hazyresearch/MMLU_with_Llama_3.1_70B_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_70B_Instruct_v1
下载链接
链接失效反馈官方服务：
资源简介：
MMLU与Llama-3.1-70B-Instruct数据集包含基于MMLU基准的719个多项选择题，每个问题由Llama-3.1-70B-Instruct模型生成了100个候选答案。每个答案的正确性通过GPT-4o-mini和Python代码的组合进行评估，并由多个奖励模型和语言模型评估器给出分数。
The MMLU with Llama-3.1-70B-Instruct dataset contains 719 multiple-choice questions based on the MMLU benchmark, each with 100 candidate answers generated by the Llama-3.1-70B-Instruct model. The correctness of each answer is evaluated by a combination of GPT-4o-mini and Python code, and scores are provided by multiple reward models and language model evaluators.
提供机构：
hazyresearch
