Update README.md
Browse files
README.md
CHANGED
|
@@ -1748,6 +1748,38 @@ configs:
|
|
| 1748 |
- split: train
|
| 1749 |
path: 2025-01/train-*
|
| 1750 |
---
|
| 1751 |
-
# Dataset Card for "bbc_images_alltime"
|
| 1752 |
|
| 1753 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1748 |
- split: train
|
| 1749 |
path: 2025-01/train-*
|
| 1750 |
---
|
|
|
|
| 1751 |
|
| 1752 |
+
|
| 1753 |
+
# RealTimeData Monthly Collection - BBC News Images
|
| 1754 |
+
|
| 1755 |
+
This datasets contains all news articles head images from BBC News that were created every months from 2017 to current.
|
| 1756 |
+
|
| 1757 |
+
To access articles in a specific month, simple run the following:
|
| 1758 |
+
|
| 1759 |
+
```
|
| 1760 |
+
ds = datasets.load_dataset('RealTimeData/bbc_images_alltime', '2020-02')
|
| 1761 |
+
```
|
| 1762 |
+
|
| 1763 |
+
This will give you all BBC news head images that were created in `2020-02`.
|
| 1764 |
+
|
| 1765 |
+
# Want to crawl the data by your own?
|
| 1766 |
+
|
| 1767 |
+
Please head to [LatestEval](https://github.com/liyucheng09/LatestEval/tree/master/data/monthly_updater) for the crawler scripts.
|
| 1768 |
+
|
| 1769 |
+
# Credit
|
| 1770 |
+
|
| 1771 |
+
This is resources is created in this AAAI'24 paper: [LatestEval: Addressing data contamination through dynamic and time-sensitive test construction](https://ojs.aaai.org/index.php/AAAI/article/view/29822).
|
| 1772 |
+
|
| 1773 |
+
If you find this collection helpful, please consider cite this paper:
|
| 1774 |
+
|
| 1775 |
+
```
|
| 1776 |
+
@inproceedings{li2024latesteval,
|
| 1777 |
+
title={Latesteval: Addressing data contamination in language model evaluation through dynamic and time-sensitive test construction},
|
| 1778 |
+
author={Li, Yucheng and Guerin, Frank and Lin, Chenghua},
|
| 1779 |
+
booktitle={Proceedings of the AAAI Conference on Artificial Intelligence},
|
| 1780 |
+
volume={38},
|
| 1781 |
+
number={17},
|
| 1782 |
+
pages={18600--18607},
|
| 1783 |
+
year={2024}
|
| 1784 |
+
}
|
| 1785 |
+
```
|