Add the inline_inbuilt_nn_modules option

[chaojun-zhang]

Disabling the inline_inbuilt_nn_modules switch will restore ALBERT-XXL's performance to PyTorch 2.4 levels

What does this PR do?

Fixes # (issue)

Before submitting

• This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
• Did you make sure to update the documentation with your changes?
• Did you write any new necessary tests?

[ssarkar2]

Adding notes for my understanding

without this change:
torch._dynamo.config.inline_inbuilt_nn_modules = True
https://github.com/pytorch/pytorch/blob/main/torch/_dynamo/config.py#L412C33-L412C42

with this change we give an option to set it to True or False. We set it to false in case of albert

[ssarkar2]

@chaojun-zhang could you please resolve conflict. We'll try to merge it in for 1.20

[chaojun-zhang]

@chaojun-zhang could you please resolve conflict. We'll try to merge it in for 1.20

Updated

[HuggingFaceDocBuilderDev]

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.

[regisss]

@chaojun-zhang One more merge conflict to solve since I just merged #1619

[ssarkar2]

@chaojun-zhang One more merge conflict to solve since I just merged #1619

Libin pushed a commit to resolve conflicts

[github-actions]

The code quality check failed, please run make style.

[regisss]

LGTM!