#!/bin/bash #should just be plug and play given dockerfile and this start-script ## but we'll see export USER="$(whoami)" export PATH=$PATH:"/home/$USER/.local/bin" # might need to run twice, test on an actual GPU, best I can tell it takes one ## round through to create the torch module so that it can actually run ## Conda might actually be a fix for this if I can make it work, not tonight though ## use ./webui.sh --listen --xformers --use-cpu all --precision full --no-half --skip-torch-cuda-test if no GPU ./webui.sh --listen --xformers ./webui.sh --listen --xformers