search

mingyuan-zhang / MotionDiffuse

MotionDiffuse: Text-Driven Human Motion Generation with Diffusion Model
https://mingyuan-zhang.github.io/projects/MotionDiffuse.html
Other
835 stars 74 forks source link

Positional encoding for gesture sequence (self.sequence_embedding) #15

Closed JeremyCJM closed 1 year ago

JeremyCJM commented 1 year ago

Hi Mingyuan,

Why the positional encoding of gesture sequence is a random initialized learnable tensor (self.sequence_embedding)? h = h + self.sequence_embedding.unsqueeze(0)[:, :T, :]

Thanks, Jeremy