metadata
dataset_info:
features:
- name: text
dtype: string
- name: quadruples
list:
list: string
splits:
- name: train
num_bytes: 1922651
num_examples: 2348
download_size: 1250303
dataset_size: 1922651
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
license: mit
language:
- en
If you use this dataset in your research, please cite:
@Article{Amalvy2026,
title = {Beyond Known Facts: Generating Unseen Temporal Knowledge to Address Data Contamination in LLM Evaluation},
author = {Amalvy, Arthur and Huang, Hen-Hsen},
year = {2026},
volume = {cs.CL},
url = {https://arxiv.org/abs/2601.13658},
pages = {2601.13658},
journal = {arXiv},
}