September 2023 Summaries

Weaviate has released a pre-release version of its new Python client API with strong typing, collections-first approach, and performance improvements. The client is designed to be more user-friendly and prioritizes client libraries for different programming languages. Key changes include strong typing through custom Python classes and a focus on individual collections for interaction. Other improvements include simplified methods, easier defaults, batch import error handling, and collection-level iterator. Performance enhancements include up to 60-80% faster imports and 3-4x faster queries. The new client is available for testing and feedback through the Experimental clients page.

Weaviate has integrated with Confluent Cloud, allowing users to access real-time data streams from across their entire business for building AI applications. This integration breaks down data silos, promotes reusability, and fosters trust throughout organizations. It enables the creation of a real-time knowledge base, real-time context at query time, governed, secured, and trusted AI, and faster experimentation, scaling, and innovation. The integration architecture is designed to be both robust and straightforward, ensuring seamless data flow from Confluent Cloud to Weaviate Cloud.

Product Quantization (PQ) is a technique used in Weaviate to compress vectors, reducing memory requirements while maintaining high search quality. PQ compression involves replacing exact vector coordinates with learned codes that represent general regions where the vector can be found. This results in a lossy algorithm, as some information is lost during compression. However, by adjusting the level of compression, users can balance memory usage and recall for their specific use case. In Weaviate v1.21, significant improvements were made to PQ, reducing recall loss while still achieving high compression ratios. These improvements include a rescoring trick that reads uncompressed vectors from disk during search to recalculate exact distances, improving recall without sacrificing performance.

Researchers have developed a new direction for Language Learning Models (LLMs) by connecting them with tools such as APIs. One example is WeaviateGraphQLGorilla, which translates natural language commands into Weaviate GraphQL Search APIs. The model was trained on a novel dataset of 46 Weaviate GraphQL Search APIs customized to a set of 50 synthetic Database schemas, totaling 2,300 examples. This project aims to help users access the power of Weaviate without needing to learn its syntax and can be integrated into the Auto API for future use.