Pipeline
How it works
Orchestrate retrieval, prompt assembly, and LLM generation in a single RagPipeline.run() call — returns RagResult with the answer and CitationRecord provenance.
The main building blocks are:
RagResult— Output ofRagPipeline.run.RagPipeline— LLM-agnostic RAG pipeline.
Under the hood, this feature spans 2 source files covering:
- Lightweight, LLM-agnostic RAG pipeline.
What connects to it
This feature relates to: pipeline, orchestration, rag, result.
Other parts of the codebase interact with pipeline through these interfaces:
| Interface | Purpose | File |
|---|---|---|
RagResult |
Output of RagPipeline.run. |
src/attune_rag/pipeline.py |
RagPipeline |
LLM-agnostic RAG pipeline. | src/attune_rag/pipeline.py |