---
dataset_info:
  features:
  - name: messages
    sequence:
    - name: role
      dtype: string
    - name: content
      dtype: string
  - name: id
    dtype: string
  splits:
  - name: train
    num_bytes: 2033912
    num_examples: 760
  download_size: 2108166
  dataset_size: 2033912
license: apache-2.0
language:
- ru
---

* Based on [oasst1](https://huggingface.co/datasets/OpenAssistant/oasst1).
* Only Russian message trees, only main branches.
* Script: [get_oasst_ru.py](https://github.com/IlyaGusev/rulm/blob/master/self_instruct/src/data_processing/get_oasst_ru.py)