runtime error while training in the .sh file

[ghost]

When im running "bash scripts/train_PBAFN_stage1.sh" its giving me the following error

RuntimeError: cuda runtime error (10) : invalid device ordinal at ..\torch\csrc\cuda\Module.cpp:33

device ordinal are:0 NVIDIA GeForce MX13

urgent help needed!!

[kris-yangjs]

@Anaghadarshan Did you solve this problem?

[mafia1990]

if you have just one gpu, open train_PBAFN_stage1.sh file and change --nproc_per_node=8 to --nproc_per_node=1