openthoughts_no_think

# Dataset Card for Ukrainian OpenThoughts 114K

## Dataset Description

**Dataset Summary**

The translated version of [OpenThoughts 114K](https://huggingface.co/datasets/open-thoughts/OpenThoughts-114k) to Ukrainian using [google/gemma-3-27b-it](https://huggingface.co/google/gemma-3-27b-it). We restructured this dataset for instruction tuning by removing reasoning traces.

<!--[Provide a brief overview of your dataset - what it contains, its purpose, and why it was created. Example: "This dataset contains X examples of Ukrainian text collected from Y sources, designed to support the development of Ukrainian language models."] -->

**Languages**

- Ukrainian (uk)

<!-- **Dataset Structure** -->

<!-- The dataset is organized into the following splits:

| Split | Examples |
|-------|----------|
| Train | [number] |
| Validation | [number] |
| Test | [number] | -->

**Data Fields**

- `system`: original system prompt
- `conversation`: list of messages in a dialog (array of objects)
- `from`: normalized sender role — `user` or `assistant` (system messages are removed)
- `value`: message text
- `original`: original conversations from [OpenThoughts 114K](https://huggingface.co/datasets/open-thoughts/OpenThoughts-114k)

## Dataset Creation

**Source Data**

- Base dataset: [OpenThoughts 114K](https://huggingface.co/datasets/open-thoughts/OpenThoughts-114k) (Apache-2.0).
- Translation: inference with [google/gemma-3-27b-it](https://huggingface.co/google/gemma-3-27b-it).

**Processing / “Fixed” Changes**

* **Removed reasoning traces**: dropped all reasoning traces for `assistant` from `conversations`.

## Considerations for Using the Data

**Intended Uses**

- Instruction/chat LLM training in Ukrainian
- Research on reasoning-heavy tasks (without reasoning traces)

**Social Impact**
This dataset was created to support Ukrainian language AI development and improve language technology accessibility for Ukrainian speakers.
<!-- **Bias and Limitations**
[Discuss any known biases, limitations, or potential issues with the dataset. Be transparent about what the dataset may not be suitable for.] -->

## Citation

TBD
<!--
**BibTeX**
```bibtex
@dataset
{dataset_name,
author = {[Your Name/Organization]},
title = {[Dataset Name]},
year = {2025},
publisher = {HuggingFace},
url = {https://huggingface.co/datasets/[your-org]/[dataset-name]}
}
```
-->

## Contact
<!-- For questions or feedback, please contact [your contact information] or open an issue on the dataset repository. -->
For questions or feedback, please open an issue on the dataset repository.

## License

CC-BY-SA-4.0
---
*This dataset is part of the "Lapa" - Ukrainian LLM initiative to advance natural language processing for the Ukrainian language.*

# 乌克兰语OpenThoughts 114K数据集卡片

## 数据集描述
**数据集概述**
本数据集是[OpenThoughts 114K](https://huggingface.co/datasets/open-thoughts/OpenThoughts-114k)的乌克兰语翻译版本，翻译工作通过[google/gemma-3-27b-it](https://huggingface.co/google/gemma-3-27b-it)模型完成。我们移除了原数据中的推理痕迹，对数据集进行重构以适配指令微调任务。

**语言**
- 乌克兰语（uk）

**数据字段**
- `system`: 原始系统提示词
- `conversation`: 对话消息列表（对象数组）
- `from`: 归一化的发送者角色——`user`（用户）或`assistant`（助手）（已移除系统消息）
- `value`: 消息文本
- `original`: 源自[OpenThoughts 114K](https://huggingface.co/datasets/open-thoughts/OpenThoughts-114k)的原始对话

## 数据集构建
**源数据**
- 基础数据集：[OpenThoughts 114K](https://huggingface.co/datasets/open-thoughts/OpenThoughts-114k)（采用Apache-2.0开源协议）。
- 翻译方式：通过[google/gemma-3-27b-it](https://huggingface.co/google/gemma-3-27b-it)模型执行推理生成翻译结果。

**处理/“固化”修改**
* **移除推理痕迹**：删除了`conversations`字段中助手角色的所有推理痕迹。

## 数据集使用注意事项
**预期用途**
- 乌克兰语大语言模型（Large Language Model, LLM）的指令/对话微调训练
- 面向无推理痕迹场景的重推理任务研究

**社会影响**
本数据集旨在支持乌克兰语人工智能产业发展，提升乌克兰语使用者的语言技术可及性。

## 引用信息
待补充

## 联系方式
如有疑问或反馈，请在数据集仓库提交Issue。

## 许可协议
CC-BY-SA-4.0
---
*本数据集隶属于“Lapa”——乌克兰语大语言模型倡议项目，旨在推动乌克兰语自然语言处理技术进步。*