v0.1.10
Highlights
Added LanceDB integrations for Python and TypeScript and a configuration-management API that distributes model changes to workers.
Features
- Weaviate document enrichment now supports asynchronous processing, chunking, and streaming. LanceDB table enrichment processes batches incrementally without materializing the entire table.
- Added
get_model()to the Python SDK and exposed queue-routing controls through Helm.
Bug fixes
- Fixed queued score-response formatting, dead-letter routing, and reconnect handling.
- LlamaIndex embedding now handles
BytesIOimages, and Weaviate classification enrichment validates its configuration.