---
license: mit
task_categories:
- text2text-generation
language:
- kk
size_categories:
- 1M<n<10M
---
## Kazakh Paraphrasing Dataset

This dataset is specifically designed for the paraphrasing task in the Kazakh language. It offers a unique resource for natural language processing applications, focusing on the development and evaluation of paraphrasing models.

### Source and Translation Process

Originally sourced from [humarin/chatgpt-paraphrases](https://huggingface.co/datasets/humarin/chatgpt-paraphrases), this dataset has been translated using Google Translate.

### Dataset Content and Structure

The dataset comprises 5.44 million phrases or sentence pairs, each consisting of an original sentence and its paraphrased counterpart in Kazakh. This structure is particularly beneficial for training algorithms to understand and generate paraphrased content while maintaining the original sentence's meaning.

### Usage and Application

Ideal for researchers and developers in the field of computational linguistics, this dataset serves as a robust tool for training and evaluating paraphrasing models in the Kazakh language. It can significantly contribute to advancements in language technologies for Kazakh.

### Acknowledgments and References

We extend our gratitude to the original dataset providers at [humarin/chatgpt-paraphrases](https://huggingface.co/datasets/humarin/chatgpt-paraphrases) and the team of linguists and translators who contributed to the adaptation of this dataset for the Kazakh language.