SmallDoge/CDMoE_dataset
收藏Hugging Face2025-07-15 更新2025-08-09 收录
下载链接：
https://hf-mirror.com/datasets/SmallDoge/CDMoE_dataset
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含两个配置：decay和stage1。每个配置都包含训练数据，其中input_ids是数据的主要特征，decay配置中input_ids为序列类型，而stage1配置中input_ids为列表类型。数据集规模较大，decay配置的训练数据包含800万个示例，而stage1配置的训练数据包含3200万个示例。
The dataset consists of two configurations: decay and stage1. Each configuration contains training data, with input_ids as the main feature, where input_ids are of sequence type in the decay configuration and list type in the stage1 configuration. The dataset is large in scale, with the decay configurations training data containing 8 million examples, and the stage1 configurations training data containing 32 million examples.
提供机构：
SmallDoge
