Closed zengyuy closed 1 year ago
Hi @zengyuy, could you update the connectomics/utils/system.py
file without the use_cpu
option? I think
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
is sufficient as we want to use GPU by default if it is available. Other changes look good to me. Please update your PR and I will merge the code. Thanks!
add --use_cpu option for model inference