You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
feat(cli): train start auto-exports version in required format
Before training, the handler now:
1. Determines the required export format for the model type using
get_model_format() (e.g., rfdetr needs coco, yolov8 needs yolov5pytorch)
2. Checks if the version is still generating; waits with progress updates
3. Checks if the version has the required export; triggers and polls if not
4. Then starts training
Also improves the "Unknown error" from the train API — adds a hint
suggesting the version may not be exported yet.
This prevents the confusing failure mode where `train start` returns
"Unknown error" because the server expects an export that doesn't exist.
405 tests pass, all linting clean.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
0 commit comments