Given the author is not fixing the HuggingFace demo, follow these steps:

1. Visit the demo on HuggingFace
2. Cry that there is an error
3. Click the 'Runtime error' button as circled in the image

4. Click the text that says 'try to fix the Space yourself by duplicating it'
5. Go to Files -> requirements.txt
6. Edit the file to the following:

git+https://github.com/zhuang2002/diffusers_cobra.git
torchaudio==2.6.0
torch==2.6.0
torchvision==0.21.0
accelerate==1.5.2
datasets==3.2.0
einops==0.8.1
gradio==5.25.0
gradio_client==1.8.0
huggingface-hub==0.30.2
matplotlib==3.10.1
numpy==1.26.4
opencv-python==4.11.0.86
peft==0.15.0
Pillow==11.1.0
safetensors==0.5.3
sentencepiece==0.2.0
tqdm==4.67.1
transformers==4.48.3

The changes I had to make:
torchaudio from 2.5.1 to 2.6.0
torch from 2.5.1 to 2.6.0
torchvision from 0.20.1 to 0.21.0
gradio from 5.22.0 to 5.25.0

It should now rebuild. Also, make sure in Settings to set publicity to Public otherwise the HTML and CSS and JS files won't load.