metadata
dataset_info:
features:
- name: instruction
dtype: string
- name: option_a
dtype: string
- name: option_b
dtype: string
- name: option_c
dtype: string
- name: option_d
dtype: string
- name: answer
dtype: string
- name: id
dtype: int64
splits:
- name: train
num_bytes: 295127
num_examples: 1588
download_size: 124202
dataset_size: 295127
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
QA dataset for testing LLM's ability to understand the meaning of Danish idioms. The dataset is based on Juunge/danske-talemaader.