Installing PyTorch 1.4 on Tesla K40m


I know that PyTorch does not support Tesla K40m starting from version 1.3. Is there any workaround for this?. I want to install PyTorch 1.4 in the cluster server which has Tesla K40m using pip command. Currently using PyTorch 1.2 with CUDA9.2. The reason for asking the question is that I need to use the COGDL library which required PyTorch 1.4 minimum.

Any help will be much appreciated.

You could build PyTorch from source as described here.

Thank you will check details!!