Selected Work
07
BetterRAG
RAG evaluation tool
A RAG evaluation toolkit that empirically benchmarks chunking strategies, fixed-size, recursive, and semantic, against comprehensive metrics like context precision, token efficiency, answer relevance, and latency. Features zero-code YAML configuration, an interactive Plotly Dash dashboard, and MongoDB-backed persistence for reusing embeddings across evaluations.
Outcome
Empirically benchmarks chunking strategies
Gallery01 images