pguso/rag-from-scratch

Recursos principais

• 01Step-by-step examples from document loading to LLM answer generation
• 02In-memory vector store with nearest-neighbor search and metadata filtering
• 03Hybrid search combining vector similarity with BM25 keyword scoring
• 04Multi-query retrieval with reciprocal rank fusion and deduplication
• 05Query rewriting with heuristic and local LLM options via node-llama-cpp
• 06Reusable library modules for loaders, splitters, embeddings, and vector stores

Casos de uso

• Learning how RAG works internally before adopting a framework like LangChain
• Building a local question-answering prototype over your own documents without cloud APIs
• Experimenting with retrieval strategies such as hybrid search or query rewriting
• Teaching embeddings and vector search to a team of JavaScript developers

Tópicos

Perguntas frequentes

O que é rag-from-scratch?

An educational JavaScript tutorial that builds retrieval-augmented generation step by step with local LLMs and no cloud APIs.

rag-from-scratch é gratuito? Qual é a licença?

rag-from-scratch é open source sob a licença MIT. Confira os termos da licença antes de usá-lo em produtos comerciais.

rag-from-scratch ainda é mantido?

Sem commits há mais de 6 meses. O último commit de rag-from-scratch foi em 11 de mar. de 2026.

Para que posso usar rag-from-scratch?

Learning how RAG works internally before adopting a framework like LangChain Building a local question-answering prototype over your own documents without cloud APIs Experimenting with retrieval strategies such as hybrid search or query rewriting Teaching embeddings and vector search to a team of JavaScript developers

Onde consigo rag-from-scratch?

O código-fonte e as instruções de instalação estão no GitHub: https://github.com/pguso/rag-from-scratch