Problem for adding a caffe custom layer in TensorRT

AI & Data Science Deep Learning (Training & Inference) TensorRT
1

hello,

I am trying to use IPlugin to add a caffe custom layer in TensorRT ,I referred the *.cu code of the caffe layer to write my custom plugin,the problem is I don’t know how to represent some sentences (from *.cu )in IPlugin. For example :

Dtype* pre_gate_data = pre_gate_.mutable_gpu_data(); (in *.cu)

or

this->blobs_[3]->mutable_gpu_data();

the pre_gate_'s definition is Blob pre_gate_; and blobs_ is vector<shared_ptr<Blob > > blobs_;

These confused me in a long time,any suggestion is great!