yufan/SFT_OpenO1_Ultra
收藏Hugging Face2024-12-23 更新2025-02-15 收录
下载链接：
https://hf-mirror.com/datasets/yufan/SFT_OpenO1_Ultra
下载链接
链接失效反馈官方服务：
资源简介：
该数据集是一个开源的SFT（Softmax Fine-tuning）数据集，包含了prompt、thought、response等字段。数据集经过清洗，只保留了英语、德语、西班牙语、中文、意大利语、日语和法语的语言数据，并且对每个prompt保留了最长的推理或解释响应，删除了响应长度超过8K字符的数据。训练集包含约110万个例子，总大小约为10GB。
This dataset is an open-sourced SFT (Softmax Fine-tuning) dataset containing fields such as prompt, thought, and response. The dataset has been cleaned to include only languages such as English, German, Spanish, Chinese, Italian, Japanese, and French, and for each prompt, the response with the longest reasoning or explanation has been retained. Data with response lengths exceeding 8K characters has been removed. The training set contains approximately 1.1 million examples and is about 10GB in total size.
提供机构：
yufan
