ZachW/nanbeige4-3b-thinking-2511_bookmia-label0-5pct-raw

该数据集名为Nanbeige/Nanbeige4-3B-Thinking-2511 on bookmia-label0-5pct-raw，是一个包含247条英语文本生成任务的数据集，属于micro-creativity推理套件的模型输出。数据集标签包括llm-outputs、creativity和micro-creativity。数据集包含多个列，如task_id（唯一任务标识符）、input（发送给模型的提示）、raw_output（完整模型输出字符串）、thinking_content（提取的思维链/思考块）、final_answer（去除思考后的最终答案）、finish_reason（完成原因）、prompt_tokens（提示中的令牌数）、completion_tokens（生成的令牌数）、temperature（采样温度）和max_tokens（令牌生成限制）。生成配置包括温度（0.0）、最大令牌数（16384）、种子（42）和后端（vllm）。The dataset is named Nanbeige/Nanbeige4-3B-Thinking-2511 on bookmia-label0-5pct-raw and is a text-generation task dataset containing 247 items in English. It is part of the micro-creativity inference suites model outputs. The dataset tags include llm-outputs, creativity, and micro-creativity. The dataset includes multiple columns such as task_id (unique task identifier), input (the exact prompt sent to the model), raw_output (full model output string), thinking_content (extracted chain-of-thought/thinking block), final_answer (extracted final answer after thinking is removed), finish_reason (completion reason), prompt_tokens (number of tokens in the prompt), completion_tokens (number of tokens generated), temperature (sampling temperature used), and max_tokens (token generation limit). The generation configuration includes temperature (0.0), max_tokens (16384), seed (42), and backend (vllm).