MOAI
Loading...

Admin Settings

RAG Configuration

Configure the Retrieval-Augmented Generation (RAG) parameters. These settings affect how documents are processed and how many chunks are retrieved for context.

Loading settings...

About RAG Settings

Chunk Size

Documents are split into chunks for embedding and retrieval. Larger chunks preserve more context within each chunk but may dilute the relevance signal.

Chunk Overlap

Overlap ensures that information at chunk boundaries isn't lost. A typical overlap is 10-20% of the chunk size.

Top K

Controls how many chunks are retrieved to form the context for the LLM. More chunks = more context but higher API costs and potential noise.

Recommendations:
  • Small documents: chunk_size=500, overlap=100, top_k=3
  • Medium documents: chunk_size=1000, overlap=200, top_k=5
  • Large documents: chunk_size=2000, overlap=400, top_k=10