search

energy-based-model / Compositional-Visual-Generation-with-Composable-Diffusion-Models-PyTorch

[ECCV 2022] Compositional Generation using Diffusion Models
https://energy-based-model.github.io/Compositional-Visual-Generation-with-Composable-Diffusion-Models/
Other
456 stars 41 forks source link

composable-diffusion package not used on the Google Colab demo? #3

Closed apolinario closed 2 years ago

apolinario commented 2 years ago

Hi, checking out the Colab demo https://colab.research.google.com/github/energy-based-model/Compositional-Visual-Generation-with-Composable-Diffusion-Models-PyTorch/blob/main/notebooks/compose_glide.ipynb

I couldn't find where the composable-diffusion library is used. If I comment out the first line that installs it 

#!pip install git+https://github.com/energy-based-model/Compositional-Visual-Generation-with-Composable-Diffusion-Models-PyTorch

everything still works just fine

Is that the intended process? I thought it is likely the case that you replicate the same techniques of the paper/library in the masks variable and model_fn, sample_64 functions and that the actual composable-diffusion library is not needed - but then I wondered why is it installed; would be great to get this clarified!

nanlliu commented 2 years ago

that can be removed for now since we haven't released some other code that we use for other models that we trained.