RAG: basic
TemplatesArchitecture
Retrieval-augmented generation in its simplest form: documents are chunked, embedded and indexed offline; each question retrieves the top matches and the model answers from them.
What is inside
- Offline ingestion vs online answering
- Corpus, chunker, embedder, vector index
- Retriever feeds the model
Use this template
Opens in the playground, no account needed. Sign up to save it.
Show the Inkline text
diagram architecture "RAG: basic"
direction right
zone offline "Ingestion (offline)" {
corpus docs "Help-centre articles"
chunker chunk "Chunk 512 tokens"
embedder embed "Embed chunks"
}
vectordb index "Vector index" use=pgvector
zone online "Answering (online)" {
client chat "Chat UI"
retriever retrieve "Retriever"
llm model "Chat model"
}
docs -> chunk -> embed
embed -> index : vectors
chat -> retrieve : question
retrieve -> index : top-k
retrieve -> model : context + question
model ~> chat : streamed answer