fix amp in inference in benchmarking suite (#103220)
Even if you passed in --amp we would run inference in float32.
`AlbertForMaskedLM` goes from 1.305 float32 to 1.724x amp, and then again to 1.910x with freezing. Benchmark numbers for amp are about to go way up lol.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/103220
Approved by: https://github.com/desertfire