Questions on Dataset Subset Selection and Filtering Criteria for WebVid-10M and Panda-70M

[junghye01]

Hello, thank you for the great work.

I have a couple of questions regarding the dataset preprocessing:

For WebVid-10M, you mentioned filtering out videos with captions that do not contain dynamic content using the LLaMA-3 model. Could you please share the criteria or code used to determine whether a caption contains dynamic content?

For Panda-70M, you stated that 5.3 million videos were downloaded. Could you clarify which subset of videos were selected and how they were chosen?

Thank you in advance for your help!