[ARM-based Apple] Proper instruction for Mac M1/M2

[liemle3893]

As you may know that CUDA is got going to support Apple ARM-based chipset. Is there anyway to get around and run this project on Mac M1/M2?

I'm go through a lot of issues in this repo and still cannot get my M1 Pro to work.

My Error

│ /opt/anaconda3/envs/visgpt/lib/python3.8/site-packages/torch/cuda/__init__.py:211  │
│ in _lazy_init                                                                                    │
│                                                                                                  │
│   208 │   │   │   │   "Cannot re-initialize CUDA in forked subprocess. To use CUDA with "        │
│   209 │   │   │   │   "multiprocessing, you must use the 'spawn' start method")                  │
│   210 │   │   if not hasattr(torch._C, '_cuda_getDeviceCount'):                                  │
│ ❱ 211 │   │   │   raise AssertionError("Torch not compiled with CUDA enabled")                   │
│   212 │   │   if _cudart is None:                                                                │
│   213 │   │   │   raise AssertionError(                                                          │
│   214 │   │   │   │   "libcudart functions unavailable. It looks like you have a broken build?   │
╰──────────────────────────────────────────────────────────────────────────────────────────────────╯
AssertionError: Torch not compiled with CUDA enabled

[swyxio]

#37

[nikolasucevic]

You can do it but must modify code on 4-5 files to replace cuda with cpu. In this case is file init.py And after torch import put this import torch

torch.backends.cudnn.enabled = False It works fine on my mac pro 2019 just a bit slower but only for image processing

[taohaowei]

现在自带的requirements.txt，是可以帮助我用cpu运转的，但是太慢了。 我尝试过看#37的问题，但是因为修改requirements，会卡在pip install -r requirements.txt 这一步，尝试很多次都不行。

你能基于最新的requirements.txt，告诉我接下来需要怎么做吗？（我现在找不到需要编辑的文件在哪，