add --cuda_only; fix some nvfuser flags (#960)
Summary:
- Added an option `--cuda_only` to only run cuda benchmarks
- For the nvfuser flags, we've already updated those settings in TIMM. See the latest at https://github.com/rwightman/pytorch-image-models/blob/2f2b22d8c7889174dbf11b92c2d72d8587f9164b/timm/utils/jit.py
Pull Request resolved: https://github.com/pytorch/benchmark/pull/960
Reviewed By: george-qi
Differential Revision: D37187084
Pulled By: xuzhao9
fbshipit-source-id: 75cdaaced599b1816f005f3345b803f3d2400e7c
