Remove fake_mode arg from torch._dynamo.export API (#106345)
#105477 removes the need of explicitly specifying `fake_mode`.
The same effect can be achieved by wrapping `torch._dynamo.export` around a `torch._subclasses.FakeTensorMode` context.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/106345
Approved by: https://github.com/ezyang
