metadata
license: mit
dataset_info:
features:
- name: question
dtype: string
- name: answer
dtype: string
splits:
- name: train
num_bytes: 1144206.5903728174
num_examples: 1695
- name: test
num_bytes: 286220.40962718264
num_examples: 424
download_size: 936129
dataset_size: 1430427
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*
Hestenet Question-Answer
The dataset is based on data from Hestenettet in the Danish Gigaword corpus. Question-answer pairs are purely extracted on the basis of heuristics, and have not been manually evaluated.
The dataset was created for aiding the training of sentence transformer models in the Danish Foundation Models project. The dataset is currently not production-ready.