Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Move Perturber to Composer.Perturber #231

Merged
merged 15 commits into from
Sep 28, 2023
Merged

Move Perturber to Composer.Perturber #231

merged 15 commits into from
Sep 28, 2023

Conversation

mzweilin
Copy link
Contributor

@mzweilin mzweilin commented Sep 27, 2023

What does this PR do?

This PR simplifies the abstraction of Adversary by moving adversary.perturber into adversary.composer.perturber

A Composer is a module that generates perturbed input from the benign input and the perturbation. The perturbation is the weights of the Composer module. Perturber is an object to manage the perturbation (e.g. initialization, projection) and return perturbation for Composer to work on.

A universal adversary should be able to use a Composer in the same way.

In the future, we may try to better integrate components in Perturber, such as Initializer and Projector, into Composer directly. For example, the image initializer should be implemented in Composer instead, so that we can easily constrain Lp bound on perturbation.

Type of change

Please check all relevant options.

  • Improvement (non-breaking)
  • Bug fix (non-breaking)
  • New feature (non-breaking)
  • Breaking change (fix or feature that would cause existing functionality to not work as expected)
  • This change requires a documentation update

Testing

Please describe the tests that you ran to verify your changes. Consider listing any relevant details of your test configuration.

  • pytest
  • CUDA_VISIBLE_DEVICES=0 python -m mart experiment=CIFAR10_CNN_Adv trainer=gpu trainer.precision=16 reports 70% (21 sec/epoch).
  • CUDA_VISIBLE_DEVICES=0,1 python -m mart experiment=CIFAR10_CNN_Adv trainer=ddp trainer.precision=16 trainer.devices=2 model.optimizer.lr=0.2 trainer.max_steps=2925 datamodule.ims_per_batch=256 datamodule.world_size=2 reports 70% (14 sec/epoch).

Before submitting

  • The title is self-explanatory and the description concisely explains the PR
  • My PR does only one thing, instead of bundling different changes together
  • I list all the breaking changes introduced by this pull request
  • I have commented my code
  • I have added tests that prove my fix is effective or that my feature works
  • New and existing unit tests pass locally with my changes
  • I have run pre-commit hooks with pre-commit run -a command without errors

Did you have fun?

Make sure you had fun coding 🙃

@mzweilin mzweilin marked this pull request as ready for review September 27, 2023 23:32
Base automatically changed from composer_functions to main September 28, 2023 04:54
@mzweilin mzweilin requested a review from dxoigmn September 28, 2023 17:00
Copy link
Contributor

@dxoigmn dxoigmn left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM. Can you make sure the DDP test runs fine?

@mzweilin
Copy link
Contributor Author

LGTM. Can you make sure the DDP test runs fine?

I confirm that DDP test runs fine. I'm merging this PR.

@mzweilin mzweilin merged commit 6fb35a5 into main Sep 28, 2023
@mzweilin mzweilin deleted the perturber_in_composer branch September 28, 2023 22:36
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants