Hierarchical Navigable Small World graphs
PulseAugur coverage of Hierarchical Navigable Small World graphs — every cluster mentioning Hierarchical Navigable Small World graphs across labs, papers, and developer communities, ranked by signal.
7 day(s) with sentiment data
-
New research reveals fundamental limits in RAG defenses against coordinated poisoning attacks
Researchers have demonstrated a fundamental limitation in current defenses against coordinated poisoning attacks on vector retrieval systems used in retrieval-augmented generation (RAG). These admission-time defenses, w…
-
Vector database admission control tackles retrieval hubs under workload drift
Researchers have developed a new admission control mechanism for vector databases to mitigate the impact of retrieval hubs, where a single document dominates search results. This system maintains a set of sentinel queri…
-
Cerebras Knowledge Base Evolves with MCP Server and Refined Retrieval
This series of posts details the development of a knowledge base system for Cerebras, focusing on its retrieval and agent capabilities. Initially, the system used a hybrid retrieval method with an LLM reranker, achievin…
-
Metadata design for retrieval systems prioritizes data recoverability and query efficiency
Designing metadata for retrieval systems requires careful consideration of data recoverability and query performance. Fields should be categorized based on whether they are unrecoverable after ingestion, expensive to re…
-
RAG Systems Enhanced with Hybrid Search and Reranking Beyond Vector Search
This article delves into enhancing Retrieval-Augmented Generation (RAG) systems by moving beyond simple vector search. It explains that while embeddings are crucial for semantic similarity, they are insufficient on thei…
-
PostgreSQL vector search faces performance cliff with filters
A technical deep-dive explains a performance issue in PostgreSQL when combining filtering with vector search using the pgvector extension. The problem arises because approximate nearest neighbor (ANN) indexes like HNSW …
-
GPU acceleration cuts HNSW vector search time by 57.7%
Researchers have optimized the Hierarchical Navigable Small World (HNSW) algorithm, a core component in many vector databases and RAG systems, for GPU acceleration using CUDA. By parallelizing distance calculations rath…
-
Developer builds fully local RAG system on Kubernetes without API keys
A developer has detailed how to build a fully local Retrieval-Augmented Generation (RAG) system using Kubernetes, bypassing the need for cloud services and API keys. The system utilizes PostgreSQL with the pgvector exte…
-
New algorithm Refnd prevents data leakage in relational datasets
Researchers have introduced Refnd, a new algorithm designed to prevent data leakage in relational datasets, particularly those in biochemical applications. Refnd formalizes the Relational Generative Process (RGP) to exp…
-
RAG Evals Flawed by Non-Deterministic Retrieval, Developer Finds
A developer encountered issues with their Retrieval-Augmented Generation (RAG) evaluation pipeline, initially suspecting the language model was degrading. However, upon closer inspection, the problem was traced to non-d…
-
New DaoQL System Separates LLM Knowledge for Improved Reasoning
Researchers have developed DaoQL, a novel system that separates deterministic knowledge from large language models (LLMs) into an explicit multimodal database. This approach aims to mitigate risks like hallucination and…
-
Ferrovec: Tiny Rust HNSW index enables browser-based semantic search
A new Rust library called Ferrovec has been developed to enable semantic search directly within a browser tab. Unlike existing solutions that rely on server-side infrastructure, Ferrovec compiles to WebAssembly, allowin…
-
TurboVec introduces cost-efficient, private vector retrieval for enterprise RAG
Researchers have developed TurboVec, an open-source vector index designed for cost-efficient and private retrieval in enterprise Retrieval-Augmented Generation (RAG) systems. TurboVec utilizes TurboQuant, a novel codebo…
-
RAG evolves beyond naive document chunking to structured knowledge infrastructure
Retrieval-Augmented Generation (RAG) is evolving beyond its initial simple document-chunking approach. The limitations of 'naive RAG' become apparent with increased complexity, particularly when dealing with over 50,000…
-
OpenSearch Optimizations for Production RAG Systems
This article details optimizations for using OpenSearch in production RAG systems, focusing on improving the semantic retrieval step. It explains how Approximate Nearest Neighbor (ANN) search, specifically using Hierarc…
-
New SHEAF method predicts query hardness in ANN search
Researchers have developed a new method called SHEAF (Self-profiled Hardness Estimation from Answer-set Flux) to predict the hardness of queries in graph-based approximate nearest neighbor (ANN) search. Unlike previous …
-
Understand HNSW: Why Vector Search Returns Garbage
This article delves into the Hierarchical Navigable Small World (HNSW) graph structure, a key component in modern vector databases that enables efficient Approximate Nearest Neighbor (ANN) search. The author highlights …
-
New framework guarantees accuracy for HNSW search algorithms
Researchers have developed a new framework called "Certify-then-Rectify" to improve the accuracy of Hierarchical Navigable Small World (HNSW) graphs, which are widely used in information retrieval. This method first use…
-
New framework guarantees HNSW graph accuracy with minimal overhead
Researchers have developed a new "Certify-then-Rectify" framework to improve the accuracy of Hierarchical Navigable Small World (HNSW) graphs, which are widely used for their speed but lack theoretical correctness guara…
-
New RACORN-1 algorithm boosts filtered vector search performance
Researchers have introduced RACORN-1, an enhancement to the ACORN-1 algorithm designed to improve filtered vector search (FVS) performance. FVS combines vector similarity with metadata filtering, crucial for RAG and ret…