Sidsidney/dolphin-r1
收藏Hugging Face2025-12-15 更新2025-12-20 收录
下载链接：
https://hf-mirror.com/datasets/Sidsidney/dolphin-r1
下载链接
链接失效反馈官方服务：
资源简介：
Dolphin R1是一个由Eric Hartford和Cognitive Computations共同创建的Apache-2.0许可数据集，旨在训练R1风格的推理模型。数据集包含800k样本，具体由300k来自DeepSeek-R1的推理样本、300k来自Gemini 2.0 flash thinking的推理样本和200k Dolphin chat样本组成。数据集分为三个配置：nonreasoning、reasoning-deepseek和reasoning-flash，每个配置对应不同的训练数据文件。
Dolphin R1 is an Apache-2.0 licensed dataset curated by Eric Hartford and Cognitive Computations, designed to train R1-style reasoning models. The dataset consists of 800k samples, including 300k reasoning samples from DeepSeek-R1, 300k reasoning samples from Gemini 2.0 flash thinking, and 200k samples of Dolphin chat. It is divided into three configurations: nonreasoning, reasoning-deepseek, and reasoning-flash, each corresponding to different training data files.
提供机构：
Sidsidney
