hyeogi/reason-data

---
dataset_info:
features:
- name: id
dtype: string
- name: conversations
list:
- name: content
dtype: string
- name: from
dtype: string
splits:
- name: train
num_bytes: 30583910
num_examples: 2959
download_size: 30538658
dataset_size: 30583910
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
---

Translating [Jackrong/Qwen3.5-27B-Claude-4.6-Opus-Reasoning-Distilled](https://huggingface.co/Jackrong/Qwen3.5-27B-Claude-4.6-Opus-Reasoning-Distilled)'s training dataset via TranslateGemma.

The dataset consists of high-quality, filtered reasoning distillation data:

| Dataset Name | Description / Purpose |
|--------------|-----------------------|
| [nohurry/Opus-4.6-Reasoning-3000x-filtered](https://huggingface.co/datasets/nohurry/Opus-4.6-Reasoning-3000x-filtered) | Provides comprehensive Claude 4.6 Opus reasoning trajectories. |
| [Jackrong/Qwen3.5-reasoning-700x](https://huggingface.co/datasets/Jackrong/Qwen3.5-reasoning-700x) | Additional curated reasoning samples designed to strengthen structured step-by-step problem solving and improve reasoning diversity. |

数据集信息：
特征字段：
- 名称：id，数据类型：字符串（string）
- 名称：conversations，类型为列表，列表项结构如下：
- 子字段：content，数据类型：字符串
- 子字段：from，数据类型：字符串
数据划分：
- 划分名称：训练集（train），数据字节数：30583910，样本数量：2959
下载大小：30538658字节
数据集总字节数：30583910字节
配置项：
- 配置名称：默认（default）
数据文件：
- 对应划分：训练集，文件路径：data/train-*

本内容通过TranslateGemma工具对[Jackrong/Qwen3.5-27B-Claude-4.6-Opus-Reasoning-Distilled](https://huggingface.co/Jackrong/Qwen3.5-27B-Claude-4.6-Opus-Reasoning-Distilled)的训练数据集进行翻译。

该数据集包含经过筛选的高质量推理蒸馏数据：

| 数据集名称 | 描述与用途 |
|-----------|------------|
| [nohurry/Opus-4.6-Reasoning-3000x-filtered](https://huggingface.co/datasets/nohurry/Opus-4.6-Reasoning-3000x-filtered) | 提供完整的Claude 4.6 Opus推理轨迹数据 |
| [Jackrong/Qwen3.5-reasoning-700x](https://huggingface.co/datasets/Jackrong/Qwen3.5-reasoning-700x) | 额外的精选推理样本集，旨在强化结构化分步问题求解能力并提升推理多样性 |