I am creating a VGG16 in pytorch and am exporting it to a .onnx model before passing it into TVM. In both TVMC and TVM with AutoTVM, when I load the model in (model_name.onnx), I get the following error:
OpNotImplemented: The following operators are not supported for frontend ONNX: aten_view, torch_nn_modules_container_Sequential_classifier_1, torch_nn_modules_container_Sequential_features_1, torch_nn_modules_pooling_AdaptiveAvgPool2d_avgpool_1
When I go to the TVM frontend directory and view the pytorch.py create_convert_map function, aten_view & AdaptiveAvgPool2d_avgpool_1 operators are present. I am assuming the issue is because the onnx.py _get_convert_map function does not contain any of these operators. If that is the case, how am I supposed to use a pytorch implementation exported to ONNX in TVM?