Video-Language Pre-training hours

[msw6468]

Thanks for the nice research!

In the case of the Vision-Language Model using the 3M dataset, I am curious about what setting you used for pertaining(which GPUs, how many GPUs) and how much time it took.

Thank you.