[compiled autograd] compile fwd in inference mode

ghstack-source-id: 8aa1e631d8bdaacfc8d3a6a1f88da32e50cbce71
Pull Request resolved: #125201

torch/_functorch/aot_autograd.py

@@ -544,6 +544,9 @@ def convert(idx, x):
  any(x.requires_grad for x in fake_flat_args if isinstance(x, Tensor))
  and torch.is_grad_enabled()
  )
+ if torch._dynamo.compiled_autograd.compiled_autograd_enabled:
+ print(f"Fwd is called within compiled autograd ctx, assume bwd will be handled by compiled autograd")
+ needs_autograd = False
 
  with enable_python_dispatcher():
  # Patch set_rng_state as set_rng_state with fake tensors is