Minifier fails on runtime compilation minification

[soumith]

I tried to use the minifier when I found an issue with examples/imagenet in #1687

But it failed with

$ TORCHDYNAMO_REPRO_AFTER="aot" python main.py /home/soumith/dataset/imagenet
/home/soumith/code/vision/torchvision/io/image.py:13: UserWarning: Failed to load image Python extension:
  warn(f"Failed to load image Python extension: {e}")
=> creating model 'resnet18'
WARNING:torch._inductor.lowering:make_fallback(aten.unfold): a decomposition exists, we should switch to it
WARNING:torch._inductor.lowering:make_fallback(aten.unfold_backward): a decomposition exists, we should switch to it
Epoch: [0][   1/5005]   Time 11.820 (11.820)    Data  1.918 ( 1.918)    Loss 6.9964e+00 (6.9964e+00)    Acc@1   0.00 (  0.00)   Acc@5   0.00 (  0.00)
WARNING:torch._dynamo.debug_utils:Writing minified repro to /tmp/minifier_soumith/minifier_launcher.py
WARNING:torch._dynamo.debug_utils:Copying minified repro from /tmp/minifier_soumith/minifier_launcher.py to /home/soumith/code/pytorch/minifier_launch
er.py for convenience
Traceback (most recent call last):
  File "/home/soumith/code/pytorch/torch/fx/interpreter.py", line 199, in placeholder
    return next(self.args_iter)
StopIteration

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/home/soumith/code/examples/imagenet/main.py", line 514, in <module>
    main()
  File "/home/soumith/code/examples/imagenet/main.py", line 121, in main
    main_worker(args.gpu, ngpus_per_node, args)
  File "/home/soumith/code/examples/imagenet/main.py", line 281, in main_worker
    train(train_loader, model, criterion, optimizer, epoch, device, args)
  File "/home/soumith/code/examples/imagenet/main.py", line 339, in train
    loss.backward()
  File "/home/soumith/code/pytorch/torch/_tensor.py", line 488, in backward
   torch.autograd.backward(
  File "/home/soumith/code/pytorch/torch/autograd/__init__.py", line 197, in backward
    Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
  File "/home/soumith/code/pytorch/torch/autograd/function.py", line 270, in apply
    return user_fn(self, *args)
  File "/home/soumith/code/pytorch/functorch/_src/aot_autograd.py", line 468, in backward
    out = call_func_with_args(
  File "/home/soumith/code/pytorch/functorch/_src/aot_autograd.py", line 255, in call_func_with_args
    out = normalize_as_list(f(args))
  File "/home/soumith/code/pytorch/torch/_dynamo/eval_frame.py", line 157, in _fn
    return fn(*args, **kwargs)
  File "/home/soumith/code/pytorch/torch/_dynamo/debug_utils.py", line 444, in deferred_for_real_inputs
    raise e
  File "/home/soumith/code/pytorch/torch/_dynamo/debug_utils.py", line 428, in deferred_for_real_inputs
    compiled_fn = compiler_fn(gm, example_inputs, **kwargs)
  File "/home/soumith/code/pytorch/torch/_inductor/debug.py", line 178, in inner
    return fn(*args, **kwargs)
  File "/home/soumith/miniconda3/envs/pytorch/lib/python3.9/contextlib.py", line 79, in inner
    return func(*args, **kwds)
  File "/home/soumith/code/pytorch/torch/_inductor/compile_fx.py", line 106, in compile_fx_inner
    graph.run(*example_inputs)
  File "/home/soumith/code/pytorch/torch/_dynamo/utils.py", line 85, in time_wrapper
    r = func(*args, **kwargs)
  File "/home/soumith/code/pytorch/torch/_inductor/graph.py", line 143, in run
    return super().run(*args)
  File "/home/soumith/code/pytorch/torch/fx/interpreter.py", line 130, in run
    self.env[node] = self.run_node(node)
  File "/home/soumith/code/pytorch/torch/_inductor/graph.py", line 311, in run_node
    result = super().run_node(n)
  File "/home/soumith/code/pytorch/torch/fx/interpreter.py", line 171, in run_node
    return getattr(self, n.op)(n.target, args, kwargs)
  File "/home/soumith/code/pytorch/torch/_inductor/graph.py", line 208, in placeholder
    example: torch.Tensor = super().placeholder(target, args, kwargs)
  File "/home/soumith/code/pytorch/torch/fx/interpreter.py", line 204, in placeholder
    raise RuntimeError(f'Expected positional argument for parameter {target}, but one was not passed in!')
RuntimeError: Expected positional argument for parameter primals_1, but one was not passed in!

While executing %primals_1 : [#users=1] = placeholder[target=primals_1]
Original traceback:
None