Why tensorrt releases this api interface but doesn’t tell us how to use this api in detail with custom layers? I am very confused about the current api document. Everything is not clear about the custom layer. I don’t want to use any parser to deal with the custom layer at all. I just want to build the network by myself. How can I do that? The official documents don’t tell anything about it, and all the examples have nothing to do with the addplugin api. It is very embarrassing.
Hi, did you find more guides on how to use the addPlugin() api? I am currently trying to build the network definition in C++ from scratch and load the weights later for inference. My network consists of a custom layer and I am not sure how to add it properly to the network.
Up till now I’ve mostly been trying to modify the code I found in the TensorRT/samples directory, e.g. sample_plugin.
I also found some repo on Github that might be helpful:
Please let me know if you come across any more clear documentation.
Thank you very much!