Hi. I’m trying to covert pytorch model to tensorrt through onnx. I can export from pytorch to onnx but when I try to export to tensorrt and got the error. For now the latest version of tensorrt does not support NonZero op of tensorrt.
So I wanted to try to change operations in pytorch that produce NonZero to smth else and try export.
I found that two operations torch.nonzero and torch.where can produce NonZero op after export. I checked my code and had not found it.
How can I check what operations in pytorch produce the NonZero op in onnx?