selfcorrexp/llama3_prompt_first_wrong_math1_processed
收藏Hugging Face2024-12-19 更新2024-12-21 收录
下载链接：
https://hf-mirror.com/datasets/selfcorrexp/llama3_prompt_first_wrong_math1_processed
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含多个特征，如索引、提示、答案序列、是否第一轮、真实值、奖励序列、解决方案序列、标志、轮次和对话内容。数据集分为训练集，包含50462个样本，总大小为631376854字节。下载大小为233927297字节。
The dataset contains multiple features such as idx (index), prompt, answers (sequence of answers), first_round (whether it is the first round), gt (ground truth), rewards (sequence of rewards), my_solu (sequence of solutions), flag, turn (round), and conversations (dialogue content). The dataset is divided into a training set, containing 50462 samples, with a total size of 631376854 bytes. The download size is 233927297 bytes.
提供机构：
selfcorrexp
