra": { |
| "last_metrics": { |
| "global_step": 1000, |
| "behavior_loss": 0.011761117406422272, |
| "behavior_microbatches": 8, |
| "behavior_singleton_microbatches": 0, |
| "oom_replayed_as_singletons": false, |
| "gradient_norm": 0.031379587948322296, |
| "learning_rate_used": 2e-05, |
| "step_seconds": 12.7, |
| "peak_reserved_gib": 24.463, |
| "optimizer_state_device": "cpu_between_updates", |
| "weighted_behavior_objective": 0.011761117406422272, |
| "validation_behavior_loss": 0.016935308971442284, |
| "lr_window_phase": "continue", |
| "lr_window_start_step": 950, |
| "lr_window_end_step": 1000, |
| "lr_window_point_count": 6, |
| "lr_window_log_slope_per_step": 0.00016107469412269816, |
| "lr_window_fitted_log_descent": -0.008053734706134907, |
| "lr_window_relative_descent": -0.008086253267493904, |
| "lr_window_residual_mad_scale": 0.0012714376993325613, |
| "lr_window_descent_to_noise": -6.334352607573851, |
| "lr_window_accepted": false, |
| "lr_window_elbow_step": 960, |
| "lr_training_window_start_step": 960, |
| "lr_training_window_end_step": 1000, |
| "lr_training_window_point_count": 5, |
| "lr_training_window_log_slope_per_step": -0.0035170239046830257, |
| "lr_training_window_fitted_log_descent": 0.14068095618732102, |
| "lr_training_window_relative_descent": 0.13123355795504654, |
| "lr_training_window_residual_mad_scale": 0.025747381477782865, |
| "lr_training_window_descent_to_noise": 5.463893728716183, |
| "lr_training_window_accepted": true, |
| "lr_training_window_elbow_step": 1000, |
| "lr_window_any_signal_accepted": true, |
| "validation_interval": 10, |
| "next_validation_step": 1010, |
| "learning_rate": 2e-05 |
| }, |
| "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": 2e-05, |
| "points": [ |
| { |
| "step": 1000, |
| "loss": 0.016935308971442284 |
| } |
| ], |
| "training_points": [], |
| "training_loss_sum": 0.0, |
| "training_loss_count": 0, |
| "last_training_step": 1000, |
| "window_start_step": 1000, |
| "next_validation_step": 1010, |
| "last_analysis": { |
| "start_step": 950, |
| "end_step": 1000, |
| "point_count": 6, |
| "log_slope_per_step": 0.00016107469412269816, |
| "fitted_log_descent": -0.008053734706134907, |
| "relative_descent": -0.008086253267493904, |
| "residual_mad_scale": 0.0012714376993325613, |
| "descent_to_noise": -6.334352607573851, |
| "accepted": false, |
| "elbow_step": 960 |
| }, |
| "last_training_analysis": { |
| "start_step": 960, |
| "end_step": 1000, |
| "point_count": 5, |
| "log_slope_per_step": -0.0035170239046830257, |
| "fitted_log_descent": 0.14068095618732102, |
| "relative_descent": 0.13123355795504654, |
| "residual_mad_scale": 0.025747381477782865, |
| "descent_to_noise": 5.463893728716183, |
| "accepted": true, |
| "elbow_step": 1000 |
| }, |
| "confirming": false, |
| "pending_rollback": null, |
| "reductions": 4 |
| } |
| }, |
| "rollback_replay": null |
| }, |
| "files": { |
| "adapters.safetensors": "4a86181524ab17680c500add76f183acd257840b8a0d18e33807ae379d52925e", |
| "training_state.pt": "e8d40cf90248f643e98c0c434cb4871e8df7f4264e991c2ddf18a609dbfdabaf" |
| } |
| } |
|
|
