---
license: cc-by-sa-4.0
language:
- hu
---

# hogyankell

Hungarian "how-to" dataset crawled from [hogyankell.hu](https://hogyankell.hu/) wiki.

## Token count

Tokenizer: [`magyar-nlp-szine-java/exotic_modernbert_128k_tokenizer_modified`](https://huggingface.co/magyar-nlp-szine-java/exotic_modernbert_128k_tokenizer_modified)

| Metric | Value |
|---|---|
| Total rows | 2,715 |
| Total tokens | 2,104,244 |
| Avg tokens/row | 775.0 |
| Min tokens | 8 |
| Max tokens | 18,772 |

## Columns

- `pageid` — Wikipedia-style page ID
- `title` — article title
- `url` — source URL
- `categories` — list of categories
- `wikitext` — raw wikitext markup
- `plain_text` — cleaned plain text
- `token_count` — number of tokens (using the tokenizer above)