What's New
New Features
- Contextual scrolling is now available for audio processing tasks and conversational analysis. This feature supports playback synchronization between an audio file and its corresponding text transcript. For more information, see the Contextual Scrolling template documentation. (#4489, #4479, #4518, #4509)
Enhancements
- The API now supports searching projects by title. (#4522)
- Added documentation for Generative AI templates. (#4486)
- Added documentation for
visibleWhenparameter: choices tag and view tag (#4443) - Stability improvements to ML backend (#286)
- New tooltips throughout the UI provide guidance on advanced features and configurations to improve labeling efficiency and quality (#4691, #4654, #4658, #4652, #4649)
Bug Fixes
- Fixed an issue that was preventing users from changing labels. (#4535)
- Fixed an issue where labels were missing from the UI when multiple labels were applied to the same text span. (#4428)
- Fixed an issue where the show/hide icon was not appearing when working in regions that were grouped by tools. (#4524)
- Fixed an issue where there was a noticeable degradation in UI performance when a large number of regions were rendered within a single task. (#4581)
- Fixed an issue that occurred when users selected keypoints and polygons within the same annotation. (#4391)
- Fixed issues to ensure more robust and uniform SSRF defenses (#4483)
- Fixed and issue in which pytests were failing (#4488)
- Fixed a security issue to ensure the secret key is securely set, persisting a new one if required (#4690). Special thanks to community member @Robbilie for identifying this issue and contributing to fix.
- Fixed an issue related to version selection for an imported ML backend. (#4363, #4703)
- Fixed an issue where users were able to import unsupported file types. (#4515)