RAG Pipeline — Reladraw template

RAG 检索增强生成. Copy the source, paste into the official playground, done. No coordinates — every position is stated relative to something else.

Preview

Source (.reladraw)

Copynode docs "Documents"
node chunk "Chunking"  below docs
node embed "Embedding model"  below chunk
node vectordb "Vector DB"  below embed
node query "User query"  right of embed
node retrieve "Retriever"  below query
node llm "LLM"  below retrieve
edge docs -> chunk
edge chunk -> embed
edge embed -> vectordb
edge query -> retrieve
edge vectordb -> retrieve "top-k"
edge retrieve -> llm "context"

How to use

1. Open the Reladraw playground
2. Paste the source above — the diagram re-solves live
3. Export SVG from the playground, or render locally: npx reladraw rag-pipeline.reladraw -o out.svg

When to use this template

Documents chunked and embedded into a vector store; queries retrieve top-k context into the LLM.. Use it to explain retrieval-augmented generation to both engineers and stakeholders — the two-column shape separates indexing from querying.

How to customize

1. Add a step or node. Insert one line: node new_id 'New label' below anchor — placement is relative, so the solver re-flows the whole diagram instantly.
2. Relabel an edge. Edit the text after the arrow: edge a -> b 'your label' — labels are the fastest way to carry rates, protocols or percentages.
3. Change reading order. Swap a placement word: below to right of rotates a branch sideways without touching any other line.

FAQ

Related templates