DavidMChan / aloha

A new reliable, localizable, and generalizable metric for hallucination detection in image captioning models.
5 stars 0 forks source link

HAT Dataset #3

Closed BrianG13 closed 3 months ago

BrianG13 commented 3 months ago

Hi,

Thank you for sharing your work!

I am wondering if the HAT dataset is available.

Thanks!

DavidMChan commented 3 months ago

The HAT dataset is already available in the github repo here: https://github.com/DavidMChan/aloha/blob/e38d69e0004a044254cef2641985c7ae4e01efd4/data/hat-trainval.json

We are working to add some documentation, and will update the documentation when it becomes available.