mlabonne/smoltalk-semhashed
收藏Hugging Face2025-01-13 更新2025-02-15 收录
下载链接：
https://hf-mirror.com/datasets/mlabonne/smoltalk-semhashed
下载链接
链接失效反馈官方服务：
资源简介：
这是一个基于semhash库创建的smoltalk数据集的近似去重版本，使用minishlab/potion-base-8M生成的embeddings进行去重，保证了相似度阈值为0.95。
This is a near-deduplicated version of the smoltalk dataset created with the semhash library, using embeddings generated by minishlab/potion-base-8M for deduplication, with a similarity threshold of 0.95.
提供机构：
mlabonne
