mlfoundations-dev/a1_code_tiny_codes

该数据集包含了响应种子(response_seed)、指令种子(instruction_seed)、推理(reasoning)、deepseek解决方案(deepseek_solution)和原始行索引(__original_row_idx)等字段。数据集被划分为训练集(train)，共有39987个样本，大小为674666605字节。提供了默认配置，指定了训练集的数据文件。

The dataset includes fields such as response seed (response_seed), instruction seed (instruction_seed), reasoning (reasoning), deepseek solution (deepseek_solution), and original row index (__original_row_idx). The dataset is split into a training set (train) with a total of 39,987 samples and a size of 674,666,605 bytes. A default configuration is provided, specifying the data files for the training set.