eliegoudout

> Do you think this PR is ready for review (it's currently in draft)? It is now ready and I would be grateful for any review :) I also fixed...

@beat-buesser Hello :) Do you think someone can review / merge this PR at some point? I'm not in a hurry, but I'm worried it gets forgotten/frozen. Thanks!

> @eliegoudout Could you please take a look at passing the DCO check? I'm not entirely sure because I'm not very familiar with this, but I think we're good 👍

> @eliegoudout Thank you, I think DCO now looks good. Could you please update your branch [eliegoudout:main](https://github.com/eliegoudout/adversarial-robustness-toolbox-fix-FGM-L1/tree/main) with the most recent version of upstream [Trusted-AI:dev_1.18.0](https://github.com/Trusted-AI/adversarial-robustness-toolbox/tree/dev_1.18.0) to minimise the differences to...

All jobs pass the tests but fail during `upload to Codecov`. Is this something I can fix or is it only a problem from the runners?

Okay, very cool, thanks! It seems that these errors occur because I forgot to change the expected value for $\Vert x-x_{\text{adv}}\Vert$ for BIM: ``` _________________________________ test_images __________________________________ [...] > assert...

> Hi @eliegoudout Could you please take a look at the failing unit tests? It seems the relative tolerance is too tight for some tests. Thank you for your message....

> Hi @eliegoudout Could you please take a look at the failing unit tests? It seems the relative tolerance is too tight for some tests. Done ✅ I realized that...

> Hi @eliegoudout Thank you very much! It looks like we need one more change, the MXNet workflow is still failing at two unit tests and requires a larger tolerance...

I'm not entirely sure but it looks to me after a quick glance that PGD was implemented as a subs class of FGSM and inherits its loss from it.