Optimizing with NVIDIA TensorRT software tools.

We are not seeing the speed we were expecting in our implementation. We would like to talk with someone who has experience in this area