| { |
| "schema_version": 1, |
| "status": "complete", |
| "config_fingerprint": "0667d4ea7a9eeb08101909c8f4b52860faadd491ce6d4ceb8094405e04a5f75e", |
| "dataset_fingerprint": "6cf9eb3d777a338eb007dcc5606b2aac6605bf899a191ea399ccd77d5527acfb", |
| "cursor": { |
| "global_step": 5000, |
| "micro_step": 40124, |
| "dataset_epoch": 2, |
| "sample_offset": 17600, |
| "accumulation_step": 0 |
| }, |
| "extra": { |
| "last_metrics": { |
| "global_step": 5000, |
| "behavior_loss": 0.016017724061384797, |
| "behavior_microbatches": 8, |
| "behavior_singleton_microbatches": 0, |
| "oom_replayed_as_singletons": false, |
| "gradient_norm": 0.012397946789860725, |
| "learning_rate_used": 1e-06, |
| "step_seconds": 11.805, |
| "peak_reserved_gib": 24.904, |
| "optimizer_state_device": "cpu_between_updates", |
| "weighted_behavior_objective": 0.016017724061384797, |
| "validation_behavior_loss": 0.016737566627562046, |
| "lr_window_phase": "collecting", |
| "validation_interval": 10, |
| "next_validation_step": 5010, |
| "learning_rate": 1e-06 |
| }, |
| "reason": "session_limit", |
| "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": 4960, |
| "loss": 0.016812349539250135 |
| }, |
| { |
| "step": 4970, |
| "loss": 0.016733078290708362 |
| }, |
| { |
| "step": 4980, |
| "loss": 0.016717635430395603 |
| }, |
| { |
| "step": 4990, |
| "loss": 0.016753521063365043 |
| }, |
| { |
| "step": 5000, |
| "loss": 0.016737566627562046 |
| } |
| ], |
| "training_points": [ |
| { |
| "step": 4970, |
| "loss": 0.015235144022153691 |
| }, |
| { |
| "step": 4980, |
| "loss": 0.013825824274681509 |
| }, |
| { |
| "step": 4990, |
| "loss": 0.013259278662735597 |
| }, |
| { |
| "step": 5000, |
| "loss": 0.013512403922504745 |
| } |
| ], |
| "training_loss_sum": 0.0, |
| "training_loss_count": 0, |
| "last_training_step": 5000, |
| "window_start_step": 4960, |
| "next_validation_step": 5010, |
| "last_analysis": { |
| "start_step": 4890, |
| "end_step": 4960, |
| "point_count": 8, |
| "log_slope_per_step": 4.04361902428918e-05, |
| "fitted_log_descent": -0.0028305333170024257, |
| "relative_descent": -0.002834543058775063, |
| "residual_mad_scale": 0.0015034014177076266, |
| "descent_to_noise": -1.882752858726446, |
| "accepted": false, |
| "elbow_step": 4890 |
| }, |
| "last_training_analysis": { |
| "start_step": 4900, |
| "end_step": 4960, |
| "point_count": 7, |
| "log_slope_per_step": -0.00015988120629852886, |
| "fitted_log_descent": 0.009592872377911732, |
| "relative_descent": 0.009547007553316611, |
| "residual_mad_scale": 0.128691405423272, |
| "descent_to_noise": 0.07454167080047287, |
| "accepted": false, |
| "elbow_step": 4900 |
| }, |
| "confirming": false, |
| "pending_rollback": null, |
| "reductions": 9 |
| } |
| }, |
| "rollback_replay": null |
| }, |
| "files": { |
| "adapters.safetensors": "39925cea8660d30a05ffe4b899903351d4c53b80d950efc0e1667837c7a1cb9c", |
| "training_state.pt": "61bd503f76b53837fbac2fccf9ba69662dace1d8e39c4b97ff8663703a58ccb2" |
| } |
| } |
|
|