@shivam I’m facing similar issues and also require gpu for inference. I did set “gpu”:true in the aicrowd.json but I get the following while submitting
Traceback (most recent call last):
File “run.py”, line 150, in
run()
File “run.py”, line 136, in run
assert torch.cuda.is_available()==True,“NO GPU AVAILABLE”
AssertionError: NO GPU AVAILABLE
Why isn’t gpu available?