---
task_categories:
- question-answering
- summarization
language:
- tk
size_categories:
- 10K<n<100K
configs:
- config_name: default
  data_files:
  - split: train
    path: dolly_15k_turkmen.jsonl
---
# Turkmen Dolly 15k Dataset

## Overview

This dataset is a Turkmen translation of the original Dolly 15k dataset. The Dolly dataset is a publicly available instruction-following dataset created by Databricks, containing 15,000 high-quality human-generated prompt-response pairs. This Turkmen version aims to extend the accessibility of instruction-following datasets to the Turkmen language community.

## Dataset Details

- **Original Dataset**: Dolly 15k
- **Language**: Turkmen
- **Number of Samples**: 15,000
- **Types of Tasks**: Various, including open-ended generation, classification, extraction, and more
- **Translation Method**: Google Translate

## File Format

The dataset is provided in JSONL (JSON Lines) format. Each line in the file represents a single JSON object with the following structure:

```json
{
  "instruction": "Original instruction in English",
  "context": "Original context in English (if applicable)",
  "response": "Original response in English",
  "category": "Category of the task",
  "instruction_tk": "Instruction translated to Turkmen",
  "context_tk": "Context translated to Turkmen (if applicable)",
  "response_tk": "Response translated to Turkmen"
}
```

Example:
```json
{
  "instruction": "In the series A Song of Ice and Fire, who is the founder of House Casterly?",
  "context": "",
  "response": "Corlos, son of Caster",
  "category": "open_qa",
  "instruction_tk": "\"Buz we ot aýdymy\" seriýasynda \"House Casterly\" -ny esaslandyryjy kim?",
  "context_tk": "",
  "response_tk": "Karlos, Kasteriň ogly"
}
```

## Acknowledgments

- Original Dolly 15k dataset creators: Databricks
- Translation: Google Translate

## Contact

For questions or issues regarding this dataset, please contact:

- Telegram: @gargamelix
- Email: 31mb41@gmail.com
- GitHub: github.com/mamed0v

## Disclaimer

The translations in this dataset were performed using Google Translate. While this approach allows for rapid translation of a large dataset, users should be aware that there might be inaccuracies, mistranslations, or loss of nuance, especially for complex or domain-specific content. Exercise caution when using this dataset for tasks requiring high precision in language understanding or generation.