This release incorporate the following:
- Improvements in the main training script.
- Bug fixes, minor enhancements, and improved documentation in the
codebook_features/code_search_cache.pyscript. - Add a "Top K" slider in the webapp to highlight only those tokens for which a code was in its top K most-similar codes
- Refactoring and code simplifications in the
codebook_features/models.pyscript.
The release is compatible with models trained using the previous release.