Beetle-Data/en-for-ja-28B

这是一个名为Beetle-Data/en-for-ja-28B的数据集，包含预分块的令牌化数据块。每个数据块由513个令牌的打包序列组成（输入ID），并确保无跨文档混合。数据以分块方式增量上传，上传完成后通过标记文件data/_finalized.json进行确认。This is a dataset named Beetle-Data/en-for-ja-28B, consisting of pretokenized chunks. Each chunk is a packed sequence of 513 tokens (input IDs) with no cross-document bleeding. The data is sharded incrementally, and the marker file data/_finalized.json is committed once all parts are uploaded.