Merge branch 'main' of https://github.com/pralab/secml-torch

README.md

@@ -23,16 +23,12 @@ The library can be installed together with other plugins that enable further fun
 
 * [Foolbox](https://github.com/bethgelab/foolbox), a Python toolbox to create adversarial examples.
 * [Tensorboard](https://www.tensorflow.org/tensorboard), a visualization toolkit for machine learning experimentation.
+* [Adversarial Library](https://github.com/jeromerony/adversarial-library), a powerful library of various adversarial attacks resources in PyTorch.
 
-Install one or more extras with the command:
-```bash
-pip install secml-torch[foolbox,tensorboard]
-```
-
-To enable the `adv_lib` extra, you have to manually install the library from the original repository:
 
+Install one or more extras with the command:
 ```bash
-pip install git+https://github.com/jeromerony/adversarial-library
+pip install secml-torch[foolbox,tensorboard, adv_lib]
 ```
 
 ## Key Features

src/secmlt/VERSION

@@ -1 +1 @@
-1.2.1
+1.2.2

src/secmlt/adv/evasion/modular_attack.py

@@ -148,7 +148,7 @@ def _init_perturbation_constraints(self) -> list[Constraint]:
         raise NotImplementedError(msg)
 
     def _create_optimizer(self, delta: torch.Tensor, **kwargs) -> Optimizer:
-        return self.optimizer_cls([delta], **kwargs)
+        return self.optimizer_cls([delta], lr=self.step_size, **kwargs)
 
     def forward_loss(
         self, model: BaseModel, x: torch.Tensor, target: torch.Tensor
@@ -181,8 +181,10 @@ def _run(
         samples: torch.Tensor,
         labels: torch.Tensor,
         init_deltas: torch.Tensor = None,
-        **optim_kwargs,
+        optim_kwargs: dict | None = None,
     ) -> tuple[torch.Tensor, torch.Tensor]:
+        if optim_kwargs is None:
+            optim_kwargs = {}
         multiplier = 1 if self.y_target is not None else -1
         target = (
             torch.zeros_like(labels) + self.y_target