VAE training error: RuntimeError: The size of tensor a (24) must match the size of tensor b (17) at non-singleton dimension 0

[akanimax]

Hello!

First, thanks a lot for releasing the source-code and models for this wonderful project, it's been super useful.

I am trying to run the VAE training pipeline, but facing the error mentioned in the title. Apparently this error is caused because of the following hard-coded is_init_image=True.

Pyramid-Flow/video_vae/causal_video_vae_wrapper.py

Line 117 in a012faa

is_init_image=True, temporal_chunk=False,)

When I change it to False, the pipeline works, but I wanted to ask here if this is the correct thing to do. Why exactly has this been hardcoded here?

Look forward to hearing from you.

Thanks and regards,
@akanimax