How krishnaik06/rag-tutorials works
This repository is a small tutorial implementation of document-grounded question answering. A user submits a query through the example app; search retrieves relevant chunks from a persisted FAISS index and asks a Groq-hosted language model to summarize them. The index-building path loads supported documents, chunks and embeds them, then persists vectors and metadata. The README names no additional product capabilities, and the notebooks are not described here because their contents were not sampled.
Main entry points: app.py Example app · vectorstore.py FAISS vector store · search.py RAG search and summary · data_loader.py Document loader
Components 9
Document ingestion
- Document loader
src/data_loader.py - Chunking and embeddings
src/embedding.py
Index and retrieval
- FAISS vector store
src/vectorstore.py - Persisted FAISS index
Answer generation
- Example app
app.py - RAG search and summary
src/search.py
Other components
- User
- Source documents
- Groq language model
Connections 9
- User → Example app: submits query
app.py - Example app → Document loader: loads documents
app.py - Source documents → Document loader: provides files
- Example app → FAISS vector store: loads index
app.py - Example app → RAG search and summary: runs search
app.py - RAG search and summary → FAISS vector store: queries vectors
src/search.py - FAISS vector store → Chunking and embeddings: chunks and embeds
src/vectorstore.py - FAISS vector store → Persisted FAISS index: reads and writes
src/vectorstore.py - RAG search and summary → Groq language model: invokes model
src/search.py