PReLU cant be converted

[cocorecoco]

trt: 6.0.1.5 cuda 10.1 cudnn 7.6.3 When I try to convert a model from caffe with PReLU layer

message PReLUParameter {
  // Parametric ReLU described in K. He et al, Delving Deep into Rectifiers:
  // Surpassing Human-Level Performance on ImageNet Classification, 2015.

  // Initial value of a_i. Default is a_i=0.25 for all i.
  optional FillerParameter filler = 1;
  // Whether or not slope parameters are shared across channels.
  optional bool channel_shared = 2 [default = false];
}

layer {
  bottom: "bn0"
  top: "relu0"
  name: "relu0"
  type: "PReLU"
}

I Got Error Like: ERROR: Parameter check failed at: ../builder/Network.cpp::addConstant::562, condition: allDimsGtEq(dimensions, 1) `

[rmccorm4]

Hi @cocorecoco,

The ONNX parser supports the PRelu op according to here: https://github.com/onnx/onnx-tensorrt/blob/master/operators.md

You could try to convert your model to ONNX and then use the ONNX parser to create a TensorRT network instead.

You can reference the samples for examples of using the ONNX parser.

If it doesn't work with the ONNX parser that comes with your TensorRT installation, you may need to update the parser (which is an open source component) by following the steps in this repo's readme: https://github.com/NVIDIA/TensorRT/blob/master/README.md

This also may be related to #179

[rmccorm4]

As far as the Caffe parser goes, this may be related to https://github.com/NVIDIA/TensorRT/issues/179.

But, ONNX parser still may work.

[lucasjinreal]

@rmccorm4 How to call PReLU in tensorrt network API? it seems this code not work it will through undefined error:

        plugin = plugin::createPReLUPlugin(serialData, serialLength);

error:

undefined reference to `nvinfer1::plugin::createPReLUPlugin(float)'

[JaySimple]

@rmccorm4 How to call PReLU in tensorrt network API? it seems this code not work it will through undefined error:

        plugin = plugin::createPReLUPlugin(serialData, serialLength);

error:

undefined reference to `nvinfer1::plugin::createPReLUPlugin(float)'

Got the same problem. Did u solve it?