Menlo/Maze-Reasoning-GRPO-v0.1

这是一个对话数据集，包含对话提示(Prompt)、对话生成的连贯性响应(Cot_Response)、实际响应(Response)、原始令牌(raw_token)以及对话内容(content)和角色(role)。数据集被拆分为训练集，其中包含181611个示例，总大小为约3.02GB。

This is a dialogue dataset containing dialogue prompts (Prompt), coherent responses generated by the dialogue (Cot_Response), actual responses (Response), raw tokens (raw_token), and dialogue content (content) and roles (role). The dataset is split into a training set, which includes 181,611 examples and is approximately 3.02GB in size.