ByteDance-Seed/mga-fineweb-edu
收藏Hugging Face2025-05-19 更新2025-05-31 收录
下载链接：
https://hf-mirror.com/datasets/ByteDance-Seed/mga-fineweb-edu
下载链接
链接失效反馈官方服务：
资源简介：
Massive Genre-Audience Augment Fineweb-Edu语料库是一个合成的预训练语料库，通过重写原有文档来扩展token数量，并保持多样性。该语料库基于SmolLM Corpus的FineWeb-EDU-Dedup子集构建，将token数量从195B扩展到770B。
The Massive Genre-Audience Augment Fineweb-Edu Corpus is a synthetic pretraining corpus that expands the token number by reformulating the original documents while maintaining diversity. It is built based on the FineWeb-EDU-Dedup subset of the SmolLM Corpus, expanding the token count from 195B to 770B.
提供机构：
ByteDance-Seed
