Kwai-Klear/KlearReasoner-MathSub-30K
收藏Hugging Face2026-01-06 更新2025-09-13 收录
下载链接：
https://hf-mirror.com/datasets/Kwai-Klear/KlearReasoner-MathSub-30K
下载链接
链接失效反馈官方服务：
资源简介：
这是一个从Klear-Reasoner Math RL数据集筛选出的30K条记录的子集，通过DeepSeek-R1-0120模型输出的结果进行筛选，确保了数学正确性和格式合规性。数据集适用于强化学习，并提供了高质量的样本，确保了准确的奖励信号。
This is a 30K-entry subset of the Klear-Reasoner Math RL dataset, filtered through the outputs of DeepSeek-R1-0120 model ensuring mathematical correctness and format compliance. The dataset is suitable for reinforcement learning and provides high-quality samples with accurate reward signals.
提供机构：
Kwai-Klear
