We’re launching Lyria 3.5 in Google Flow Music, with advances across musicality, lyrics, vocals, and creative control
deepmind.google·6d ago
AI Engineering
Hacker News·Jun 22·5 min read
Manticore Search 27.1.5: Auth, sharding, conversational and faster vector search
TL;DR
Manticore Search previously lacked built-in user authentication and efficient data management for large deployments. The latest version, 27.1.5, introduces features like user authentication, sharded tables for better data distribution, and conversational search capabilities.
✦ Why It Matters
Engineers can leverage built-in authentication and sharding to simplify deployment and enhance data management.
Key Takeaways
- Implement built-in authentication for secure access control.
- Utilize sharded tables for efficient data management.
- Adopt conversational search for improved user interaction.
- Leverage faster HNSW builds for enhanced search performance.
Full Summary
Manticore Search is a powerful search engine that previously required external solutions for user authentication and data sharding, complicating deployment and management. Version 27.1.5 introduces built-in authentication and authorization, allowing users to manage access with usernames, passwords, and bearer tokens across various protocols like MySQL and HTTP/HTTPS.
Additionally, it now supports sharded tables, which distribute data across multiple segments, simplifying operations for write-heavy applications. The release also features conversational search, enabling users to interact with vectorized data directly without needing a separate retrieval layer.
Improvements in the Hierarchical Navigable Small World (HNSW) algorithm lead to faster builds, enhancing performance for nearest neighbor searches. Overall, these updates reduce the complexity of managing large datasets and improve user experience.
Engineers can expect a more integrated and efficient search solution with these advancements.