Version 2.4.0.0
Version 2.4.0.0 Release Notes
Compatible with OpenSearch 2.4.0
Experimental Features
- Add MLCommonsClientAccessor and MLPredict TransportAction for accessing the MLClient's predict API (#16)
- Add parsing logic for neural query (#15)
- Integrate model inference to build neural query (#20)
- Add text embedding processor to neural search (#18)
Enhancements
- Change text embedding processor to async mode for better isolation (#27)
Bug Fixes
- Update the model function name from CUSTOM to TEXT_EMBEDDING as per the latest changes in MLCommons (#17)
- Fix the locale changes from locale.default to locale.ROOT to fix the tests failing on Windows (#43)
Infrastructure
- Initial commit for setting up the neural search plugin (#2)
- Fix CI and Link Checker GitHub workflows (#3)
- Enable the K-NN plugin and ML plugin for integ test cluster (#6)
- Add dependency on k-NN plugin (#10)
- Switch pull_request_target to pull_request in CI (#26)
- Fix minor build.gradle issues (#28)
- Fix group id for ml-commons dependency (#32)
- Add Windows support for CI (#40)
- Add opensearch prefix to plugin name (#38)
- Add integration tests for neural query (#36)
- Switch processor IT to use Lucene (#48)
- Add release note draft automation (#52)
Documentation
Maintenance
- Upgrade plugin to 2.4 and refactor zip dependencies (#25)
Refactoring
- Refactor project package structure (#55)