SmallDoge/DMA-Pretrain
收藏Hugging Face2025-04-30 更新2025-05-31 收录
下载链接：
https://hf-mirror.com/datasets/SmallDoge/DMA-Pretrain
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含两种配置：decay_8192_8B和stable_2048_32B。每种配置都包括input_ids和attention_mask两个字段，分别为int32类型的序列和int8类型的序列。数据集被划分为训练集，其中decay_8192_8B配置的训练集包含100万示例，大小为28666203830字节；stable_2048_32B配置的训练集包含1600万示例，大小为147526268015字节。
The dataset consists of two configurations: decay_8192_8B and stable_2048_32B. Each configuration includes two fields, input_ids and attention_mask, which are sequences of type int32 and int8 respectively. The dataset is split into training sets, with the decay_8192_8B configurations training set containing 1,000,000 examples and being 28,666,203,830 bytes in size; the stable_2048_32B configurations training set contains 16,000,000 examples and is 147,526,268,015 bytes in size.
提供机构：
SmallDoge
