RAG Chunk Calculator
Calculate and Preview RAG Chunks
Set chunk size and overlap, preview sections, and export JSONL for a knowledge base.
PRIVATE WORKSPACE
RAG Chunk Calculator
Plan document splitting and embedding counts before building RAG.
HOW TO
How to use RAG Chunk Calculator
Plan document splitting and embedding counts before building RAG.
- 1
Paste the document text into the workspace.
- 2
Choose the chunk size and overlap in tokens, then process the text.
- 3
Inspect chunk boundaries and download JSONL with each chunk’s ID, token count and text.
INPUT AND OUTPUT
Inspect text chunks before building a retrieval pipeline
Chunk size controls the maximum tokens in each section; overlap repeats tokens at neighboring boundaries. The preview lets you inspect what context a retrieval system would receive.
This is a token-window splitter, not an embedding model or retrieval evaluation. It can split a paragraph or table mid-thought. The token estimate uses the bundled tokenizer and may differ from the tokenizer in your chosen model.
FAQ
Common questions
Are my files uploaded?+
No. This tool processes files in browser memory on your device.
Is the tool free?+
Yes. Core processing is free, requires no account, and adds no watermark.
Does a RAG chunk count tell me how many embeddings I have?+
It tells you how many text chunks were produced. If your later pipeline embeds each chunk once, that count can help plan the workload; this tool does not create embeddings.
What does overlap do to neighboring chunks?+
It repeats part of the preceding token window in the next one. More overlap carries extra context but increases repeated text and the amount your later pipeline may process.