hazyresearch/MMLU_with_Llama_3.1_8B_Instruct_v1
收藏Hugging Face2025-06-24 更新2025-07-05 收录
下载链接：
https://hf-mirror.com/datasets/hazyresearch/MMLU_with_Llama_3.1_8B_Instruct_v1
下载链接
链接失效反馈官方服务：
资源简介：
MMLU_with_Llama-3.1-8B-Instruct数据集包含了基于MMLU基准生成的多选问题，每个问题由Llama-3.1-8B-Instruct模型生成了100个候选答案。这些答案经过混合的GPT-4o-mini和Python代码验证其正确性，并通过多个奖励模型和语言模型进行评分。数据集包含指示、生成的答案、提取的最终答案、答案正确性以及多种模型的评分信息。
The MMLU_with_Llama-3.1-8B-Instruct dataset contains multiple-choice questions generated based on the MMLU benchmark, with each question having 100 candidate responses produced by the Llama-3.1-8B-Instruct model. These responses are verified for correctness using a mixture of GPT-4o-mini and procedural Python code, and scored by multiple reward models and language models. The dataset includes information such as instructions, generated responses, extracted final answers, answer correctness, and scores from various models.
提供机构：
hazyresearch
