·
1 commit
to update.redisearch
since this release
What’s Changed
- Expanded dataset support (LAION up to 400M, DBpedia multiple dims, Cohere, 768-1G cosine)
- Added Redis Vector Sets + SVS-VAMANA (replacing SVS/SVS_TIERED)
- Improved benchmarking accuracy (calibration, timing fixes, precision summary: QPS/P50/P95)
- Scalability and performance fixes
- Enhanced dataset handling (auto-discovery, validation, .bz2 unzip, S3 fallback)
- Added memory reporting and improved monitoring
- CLI improvements (
--only-configure,--engines-file,--describe) - Docker updates, permissions fixes, org update to
redis - Added PyPI publishing support and tests
New Contributors
Full Changelog: https://github.com/redis-performance/vector-db-benchmark/commits/0.1.0