How was the STAR dataset preprocessed for this code
Opened this issue · 1 comments
SimMLPhys commented
I've been exploring the STAR dataset as described on the official STAR website and noticed some potential differences in the dataset's structure and usage in this repository. Could you please clarify how the STAR dataset used here differs from the version detailed on the official website? Specifically, I'm interested in any modifications or preprocessing steps that were applied to the dataset for integration into Flipped-VQA.
ikodoh commented
Hi,
I have used the STAR dataset from the official website.
Could you tell me what is different from the official website?
Thank you.