IP adapter image embeds dimension

[caiqi]

Describe the bug

https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion.py#L537C1-L538C1

In this line, single_image_embeds = torch.stack([single_image_embeds] * num_images_per_prompt, dim=0) will add a new dimension to single_image_embeds,making the image_embedding has 4 dimensions. Would be better to use torch.cat()?

Reproduction

It won't cause errors for now since the embedding is reshaped in attention processor. But current format looks strange.

Logs

No response

System Info

Latest

Who can help?

@yiyixuxu

[yiyixuxu]

we need the additional dimension as batch size

[sayakpaul]

@caiqi did you get to try out @yiyixuxu's suggestion?

[github-actions]

This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.

Please note that issues that do not follow the contributing guidelines are likely to be ignored.

[sayakpaul]

Closing due to inactivity. Feel free to reopen.