---
license: cc-by-sa-3.0
language:
- de
---

# German DBMDZ BERT Corpus

This datasets includes all corpora that were used for pretraining the [German DBMDZ BERT Models](https://github.com/dbmdz/berts?tab=readme-ov-file#german-bert).

It consists of Wikipedia dump and corpora from [OPUS](https://opus.nlpl.eu/):

| Filename            | Description        | Creation Date | File Size |
| ------------------- | ------------------ | ------------  | --------- |
| `dewiki.txt`        | Wikipedia Dump     | May 2019      | 5.1GB     |
| `eubookshop.txt`    | OPUS EUbookshop    | November 2018 | 2.2GB     |
| `news.2018.txt`     | OPUS News corpora  | January 2019  | 4.1GB     |
| `opensubtitles.txt` | OPUS OpenSubtitles | November 2018 | 1.3GB     |
| `paracrawl.txt`     | OPUS ParaCrawl     | November 2018 | 3.1GB     |