All Tools
C
RAGFreeOpen Source
CHONKIE
Fast, lightweight chunking for RAG pipelines
MIT
ABOUT
RAG quality often dies at chunking, and teams either hand-roll splitters or import a full ingestion stack. Chonkie is a small library of token, sentence, semantic, and late chunkers with a consistent API, so you can swap strategies and feed a vector store without a heavy pipeline framework.
INSTALL
pip install chonkieINTEGRATION GUIDE
1. Token-aware chunk PDFs and markdown before embedding for RAG
2. Try semantic or late chunking without rewriting the rest of the pipeline
3. Keep ingestion light in a custom RAG service that already has a vector DB
4. Benchmark chunkers with hyperfine or evals before locking a split strategy
TAGS
pythonragchunkingingestionembeddingsopen-source