metadata
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
task_categories:
- conversational
- text-generation
size_categories:
- 1K<n<10K
- Based on oasst1.
- Only Russian message trees, only main branches.
- Script: get_oasst_ru.py