Fixes:
- Resolved the issue with cosine vector similarity by normalizing the results.
Breaking changes:
- Deprecated text truncation.
- Added autologic to generate average embeddings when the text size exceeds the limit for the model we're using.
Fixes:
Breaking changes: