Update correctness checking code for fx2trt (#981)
Summary:
TensorRT generates less accurate results in both FP32 and FP16 precisions. Therefore, we are using more relaxed cosine-similarity to check accuracy.
Pull Request resolved: https://github.com/pytorch/benchmark/pull/981
Reviewed By: frank-wei
Differential Revision: D37329822
Pulled By: xuzhao9
fbshipit-source-id: 2d569ea4b3bd6511ecf6179ab670cc5d0052176f
