ONNX dynamo exporter uses custom domains

I have exported the usrnet (GitHub - cszn/USRNet: Deep Unfolding Network for Image Super-Resolution (CVPR, 2020) (PyTorch)) model to onnx using the dynamo exporter because the legacy one does not support FFT. However, the exported onnx model contains nodes with custom domains like “pkg.onnxscript.torch_lib”. When converting the onnx model to tensorrt those custom domains are not supported and throw errors. Is there a way to export a pytorch model to onnx using only the ai.onnx domain?