search

yandex-research / ddpm-segmentation

Label-Efficient Semantic Segmentation with Diffusion Models (ICLR'2022)
https://yandex-research.github.io/ddpm-segmentation/
MIT License
662 stars 60 forks source link

Some questions about mpi4py #4

Closed HoJ-Onle closed 2 years ago

HoJ-Onle commented 2 years ago

Hi. Thanks for your work. I had some problems when I was following your work. In fact, I can't install the "mpi4py". How can I run the code without this requirement?

dbaranchuk commented 2 years ago

Hi,

Actually, this requirement comes from the guided_diffusion repo and isn't needed for this code. So, just comment/remove the lines here and probably here

HoJ-Onle commented 2 years ago

Hi,

Actually, this requirement comes from the guided_diffusion repo and isn't needed for this code. So, just comment/remove the lines here and probably here

Thank you very much. I had solved the problem with your help. But I met a new trouble as follows:

Total dimension 8448 scripts/ddpm/train_interpreter.sh: line 6: 169447 Killed python train_interpreter.py --exp experiments/${DATASET}/ddpm.json $MODEL_FLAGS

I found this error in line 57 "X = X.permute(1,0,2,3).reshape(d, -1).permute(1, 0)". ("train_interpreter.py") I think my computer with memory 512G is enough to finish this work. My pytorch version is 1.8.0. I don't know if it matters.