Closed qtli closed 3 weeks ago
Thanks for your high-quality work. Could you please provide the blank video or the code you utilized to generate the ablation model VideoChat2_text? Thanks very much!
Hi! For VideoChat2_text, we simply input a video tensor of 0, like torch.zeros_like(video_emb).
VideoChat2_text
torch.zeros_like(video_emb)
Thanks for your high-quality work. Could you please provide the blank video or the code you utilized to generate the ablation model VideoChat2_text? Thanks very much!