---
license: mit
---

A subset of [FineWeb-Edu](https://huggingface.co/datasets/HuggingFaceFW/fineweb-edu) randomly sampled from the whole dataset of around 1B gpt2 tokens. 

This dataset is created for illustration purpose in [retrieval-scaling](https://github.com/RulinShao/retrieval-scaling). Please do not distribute.