selfcorrexp/llama3_8b_non_delete_40k_scaling_tmp07
收藏官方服务:
资源简介:
这是一个包含多个字段的数据集,包括索引、真实值、提示信息、难度等级、类型、标准解决方案、用户书写的解决方案、预测结果以及是否获得奖励的标记。数据集被划分为训练集,其中包含了320000个示例,总文件大小为1,106,567,980字节。
This dataset includes multiple fields such as index, ground truth, prompt, level, type, standard solution, user-written solution, prediction result, and a reward marker. The dataset is split into a training set, which contains 320,000 examples with a total file size of 1,106,567,980 bytes.
提供机构:
selfcorrexp


