Expand regional corpus and aerial finetuning controls

2026-07-27 06:19:11 +02:00
parent 8a4a995659
commit 2316ff28f5
6 changed files with 105 additions and 2 deletions
@@ -80,6 +80,10 @@ The active production model remains unchanged while any gate fails.

 Every failed assessment returns `continue_training_loop`. Only a report with
 `training_complete` may proceed to final human review and guarded activation.
+Optimizer, initial learning rate, image size and geometric augmentation are
+explicit loop inputs. This permits a conservative aerial-imagery finetune
+(for example AdamW with mosaic disabled) without changing calibration, test
+or release gates.

 After a failed assessment,
 `scripts/build_failure_driven_yolo_sampling.py` creates a checksummed,