Posted by Laurence Moroney (Google) and Siddarth Sharma (NVIDIA) Today we are announcing integration of NVIDIA® TensorRTTM and TensorFlow. TensorRT is a library that optimizes deep learning models for inference and creates a runtime for deployment on GPUs in production environments. It brings a number of FP16 and INT8 optimizations to TensorFlow and automatically selects platform specific kernels
![Announcing TensorRT integration with TensorFlow 1.7](https://cdn-ak-scissors.b.st-hatena.com/image/square/4c14a9c08429620d65abb863ef8f4e840251dd58/height=288;version=1;width=512/https%3A%2F%2F3.bp.blogspot.com%2F-Pl-1BEKV_TI%2FWrlazTVybEI%2FAAAAAAAAE1w%2FRqEufGTPqIw81tuN2EBd-_6YWsYA5qnbQCLcBGAs%2Fw1200-h630-p-k-no-nu%2Fimage2.png)