rus_news_classifier / README.md
data-silence's picture
Update README.md
b5a11a2 verified
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

This is a dataset for training models on the task of multiclass classification of news texts in Russian. It consists of a set of news items for the last 5 years and is well balanced.

categories_translator = 
{'climate': 0,
 'conflicts': 1,
 'culture': 2,
 'economy': 3,
 'gloss': 4,
 'health': 5,
 'politics': 6,
 'science': 7,
 'society': 8,
 'sports': 9,
 'travel': 10}