reciprocal rank fusion
PulseAugur coverage of reciprocal rank fusion — every cluster mentioning reciprocal rank fusion across labs, papers, and developer communities, ranked by signal.
11 day(s) with sentiment data
-
Local code search tool 'ctx' combines BM25 and dense vectors
The article details the inner workings of `ctx`, a TypeScript command-line tool designed for local code search. It explains how `ctx` combines BM25 lexical search with dense vector embeddings, fused using reciprocal ran…
-
RAG systems focus on retrieval improvements and reliability testing · 8 sources tracked
Researchers are exploring advanced techniques to improve Retrieval-Augmented Generation (RAG) systems, focusing on enhancing the retrieval component which is often the bottleneck. Apple Inc. has developed a method using…
-
Nexariza AI unveils hybrid RAG for zero-hallucination enterprise knowledge engines
Nexariza AI is developing a new approach to Retrieval-Augmented Generation (RAG) for enterprise clients, aiming to eliminate hallucinations and ensure data sovereignty. Their system combines dense vector search with spa…
-
Research paper reveals 'fusion tax' in hybrid AI retrieval methods
A new research paper titled "The Fusion Tax" analyzes the drawbacks of hybrid retrieval methods, which combine sparse and dense retrieval techniques. The study found that while hybrid retrieval often shows aggregate gai…
-
Hybrid Search Enhances Enterprise RAG Systems Over Pure Vector Search
This article explains the limitations of pure vector search in enterprise Retrieval-Augmented Generation (RAG) systems, particularly for exact-match queries like product codes or financial tickers. It proposes hybrid se…
-
Developer highlights vector search limitations for exact identifier retrieval
A developer encountered issues with a vector search system failing to retrieve specific error codes like E4012, instead returning results for similar codes such as E4021 and E4013. This problem arises because embedding …
-
Naive RAG fails in production; 7 architectures offer solutions
Naive Retrieval-Augmented Generation (RAG) pipelines often fail in production despite appearing effective during initial demos. These failures stem from five key retrieval issues: vocabulary mismatch, lost context, ques…
-
Study reveals optimal fusion weights for hybrid retrieval are intervals, not points
A new study on arXiv explores the optimal fusion weights for combining sparse and dense retrievers in hybrid information retrieval systems. The research found that the ideal fusion weight for a given query is typically …
-
LLM reranker bug fixed, boosting text-to-SQL accuracy
A developer discovered a bug in their LLM reranker for the schemagate text-to-SQL tool where the model was being called but not providing answers due to token limits, causing the system to silently fall back to a less e…
-
Hybrid Search for RAG: Python, BM25, and Vector Fusion
This article explores how to build a hybrid search system for Retrieval-Augmented Generation (RAG) using Python. It highlights the limitations of pure vector search, particularly in retrieving specific error codes or pr…
-
Developer exposes docs as grounded MCP server using Cloudflare AI Search
A developer has created an open-source project that exposes documentation as a grounded MCP server, utilizing Cloudflare AI Search and a stateless Worker. This setup allows AI clients to query the documentation, with th…
-
Local RAG app development reveals surprising effectiveness of simple techniques
The author details their experience building LocalCortex, a fully local RAG application, and the lessons learned from evaluating various retrieval and generation techniques. They found that many seemingly promising opti…
-
5 RAG mistakes that break production systems
Building a production-ready retrieval-augmented generation (RAG) system requires careful consideration beyond simple demonstrations. Key mistakes include relying on subjective quality assessments instead of quantitative…
-
New research probes LLM long-context retrieval and RAG strategies · 9 sources tracked
Recent research explores how large language models (LLMs) handle long contexts, with studies investigating the mechanisms behind improved performance. One paper examines Chain-of-Thought (CoT) reasoning, finding that it…
-
ShopEase framework uses LLaMA 3.2 for enterprise customer support
A new research paper introduces ShopEase, a multi-agent framework designed for intelligent enterprise customer support. This system integrates six components, including intent recognition, CRM interaction, memory manage…
-
New framework unifies solar analytics, Q&A, and forecasting
A new research paper introduces Solar Intelligence, a hybrid framework designed to unify solar energy analytics, scientific question answering, and machine learning forecasting. This system integrates data from NASA POW…
-
Perplexity launches Q2D-Web benchmark for retrieval in agentic RAG systems
Perplexity has introduced Q2D-Web, a new benchmark and leaderboard designed to evaluate retrieval performance in agentic RAG systems. The benchmark utilizes a large corpus of 190 million web documents and over 69,000 ag…
-
New agent system automates data cataloging with LLMs · 1 source tracked
Researchers have developed Glyph, a multi-strategy agentic system designed to automate the description and tagging of enterprise data catalogs. This system uses cooperating LLM agents to generate column descriptions by …
-
Uteke 0.17 enhances memory store explainability and trust
Codecora has released version 0.17 of its Uteke memory store, focusing on enhanced explainability and trust. The new version introduces an "explain mode" for recall, providing a detailed scoring breakdown including vect…
-
Developer builds RAG platform to prevent confident hallucinations
A developer has created RAG.NextUpgrad, a platform designed to prevent retrieval-augmented generation (RAG) systems from confidently hallucinating answers. The platform prioritizes running on low-resource, free-tier hos…