Torch7 in TX2 cuda num_thread

i met a problem that “cuda time error ,too many resources requested for launch” when i run a ENet moedl in TX2 with torch7.
could you help me? when i run this model, the useage of GPU reached 99% at a short time.

There is relevant topic about torch7:

Also we have PyTorch releases:

Please check above links and see if it helps your usecase.

I have met the same issue “too many resources requested for launch”, have you solved this issue?