fix missing sentence

Author: soumith

_posts/2018-01-19-a-year-in.md

@@ -153,7 +153,7 @@ As we embarked upon a redesign of the PyTorch internals, we built the [ATen C++1
 One of the common requests we've received was to export PyTorch models to another framework. Users engaged in a rapid research cycle in PyTorch and when they were done, they wanted to ship it to larger projects with C++ only requirements.
 
 With this in mind, we built a tracer for PyTorch — which can export PyTorch models into an intermediate representation.
-The subsequent trace can be either used to run the current PyTorch model more efficiently (by running optimization passes on it), or be converted to the [ONNX](http://onnx.ai/) format to be shipped to other frameworks such as Caffe2, MXNet, TensorFlow and others or directly to the hardware accelerated . Over the next year, you will hear more about the JIT compiler for performance improvements.
+The subsequent trace can be either used to run the current PyTorch model more efficiently (by running optimization passes on it), or be converted to the [ONNX](http://onnx.ai/) format to be shipped to other frameworks such as Caffe2, MXNet, TensorFlow and others or directly to the hardware accelerated libraries like CoreML or TensorRT. Over the next year, you will hear more about the JIT compiler for performance improvements.
 
 
 # Users being funny :)