How can I run it on a Mac M2 local environment using the CPU?

[SukhleenSingh7]

Can anyone help me understand how to run this on a Mac M2 local environment using only the CPU, without a GPU?

[0xprasanth]

I tried running the app on my MacBook Pro (M2) using a virtual environment, without any additional flags:

python3 app.py

However, it appears that a CUDA-compatible GPU is required to run the model successfully. Also, when I ran it locally I noticed it consumed excessive memory and froze, forcing me to quit Python manually.

After that, I switched to Google Colab and connected to T4 GPU runtime. I asked to ChatGPT for running steps, but the process got stuck for more than 30mintues without any progress.

After stopping the execution, I noticed the following output:

---

Note:
I modified the code in app.py at line 381, changing the device from cuda:3 to cuda:0 to match the Colab GPU. Have a look in Google Colab
For testing purposes, I also used ngrok to expose the Gradio interface in Colab for remote access.