[eager_fail_to_run] opacus_cifar10

Running from pytorch main

python benchmarks/dynamo/torchbench.py --only opacus_cifar10 --accuracy --no-translation-validation --training --amp --backend inductor --device cuda --output out.csv

cuda train opacus_cifar10
Traceback (most recent call last):
  File "/data/users/ivankobzarev/a/pytorch/benchmarks/dynamo/common.py", line 2744, in validate_model
    self.model_iter_fn(model, example_inputs)
  File "/data/users/ivankobzarev/a/pytorch/benchmarks/dynamo/torchbench.py", line 455, in forward_and_backward_pass
    self.grad_scaler.scale(loss).backward()
  File "/data/users/ivankobzarev/a/pytorch/torch/_tensor.py", line 648, in backward
    torch.autograd.backward(
  File "/data/users/ivankobzarev/a/pytorch/torch/autograd/__init__.py", line 353, in backward
    _engine_run_backward(
  File "/data/users/ivankobzarev/a/pytorch/torch/autograd/graph.py", line 815, in _engine_run_backward
    return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
  File "/data/users/ivankobzarev/a/pytorch/torch/nn/modules/module.py", line 86, in __call__
    return self.hook(module, *args, **kwargs)
  File "/data/users/ivankobzarev/a/opacus/opacus/grad_sample/grad_sample_module.py", line 327, in capture_backprops_hook
    activations, backprops = self.rearrange_grad_samples(
  File "/data/users/ivankobzarev/a/opacus/opacus/grad_sample/grad_sample_module.py", line 383, in rearrange_grad_samples
    raise ValueError(
ValueError: No activations detected for <class 'torch.nn.modules.linear.Linear'>, run forward after add_hooks(model)

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
  File "/data/users/ivankobzarev/a/pytorch/benchmarks/dynamo/common.py", line 4857, in run
    ) = runner.load_model(
  File "/data/users/ivankobzarev/a/pytorch/benchmarks/dynamo/torchbench.py", line 372, in load_model
    self.validate_model(model, example_inputs)
  File "/data/users/ivankobzarev/a/pytorch/benchmarks/dynamo/common.py", line 2746, in validate_model
    raise RuntimeError("Eager run failed") from e
RuntimeError: Eager run failed

eager_fail_to_run

Reproduce the failure with the listed benchmarks/dynamo/torchbench.py command, then start at forward_and_backward_pass and validate_model in the referenced files. Investigate the opacus_cifar10 eager training traceback; done when the benchmark completes its eager training run without the reported missing-activations error.