jadolfbr
commented
10 months ago I’ve tried a few things to this end (though ultimately I was not successful). See the discussion page for more.
Open njrollins opened 10 months ago
jadolfbr
commented
10 months ago I’ve tried a few things to this end (though ultimately I was not successful). See the discussion page for more.
mehedihasandesu
commented
1 month ago Hi, do you have any update? I could install RFdiffusion on M1 and M2 by following the ways, suggested in this fork (https://github.com/YaoYinYing/RFdiffusion#conda-install-se3-transformer). But while executing the example commands, I found the following error message: FileNotFoundError: Cannot find DGL C++ graphbolt library at /Users/mehedi/miniconda3/envs/SE3nv/lib/python3.9/site-packages/dgl/graphbolt/libgraphbolt_pytorch_2.3.1.dylib.
Aminoacid1226
commented
2 weeks ago 你好，有什么更新吗？我可以按照此 fork 中建议的方法（https://github.com/YaoYinYing/RFdiffusion#conda-install-se3-transformer）在 M1 和 M2 上安装 RFdiffusion。但在执行示例命令时，我发现以下错误消息：FileNotFoundError：无法在 /Users/mehedi/miniconda3/envs/SE3nv/lib/python3.9/site-packages/dgl/graphbolt/libgraphbolt_pytorch_2.3.1.dylib 找到 DGL C++ graphbolt 库。
Try reducing the torch grade, I according to the steps in the https://github.com/YaoYinYing/RFdiffusion/tree/main was successful
Would be wonderful to update the code to be compatible with new Macs and any other devices incompatible with cuda. Currently does not install due to cudatoolkit and dgl, which are unavailable for M1/M2.
Possibly the code may only need to change in a small number of places, because recent versions of pytorch are compatible with M1/M2. If I understand, might only need to change the dependencies yml and replace the use of cuda/dgl modules e.g. 'torch.cuda.nvtx'.