Drew Thomasson

.... This sounds... complex... Perhaps we should just lock it as the build arguments to only [cuda121, cuda118, cuda128, rocm, xpu, cpu] with pre-set pip installs and take away the...

How does this look? https://github.com/DrewThomasson/ebook2audiobook/pull/1105 I simplified it and made it be specific if its a cuda 11.8 torch

Should I be locking the torch/torchaudio to version to 2.8 for ALL CUDA Versions?

What was spoken about here has been applied to this Dockerfile update https://github.com/DrewThomasson/ebook2audiobook/pull/1106

Building from the main branch is still broken till the next update (It's a version behind but 🤷) But we should have a pre-built image that might work with cuda...

### Try replacing your `docker-compose.yml` and `requirements.txt` with these provided: -> [docker-compose.yml](https://github.com/user-attachments/files/23293381/docker-compose.yml) [requirements.txt](https://github.com/user-attachments/files/23293380/requirements.txt) ### Replace the contents of your `Dockerfile` with the contents of this `Dockerfile.txt` -> [Dockerfile.txt](https://github.com/user-attachments/files/23293386/Dockerfile.txt) ### Use...

Oh my bad I forgot Set GPU Support (disabled by default) To enable GPU support, modify 'docker-compose.yml' and change *gpu-disabled to *gpu-enabled

> wouldn't be more simpler when TORCH_VERSION is cudaXXX or ROCM so GPU is enabled? Idk if I can make a compose work programmatically like if torch version is on...

@stratus-ss OMG I made a mistake 🤦 you need `cuda128 ` cause of your blackweld GPU I gave you a compose that does `cuda118` lol Here Try this one I...