OctoMed/Gemini-Hard-Reasoning

--- license: apache-2.0 task_categories: - question-answering - text-generation language: - en tags: - reasoning - chain-of-thought - hard --- # OctoMed/Gemini-Hard-Reasoning Hard reasoning examples with explicit chain-of-thought, converted to OctoMed format for SFT training. ## Source Derived from [Roman1111111/gemini-3.1-pro-hard-high-reasoning](https://huggingface.co/datasets/Roman1111111/gemini-3.1-pro-hard-high-reasoning) by [Roman1111111](https://huggingface.co/Roman1111111). All credit for the original data collection and model distillation goes to the original authors. ## Format Each example contains: - `question`: the prompt text (extracted from `original_input.text`) - `responses`: a single reasoning trace combining `model_thoughts` in `<think>` blocks with `model_response` - `answer`: the model's final response without the thinking block许可证：Apache 2.0 任务类别： - 问答（question-answering） - 文本生成（text-generation） 语言： - 英语 标签： - 推理（reasoning） - 思维链（Chain-of-Thought） - 高难度 # OctoMed/Gemini高难度推理数据集 本数据集包含带有清晰思维链（Chain-of-Thought）的高难度推理样本，已转换为OctoMed格式以用于监督微调（Supervised Fine-Tuning，SFT）训练。 ## 来源 本数据集改编自[Roman1111111](https://huggingface.co/Roman1111111)发布的`Roman1111111/gemini-3.1-pro-hard-high-reasoning`数据集（链接：https://huggingface.co/datasets/Roman1111111/gemini-3.1-pro-hard-high-reasoning）。原始数据采集与模型蒸馏工作的全部成果归于原作者。 ## 格式 每个样本包含以下字段： - `question`：提示文本（从`original_input.text`中提取） - `responses`：单条推理轨迹，将包裹在`<think>`标签内的`model_thoughts`与`model_response`相结合 - `answer`：模型的最终输出结果，不含思考过程模块