search

salesforce / awd-lstm-lm

LSTM and QRNN Language Model Toolkit for PyTorch
BSD 3-Clause "New" or "Revised" License
1.96k stars 488 forks source link

THCudaCheck FAIL file=/tmp/luarocks_cutorch-scm-1-3884/cutorch/lib/THC/THCGeneral.c line=70 error=35 : CUDA driver version is insufficient for CUDA runtime version #80

Open AnuragRoy82 opened 6 years ago

AnuragRoy82 commented 6 years ago

I am getting the error on the title _THCudaCheck FAIL file=/tmp/luarockscutorch-scm-1-3884/cutorch/lib/THC/THCGeneral.c line=70 error=35 : CUDA driver version is insufficient for CUDA runtime version while running main.py. Is there any way to run main.py if I am using Pytorch CPU?

egrefen commented 6 years ago

There's a bug in the flag declaration (AFAIK). The lines 48-49 in main.py should be:

parser.add_argument('--cuda', action='store_true', default=False,
                    help='use CUDA')

instead of 

parser.add_argument('--cuda', action='store_false',
                    help='use CUDA')

For now, you can (counter-intuitively) deactivate cuda (e.g. on a machine with no GPU) by adding the flag --cuda to your arguments.