gemini-3-pro-preview-high-reasoning-1000x

This is a reasoning dataset created using Gemini 3 Pro Preview with a reasoning depth set to high. Some of these questions are from reedmayhew and the rest were generated. The dataset is meant for creating distilled versions of Gemini 3 Pro Preview by fine-tuning already existing open-source LLMs on the [**summarized reasoning traces**](https://ai.google.dev/gemini-api/docs/thinking) provided from their API. This dataset includes 250x from [TeichAI/gemini-3-pro-preview-high-reasoning-250x](https://huggingface.co/datasets/TeichAI/gemini-3-pro-preview-high-reasoning-250x) ## Stats - Costs: $ 32.7 (USD) - Tokens: 2.73 M (total)本数据集为基于Gemini 3 Pro预览版（Gemini 3 Pro Preview）构建的推理数据集，推理深度设置为高等级。数据集内的部分问题源自reedmayhew，其余问题由Gemini 3 Pro预览版生成。 本数据集旨在通过在其API提供的**总结式推理轨迹（summarized reasoning traces）**上对现有开源大语言模型（Large Language Model, LLM）进行微调，以构建Gemini 3 Pro预览版的蒸馏版本。 本数据集包含来自[TeichAI/gemini-3-pro-preview-high-reasoning-250x](https://huggingface.co/datasets/TeichAI/gemini-3-pro-preview-high-reasoning-250x)的250份数据。 ## 统计信息 - 成本：32.7美元（USD） - 总Token数：2.73 M（总计）

搜集汇总

背景与挑战

背景概述

该数据集基于Gemini 3 Pro Preview创建，专注于高推理深度问题，旨在通过开源大语言模型对API生成的推理轨迹进行微调，以蒸馏模型版本。它包含250倍扩展的样本，涉及总成本32.7美元和2.73百万令牌。