kurakurai/luth-sft
收藏Hugging Face2025-10-12 更新2025-09-13 收录
下载链接:
https://hf-mirror.com/datasets/kurakurai/luth-sft
下载链接
链接失效反馈官方服务:
资源简介:
Luth-SFT法语数据集,用于指令遵循、数学和常识任务的微调,包含约3.38亿个法语标记,由Scholar、Tulu 3 Persona Instruct、Tulu 3 Persona Math、Smoltalk2、Aya Dataset、OpenHermes-fr 和 CroissantLLM等多个数据源组成。
Luth-SFT French dataset for fine-tuning on instruction following, math, and general knowledge tasks, containing approximately 338 million French tokens, compiled from multiple sources including Scholar, Tulu 3 Persona Instruct, Tulu 3 Persona Math, Smoltalk2, Aya Dataset, OpenHermes-fr, and CroissantLLM.
提供机构:
kurakurai



