Dataset Splitter

Split a CSV, TSV or JSON Lines dataset into train / validation / test sets with a reproducible seed, ready for fine-tuning or evaluation.

Your dataset

Paste JSON Lines, a JSON array or CSV / TSV. Splitting is seeded, so the result is reproducible.

Nothing split yet

Paste your dataset above, then press “Split dataset”.

Runs entirely in your browser. Your data is never uploaded.

How it works

This utility runs locally in your browser for privacy and speed.

1 ⚡

Unlimited local tools

This run costs 1 credit(s) and is saved to your history automatically.

Log in to save and view your usage history.

My usage history

No runs recorded yet.

Frequently asked questions

Do I get free credits to try Dataset Splitter?
Yes — new accounts get 3 free credits to spend on any tool, Dataset Splitter included. After that you only pay for what you run.
How many credits does Dataset Splitter cost?
Everything runs in your browser and nothing is uploaded: free accounts pay 1 credit per run, and Pro or Pro+ members run it as often as they like.
Is the content I paste private?
Dataset Splitter uses your input only to generate the result. Stored results are deleted automatically by our retention policy, and you can remove any run from your history.
Do I need to install anything?
No. Dataset Splitter works in any modern browser — nothing to download or install.

Related tools

Tools people usually need next, in the same workflow.

Text Chunker

1 ⚡

Split long text into overlapping chunks for RAG, embeddings or long-context prompts — with a preview of every chunk.

Token Counter

1 ⚡

Estimate the tokens in your prompt, the cost per run and the total for N runs, and see how much of a model's context window you are using.

JSON to JSON Schema

1 ⚡

Turn a sample JSON value into a JSON Schema (draft 2020-12 subset) you can use for structured output or function calling.

JSON Repair

1 ⚡

Fix broken JSON from a model or a copy-paste: code fences, comments, single quotes, curly quotes, trailing commas, unquoted keys, Python literals and truncation.