AI » RAG
Retrieval-Augmented Generation, chat with your documents
Jedan od najboljigh resursa RAG tool interesatno da razumeš je RAG Playground - Interactive RAG Pipeline Visualization
Embedding models
Po njima, najbolje su se pokazali embeddings modeli:
Embeddings in the Gemini API | Google AI for Developers Best-in-class embedding models and rerankers Voyage AI | Home
Reranking models
There are several reranking models on the market. We ran our tests with the Cohere reranker. Voyage also offers a reranker, though we did not have time to test it.
Contextual Retrieval as RAG alternative
Introducing Contextual Retrieval \ Anthropic, a tu je bitan i prompt caching da to sve ne bude preskupo, pošto ceo dokument ubaciš u cache.
Ovde vidi o prompt caching…
anthropic-cookbook/skills/contextual-embeddings at main · anthropics/anthropic-cookbook
RAG Contextual Retrieval je dodatna fora u embedding za RAG i odlično sve objašnjeno ovde: Anthropic’s New RAG Approach. If you want a more detailed article… | by Barhoumi Mosbeh | Towards AI
Objašnjeno ovde The Best RAG Technique Yet? Anthropic’s Contextual Retrieval Explained! - YouTube ali je to ustvari embeddings procesiranje zajedno sa LLM sumarizacijom kontexta.
RAG is a technique where a model improves its answers by retrieving information from document collection or database or similar.
GPT4All LocalDocs is open-source RAG solution. Formulaic: create and deploy AI apps faster. from Mozilla is a new RAG experiment koji odmah možeš da probaš.
Evo ga najbolji:
RAGFlow | RAGFlow sa repo na infiniflow/ragflow: RAGFlow is an open-source RAG (Retrieval-Augmented Generation) engine based on deep document understanding.
CustomGPT.ai | Support & Research AI Agents For Business je prvenstveno RAG tool, ali je potrebna subscription za njega i njegov API key kako bi koristio CustomGPT.ai Researcher.
RAG on source code
Greptile | AI Code Reviews | Full Codebase Context Refact.ai AI Coding Assistant | Fine-tuned, Secure, and Open-source - Refact.ai is open-source at smallcloudai/refact: WebUI for Fine-Tuning and Self-hosting of Open-Source Large Language Models for Coding
Onyx AI | Open Source Enterprise Search & AI Assistant us RAG as open-source onyx-dot-app/onyx: Gen-AI Chat for Teams - Think ChatGPT if it had access to your team’s unique knowledge., formerly Danswer
Preparing documents for AT, scraping sites for RAG
RAG on Website , scraping for RAG
convert a webpage into markdown - Home - Crawl4AI Documentation (v0.4.3bx) with repo at unclecode/crawl4ai: 🚀🤖 Crawl4AI: Open-source LLM Friendly Web Crawler & Scraper A dobar primer je ottomator-agents/crawl4AI-agent/crawl4AI-examples/3-crawl_docs_FAST.py at main · coleam00/ottomator-agents koji scrape-uje ceo sajt i snima u markdown
Crawl4AI is like self-hosted Firecrawl, koji je takođe self-hosted.
DS4SD/docling: Get your documents ready for gen AI
Vector Databases
Qdrant - Vector Database - Qdrant is open-source and self-hosted Vector Database and Vector Search Engine at qdrant/qdrant: Qdrant - High-performance, massive-scale Vector Database and Vector Search Engine for the next generation of AI. Also available in the cloud https://cloud.qdrant.io/
RAG as mcp using Qdrant - qpd-v/mcp-ragdocs