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

Add tests for Adversary at arbitrary layer in model. #111

Open
dxoigmn opened this issue Mar 20, 2023 · 0 comments
Open

Add tests for Adversary at arbitrary layer in model. #111

dxoigmn opened this issue Mar 20, 2023 · 0 comments

Comments

@dxoigmn
Copy link
Contributor

dxoigmn commented Mar 20, 2023

Right now, the Adversary checks whether model is present in order to determine when to attack:

if model is not None:

However, because an Adversary can live at any layer, it should really check whether sequence is present:

output = module(step=step, sequence=sequence, **kwargs)

The reason is that the sequence represents the "continuation" of the model at the point where the Adversary lives in the model. Meaning that the Adversary can call the sequence to get a loss. Calling the model may be incorrect depending upon at which layer of model the Adversary lives.

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

No branches or pull requests

1 participant