Skip to content

Dataset · hausmer

Cleaned Telegram channel posts in Russian and Ukrainian

hausmer/ua-ru-tg-channels-clean

Over 1.3M PII-free posts from 110 news, IT, and official Telegram channels for language model training.

What's inside

This corpus contains 1,312,360 Russian and Ukrainian posts from 110 public Telegram channels covering news, official and military briefings, IT (including DOU), job listings, and blogs (the non-humorous portion of the Truha corpus). The data is scrubbed of personal information (PII-clean), nearly every entry (99.9%) includes a message ID and link, and it can be loaded either as a single split or by individual channel.

What you can do with it

  • Pretraining and SFT fine-tuning of Russian and Ukrainian LLMs on authentic news and IT texts
  • Topic classification of posts, media analytics, and named entity recognition
Tasks
Text generation
Languages
Ukrainian, Russian
Size
1M–10M rows
Downloads this month
42.1K
Likes
0
Access
on request on Hugging Face

How to load it

from datasets import load_dataset

ds = load_dataset("hausmer/ua-ru-tg-channels-clean")