pietrolesci/unimixlm
收藏Hugging Face2025-07-25 更新2025-10-25 收录
下载链接:
https://hf-mirror.com/datasets/pietrolesci/unimixlm
下载链接
链接失效反馈官方服务:
资源简介:
该数据集包含了五种不同的配置,分别为bpe128k、langspec128k、multigram128k、tokmix128k和unigramlm128k。每个配置都有对应的训练集和验证集,其中训练集和验证集分别提供了数据集的字节数和示例数。特征为input_ids,数据类型为uint32序列。具体描述未知。
The dataset consists of five different configurations: bpe128k, langspec128k, multigram128k, tokmix128k, and unigramlm128k. Each configuration has its corresponding training and validation sets with provided byte sizes and example counts. The feature is input_ids, with data type as uint32 sequence. Specific description is not available.
提供机构:
pietrolesci



