Eugene Brodsky
Eugene Brodsky
> I feel like whether we use xformers should be controlled via the config rather than whether xformers is installed. Is there a reason for doing it this way? This...
I added the suggested runtime change in the latest commit
> I guess if for some reason a person with a new card wanted to use xformers, they'd have to download it manually There's no good reason for someone with...
You need to `docker compose build` first. Or if you just want to pull and run the official image, see https://github.com/invoke-ai/InvokeAI?tab=readme-ov-file#docker-container
Did you see the link I posted? Just run this: `docker run --runtime=nvidia --volume ${HOME}/invokeai:/invokeai --gpus=all --publish 9090:9090 ghcr.io/invoke-ai/invokeai`
You're installing in the root of the C: drive. Try to select a location inside your user dir or another drive where your user has write permissions.
We are pinning `scikit-image=~0.21.1`, which does not support Python 3.12. @JamesClarke7283 - you might have a later version cached, which makes the build work for you, but not for @leder11011...
Our image is based on ubuntu23.04 which ships with python3.11. python3.12 as default was introduced in 24.04, and we are not in a hurry to upgrade until we can support...
I've been trying to build this with Strix Halo support for days, and no luck. With either `vulkan` or `hipblas` (I did add the `gfx1151` arch to the Makefile), on...
How critical is it to remove `xformers` completely vs leaving it as an option? `torch-sdp` is 4.8x slower on older generation Pascal GPUs, like P40 or P100 or the Nvidia...