Eric Curtin

macOS is a packaging effort. Could we consider running RamaLama inside podman-machine or WSL2 for Windows? Porting it to Windows will be a significant effort. Note if we run RamaLama...

> > Note if we run RamaLama directly on Windows and/or macOS you lose all the container features of RamaLama, which is kind of a key goal of RamaLama and...

> > Because containers don't exist in Windows or macOS, but if you run RamaLama inside a Linux VM like podman-machine or WSL2 (WSL2 already should have the GPU passthrough...

It would not be a terrible idea just to ship this with podman desktop/podman machine, it's small, like we do with krunkit

@sayap we would love this feature in Docker Model Runner: https://github.com/docker/model-runner Please star, fork and contribute!

Thanks for testing @Ben-Epstein ! @rhatdan PTAL when you get back

@hosekadam @afazekas @rhatdan @marceloleitner @maxamillion I think this patch could fix it: https://github.com/containers/ramalama/pull/1475 please test and review.

@hosekadam does: ``` ramalama serve qwen-coder:7b ``` work with this patch? If HSA_OVERRIDE_GFX_VERSION=10.3.0 is required, it is possible to add a patch to detect your gpu gfx number and set...

> @ieaves We already apply `black` and `isort` in the make target `format`/`check-format` ([see here](https://github.com/containers/ramalama/blob/main/Makefile#L132-L135)) in the CI during the Lint Code job [here](https://github.com/containers/ramalama/blob/main/.github/workflows/ci.yml#L22-L24). I'd consider the results in the...

Like for example if I do a quick documentation change and I want to commit it quickly and push, I don't want this to occur everytime I do a micro...