To mitigate potential abuse of the access token, you might want to consider revoking it and (at least in the colab) use the login-widget
in the python file, you might want to consider passing use_auth_token=True which will use the access token stored under ~/.huggingface/token generated by huggingface-cli login
It appears in both your Colab notebook as well as here:
https://github.com/energy-based-model/Compositional-Visual-Generation-with-Composable-Diffusion-Models-PyTorch/blob/7e996917a7e964f63f02af14962038211e63e777/scripts/image_sample_compose_stable_diffusion.py#L28
To mitigate potential abuse of the access token, you might want to consider revoking it and (at least in the colab) use the login-widget
in the python file, you might want to consider passing use_auth_token=True which will use the access token stored under ~/.huggingface/token generated by huggingface-cli login