It seems that the call from external is OK, but wh...
# ask-for-help
j
It seems that the call from external is OK, but when the internal call to runner, the PyTorchTensorContainer is lost