metadata
dataset_info:
features:
- name: text
dtype: string
- name: answer
dtype: string
splits:
- name: train
num_bytes: 67126095
num_examples: 1257
- name: test
num_bytes: 11649502
num_examples: 200
download_size: 33315639
dataset_size: 78775597
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*
The test dataset is from LongBench's QMSum task. The train dataset is from the original QMSum's repository. There is no built-in validation set. For validation, please take a portion of the training dataset.