Glaciohound/VCML

Args object has no attribute 'cudas'

Closed this issue · 2 comments

Hi,

I am trying to replicate your project, but when running your sample single-experiment command, I am getting the cudas attribute not found error. Could you please give me some help? Thank you!

Traceback (most recent call last):
File "scripts/main.py", line 38, in
main()
File "scripts/main.py", line 28, in main
run_experiment.run(args)
File "/home/VCML/scripts/run_experiment.py", line 164, in run
device = args.cudas[i * args.num_gpus // args.num_parallel]
AttributeError: 'Args' object has no attribute 'cudas'

/home/VCML/scripts/run_experiment.py(164)run()
162
163 for i in range(0, args.num_parallel):
--> 164 device = args.cudas[i * args.num_gpus // args.num_parallel]
165 p = ctx.Process(target=ready_go,
166 args=(args, i, message[i], control[i],

ipdb>

I think I got the reason, it is because of torch.cuda.device_count() gives 0 while it needs to be greater than 0

You should use a server with GPU installed. Closing.