Daemontatox/natural_reasoning
收藏Hugging Face2025-03-01 更新2025-04-12 收录
下载链接:
https://hf-mirror.com/datasets/Daemontatox/natural_reasoning
下载链接
链接失效反馈官方服务:
资源简介:
NaturalReasoning是一个大规模的通用推理任务数据集,包含从预训练语料库DCLM和FineMath反译出的大量高质量挑战性推理问题。这些问题已经经过去重和净化,来自流行的推理基准测试如MATH、GPQA、MMLU-Pro和MMLU-STEM。对于每个问题,如果可能,我们从预训练语料库的原始文档中提取参考最终答案。此外,还提供了Llama3.3-70B-Instruct模型生成的响应。
NaturalReasoning is a large-scale dataset for general reasoning tasks, consisting of high-quality challenging reasoning questions backtranslated from pretraining corpora DCLM and FineMath. The questions have been deduplicated and decontaminated from popular reasoning benchmarks including MATH, GPQA, MMLU-Pro, and MMLU-STEM. For each question, we extract the reference final answer from the original document in the pretraining corpora if possible, and also provide a model-generated response from Llama3.3-70B-Instruct.
提供机构:
Daemontatox



