mlfoundations-dev/seed_code_r1_star_coder_instruct

该数据集包含了对话生成的相关特征，如响应种子(response_seed)、指令种子(instruction_seed)、推理过程(reasoning)、解决方案(deepseek_solution)等。数据集被划分为训练集(train)，包含大约10000个示例，总文件大小为422,367,006字节。此外，数据集中的对话被进一步结构化为发送者(from)和消息内容(value)。

The dataset includes features related to dialogue generation, such as response seed (response_seed), instruction seed (instruction_seed), reasoning process (reasoning), solution (deepseek_solution), etc. The dataset is split into a training set (train) containing approximately 10,000 examples, with a total file size of 422,367,006 bytes. Moreover, the conversations in the dataset are further structured with sender (from) and message content (value).