I have an onnx model built with ir version 0.0.3 and opset 7. It has an un-supported layer “PReLU”. I implemented the layer by extending the class IPluginCreator and IPluginV2. Afterwards, I registered the plugin using REGISTER_TENSORRT_PLUGIN(myPluginCreator). However, I still get the error, “No importer registered for op: PRelu”. My question is, is there any other step to register the custom plugin that I am missing? My understanding is, if the issue is in implemented plugin, it should parse the model, and throw an error while building phase. Any pointers are welcome!
Were you able to figure out your issue here?
https://docs.nvidia.com/deeplearning/sdk/tensorrt-developer-guide/index.html#extending should be able to help.
NVIDIA Enterprise Support