Datasets:
metadata
language:
- ru
size_categories:
- 10K<n<100K
task_categories:
- text-classification
pretty_name: Classified well-balanced russian-language news for 2019-2024
dataset_info:
features:
- name: news
dtype: string
- name: labels
dtype: int32
splits:
- name: train
num_bytes: 132863829.81185599
num_examples: 57530
- name: test
num_bytes: 33217112.188144006
num_examples: 14383
download_size: 86321663
dataset_size: 166080942
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*
tags:
- russian
- news
- media