Dataset · deepset
German dataset for question-answering systems
deepset/germanquad
Corpus of 13,722 manually annotated questions and answers in German.
What's inside
Created by deepset in German, this dataset includes 13,722 manually annotated questions, including a test split with triple annotation. The questions are open-domain self-contained and often require detailed answers, with non-overlapping train and test splits.
What you can do with it
- Training and evaluating German question-answering models
- Testing text-retrieval systems on German texts
- Tasks
- Question answering, Text retrieval
- Languages
- German
- Downloads this month
- 3.2K
- Likes
- 44
Models trained on it
- deepset/gelectra-large-germanquadQuestion answering 2.1K
- mradermacher/smollm3-3b-german-embed-GGUF 708
- deepset/gelectra-base-germanquadQuestion answering 475
- keyvan-ai/Saturn-Duplex-5B-DEAudio → audio 250
- dehio/german-qg-t5-quad 184
- Beko2210/Bel-KI-v1-GGUFText generation 151
- LLukas22/all-MiniLM-L12-v2-embedding-allSentence similarity 146
- Zeolit/lettuce-eidos-768d-v5Sentence similarity 134
How to load it
from datasets import load_dataset
ds = load_dataset("deepset/germanquad")