moondream/long-proof-pile
收藏Hugging Face2025-08-19 更新2025-09-13 收录
下载链接：
https://hf-mirror.com/datasets/moondream/long-proof-pile
下载链接
链接失效反馈官方服务：
资源简介：
Proof Pile文档数据集，包含超过128K个token，适用于测量长文本上下文的交叉熵。数据集包含两个字段：文本内容(text)和元数据(meta)，均为字符串类型。数据集划分为训练集，共有100个样本，总大小为52428748字节。
Proof Pile document dataset containing over 128K tokens, intended for measuring cross-entropy of long text contexts. The dataset includes two fields: text content (text) and metadata (meta), both of which are string types. The dataset is split into a training set with a total of 100 samples and a size of 52428748 bytes.
提供机构：
moondream
