Weaviate 1.39 adds 4-bit Rotational Quantization
Weaviate says the new 4-bit option reduces memory use while maintaining strong recall, alongside faster imports and search improvements across its Rotational Quantization family.
TLDR
Weaviate announced 4-bit Rotational Quantization (RQ) in version 1.39 for users prioritizing lower RAM use and cost. The company says 8-bit RQ remains the default in Weaviate Cloud and now imports 16% faster. It describes its accompanying blog as covering the performance work, recall as datasets scale and comparisons with TurboQuant.
Combined views
2.7K
4 Sources, first seen ago