---
dataset_info:
  features:
  - name: url
    dtype: string
  - name: channel_id
    dtype: string
  - name: post_id
    dtype: int64
  - name: views
    dtype: int64
  - name: pub_time
    dtype: int64
  - name: text
    dtype: string
  - name: fetch_time
    dtype: int64
  - name: images
    sequence: string
  - name: links
    sequence: string
  - name: videos
    sequence: string
  - name: reply_to
    dtype: string
  - name: forward_from
    dtype: string
  - name: channel_title
    dtype: string
  - name: has_obscene
    dtype: bool
  - name: patched_text
    dtype: string
  - name: groups
    struct:
    - name: economy
      dtype: string
    - name: main
      dtype: string
    - name: tech
      dtype: string
  - name: issue
    dtype: string
  - name: language
    dtype: string
  splits:
  - name: train
    num_bytes: 3508000056
    num_examples: 1672028
  download_size: 1827333867
  dataset_size: 3508000056
license: cc-by-4.0
task_categories:
- text-generation
language:
- ru
pretty_name: Nyan Documents
size_categories:
- 1M<n<10M
---

# Nyan documents

Documents scraped for [НЯН](https://t.me/nyannews) Telegram channel from March 2022 to December 2023. The dataset includes documents from 100+ different Telegram news channels.

## Usage
```bash
pip3 install datasets
```

```python
from datasets import load_dataset
for row in load_dataset("NyanNyanovich/nyan_documents", split="train", streaming=True):
    print(row)
    break
```

## Other datasets
* Documents (this dataset): https://huggingface.co/datasets/NyanNyanovich/nyan_documents
* Clusters: https://huggingface.co/datasets/NyanNyanovich/nyan_clusters