Nysson3624/Conversations

---
annotations_creators: []
language:
- ru
language_creators:
- machine-generated
license:
- cc-by-4.0
multilinguality:
- monolingual
pretty_name: 'Conversations'
size_categories:
- 1M<n<10M
source_datasets: []
tags:
- conversations
task_categories:
- text-generation
task_ids:
- dialogue-modeling
---

# Russian-Language Dialogues Dataset (RLDD)

## Dataset Overview

This dataset contains approximately 9 million Russian-language dialogues collected from diverse sources, including:

- [Jokes](https://huggingface.co/datasets/inkoziev/jokes_dialogues)
- [Literary works](https://github.com/Koziev/NLP_Datasets?tab=readme-ov-file#%D0%B4%D0%B8%D0%B0%D0%BB%D0%BE%D0%B3%D0%B8-%D0%B8-%D0%BE%D0%B1%D0%BC%D0%B5%D0%BD%D1%8B-%D1%80%D0%B5%D0%BF%D0%BB%D0%B8%D0%BA%D0%B0%D0%BC%D0%B8)
- Online forums
- [A small portion of synthetic dialogues](https://huggingface.co/datasets/inkoziev/arithmetic)

Each dialogue is tagged with a generalized source identifier, enabling targeted filtering. The dataset has undergone basic preprocessing, including deduplication, cleaning, typographical normalization, and minor typo correction.

Sample dialogue:

```
- Вы уже кому-нибудь читали свои стихи?
- Нет, а что?
- Просто у Вас глаз подбит.
```

Each conversation turn is started with em dash symbol.

## Intended Use

This dataset is designed for:

- Statistical research in linguistics

- Supplementary training data for pretraining small language models

- Experiments in conversational AI and NLP tasks

## Usage

```python
from datasets import load_dataset

for sample in load_dataset("inkoziev/Conversations", split="train", trust_remote_code=True, streaming=True):
print(sample['conversation'])
```

## Citing

If you use this dataset in your research or projects, please cite it as follows:

```bibtex
@misc{Conversations,
author = {Ilya Koziev},
title = {Russian-Language Dialogues Dataset},
year = {2025},
publisher = {Hugging Face},
howpublished = {\url{https://huggingface.co/datasets/inkoziev/Conversations}},
}
```

## License

This dataset is licensed under the [CC-BY-NC-4.0](https://creativecommons.org/licenses/by-nc/4.0/) license, which permits non-commercial use only.
For commercial use, please contact the author at [[e-mail]].

By using this dataset, you agree to:
- Provide proper attribution to the author.
- Refrain from using the dataset for commercial purposes without explicit permission.

## Limitations

- **Automated Processing**: The dataset was processed automatically with only selective manual checks. As a result, some dialogues may contain errors such as neighboring lines incorrectly split between participants.

- **Uncorrected Errors**: Grammatical errors, typos, and spelling mistakes present in the original sources have been preserved. These may impact the performance of language models trained on this dataset.

- **Data Quality Variability**: Due to the diverse sources of the dialogues, the quality and consistency of the data may vary significantly.

## Ethical considerations

The dataset includes dialogues sourced from internet forums, which may contain:

- **Profanity and offensive language**

- **Insults or derogatory remarks**

- **Misinformation or fake facts**

Users of this dataset should be aware of these potential issues and take appropriate measures to filter or handle such content responsibly, especially when deploying models trained on this data in real-world applications.