search

Vchitect / Latte

Latte: Latent Diffusion Transformer for Video Generation.
Apache License 2.0
1.45k stars 147 forks source link

Import error #12

Open Taldhi opened 4 months ago

Taldhi commented 4 months ago

ImportError: cannot import name 'CaptionProjection' from 'diffusers.models.embeddings' can anyone please help with this errror?

maxin-cn commented 4 months ago

ImportError: cannot import name 'CaptionProjection' from 'diffusers.models.embeddings' can anyone please help with this errror?

Maybe you can update the version of diffusers to 0.24.0 for text-to-video generation.

Taldhi commented 4 months ago

ImportError: cannot import name 'CaptionProjection' from 'diffusers.models.embeddings' can anyone please help with this errror?

Maybe you can update the version of diffusers to 0.24.0 for text-to-video generation.

thanks that works .

zgdjcls commented 4 months ago

Hi, I've checked https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/embeddings.py, could we use PixArtAlphaTextProjection and PixArtAlphaCombinedTimestepSizeEmbeddings to replace CaptionProjection and CombinedTimestepSizeEmbeddings in your code? Also, the cudatoolkit is not written into enviroment.yml, could you please update the environment file or provide the version of cudatoolkit?

maxin-cn commented 4 months ago

Hi, I've checked https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/embeddings.py, could we use PixArtAlphaTextProjection and PixArtAlphaCombinedTimestepSizeEmbeddings to replace CaptionProjection and CombinedTimestepSizeEmbeddings in your code? Also, the cudatoolkit is not written into enviroment.yml, could you please update the environment file or provide the version of cudatoolkit?

If you want to use PixArtAlphaTextProjection and PixArtAlphaCombinedTimestepSizeEmbeddings to replace CaptionProjection and CombinedTimestepSizeEmbeddings in my codes, please ensure the implementation of the both are consistent.

The version of cudatoolkit for my pytorch (2.0.0) is 11.8.