style: fix formatting and linting issues

pyproject.toml

@@ -144,7 +144,6 @@ src = ["src"]
 docstring-code-format = true
 
 [tool.ruff.lint]
-ignore-init-module-imports = true
 # Enable every possible rule
 select = ["ALL"]
 ignore = [

src/cogelot/app/overfit_model_explore.py

@@ -50,12 +50,12 @@ def load_datamodule(
     return datamodule
 
 
-def simplify_target_actions(actions: Any) -> dict[str, dict[str, torch.Tensor]]:  # noqa: ARG001
+def simplify_target_actions(actions: Any) -> dict[str, dict[str, torch.Tensor]]:
     """Simplify the target actions."""
     raise NotImplementedError
 
 
-def simplify_predicted_actions(actions: Any) -> dict[str, dict[str, torch.Tensor]]:  # noqa: ARG001
+def simplify_predicted_actions(actions: Any) -> dict[str, dict[str, torch.Tensor]]:
     """Simplify the predicted actions."""
     raise NotImplementedError
 

src/cogelot/common/system_deps.py

@@ -6,8 +6,8 @@ def verify_ffmpeg_is_available() -> None:
     """Verify ffmpeg is available."""
     is_available = False
     with suppress(FileNotFoundError):
-        outcome = subprocess.run(
-            ["ffmpeg", "-version"],  # noqa: S603 S607
+        outcome = subprocess.run(  # noqa: S603
+            ["ffmpeg", "-version"],  # noqa: S607
             capture_output=True,
             check=False,
         )

src/cogelot/data/datasets.py

@@ -14,7 +14,7 @@ def load_instance_from_path(
     """Load the instance from the pickled path."""
     try:
         return instance.model_validate(load_from_path_fn(path)).model_dump()
-    except Exception as err:
+    except Exception as err:  # noqa:  BLE001
         logger.exception(f"Something went wrong with {path}")
         raise err from None
 

src/vima_bench/tasks/utils/misc_utils.py

@@ -9,7 +9,6 @@
 import torch
 from transforms3d import euler
 
-
 # -----------------------------------------------------------------------------
 # HEIGHTMAP UTILS
 # -----------------------------------------------------------------------------
@@ -335,7 +334,7 @@ def preprocess(img, dist="transporter"):
         depth_std = transporter_depth_std
 
     # convert to pytorch tensor (if required)
-    if type(img) == torch.Tensor:
+    if type(img) is torch.Tensor:
 
         def cast_shape(stat, img):
             tensor = torch.from_numpy(np.array(stat)).to(device=img.device, dtype=img.dtype)

tests/test_experiments.py

@@ -108,7 +108,7 @@ def test_experiment_can_instantiate_from_config(hydra_config: DictConfig) -> Non
     assert instantiated_module is not None
 
 
-@pytest.mark.trylast()
+@pytest.mark.trylast
 def test_experiment_can_fast_dev_run(
     hydra_config: DictConfig, vima_dataloader: DataLoader[list[PreprocessedInstance]]
 ) -> None:

tests/test_hardware.py

@@ -3,6 +3,6 @@
 
 
 @pytest.mark.env("cuda")
-@pytest.mark.tryfirst()
+@pytest.mark.tryfirst
 def test_cuda_is_available() -> None:
     assert torch.cuda.is_available() is True

tests/test_instance_transforms.py

@@ -49,7 +49,7 @@ def test_reword_transform_works(vima_instance: VIMAInstance) -> None:
     assert new_instance.prompt != vima_instance.prompt
 
 
-@pytest.mark.skip()
+@pytest.mark.skip
 def test_count_all_reword_transforms(vima_instance: VIMAInstance) -> None:
     reword_transform = RewordPromptTransform()
     all_possible_templates = reword_transform.generate_all_possible_prompts(vima_instance)