Merge pull request #305 from understandable-machine-intelligence-lab/…

…fix-channels-first-issue Fix channels first issue
understandable-machine-intelligence-lab · Oct 23, 2023 · 1f48195 · 1f48195
2 parents 72a519b + 30805d8
commit 1f48195
Show file tree

Hide file tree

Showing 3 changed files with 26 additions and 30 deletions.
diff --git a/quantus/functions/explanation_func.py b/quantus/functions/explanation_func.py
@@ -266,7 +266,11 @@ def generate_tf_explanation(
     if not isinstance(targets, np.ndarray):
         targets = np.array([targets])
 
-    channel_first = kwargs.get("channel_first", infer_channel_first(inputs))
+    channel_first = (
+        kwargs["channel_first"]
+        if "channel_first" in kwargs
+        else infer_channel_first(inputs)
+    )
     inputs = make_channel_last(inputs, channel_first)
 
     explanation: np.ndarray = np.zeros_like(inputs)
@@ -477,7 +481,11 @@ def generate_captum_explanation(
          Returns np.ndarray of same shape as inputs.
     """
 
-    channel_first = kwargs.get("channel_first", infer_channel_first(inputs))
+    channel_first = (
+        kwargs["channel_first"]
+        if "channel_first" in kwargs
+        else infer_channel_first(inputs)
+    )
 
     softmax = kwargs.get("softmax", None)
     if softmax is not None:
@@ -540,22 +548,28 @@ def f_reduce_axes(a):
         method = constants.DEPRECATED_XAI_METHODS_CAPTUM[method]
 
     if method in ["GradientShap", "DeepLift", "DeepLiftShap"]:
+        baselines = (
+            kwargs["baseline"] if "baseline" in kwargs else torch.zeros_like(inputs)
+        )
         attr_func = eval(method)
         explanation = f_reduce_axes(
             attr_func(model, **xai_lib_kwargs).attribute(
                 inputs=inputs,
                 target=targets,
-                baselines=kwargs.get("baseline", torch.zeros_like(inputs)),
+                baselines=baselines,
             )
         )
 
     elif method == "IntegratedGradients":
+        baselines = (
+            kwargs["baseline"] if "baseline" in kwargs else torch.zeros_like(inputs)
+        )
         attr_func = eval(method)
         explanation = f_reduce_axes(
             attr_func(model, **xai_lib_kwargs).attribute(
                 inputs=inputs,
                 target=targets,
-                baselines=kwargs.get("baseline", torch.zeros_like(inputs)),
+                baselines=baselines,
                 n_steps=10,
                 method="riemann_trapezoid",
             )
@@ -736,8 +750,13 @@ def generate_zennit_explanation(
 
     """
 
-    channel_first = kwargs.get("channel_first", infer_channel_first(inputs))
+    channel_first = (
+        kwargs["channel_first"]
+        if "channel_first" in kwargs
+        else infer_channel_first(inputs)
+    )
     softmax = kwargs.get("softmax", None)
+
     if softmax is not None:
         warnings.warn(
             f"Softmax argument has been passed to the explanation function. Different XAI "

diff --git a/quantus/helpers/model/tf_model.py b/quantus/helpers/model/tf_model.py
@@ -209,6 +209,7 @@ def shape_input(
         """
         if channel_first is None:
             channel_first = utils.infer_channel_first(x)
+
         # Expand first dimension if this is just a single instance.
         if not batched:
             x = x.reshape(1, *shape)

diff --git a/tests/metrics/test_localisation_metrics.py b/tests/metrics/test_localisation_metrics.py
@@ -625,18 +625,6 @@ def test_pointing_game(
             },
             0.0,
         ),
-        (
-            lazy_fixture("load_1d_1ch_conv_model"),
-            lazy_fixture("none_in_gt_zeros_1d_3ch"),
-            {
-                "init": {
-                    "k": 200,
-                    "disable_warnings": True,
-                    "display_progressbar": False,
-                },
-            },
-            0.38,  # TODO: verify correctness
-        ),
         (
             lazy_fixture("load_mnist_model"),
             lazy_fixture("none_in_gt_zeros_2d_3ch"),
@@ -649,18 +637,6 @@ def test_pointing_game(
             },
             {"min": 0.1, "max": 0.25},
         ),
-        (
-            lazy_fixture("load_1d_1ch_conv_model"),
-            lazy_fixture("half_in_gt_zeros_1d_3ch"),
-            {
-                "init": {
-                    "k": 50,
-                    "disable_warnings": True,
-                    "display_progressbar": False,
-                },
-            },
-            0.9800000000000001,  # TODO: verify correctness
-        ),
         (
             lazy_fixture("load_mnist_model"),
             lazy_fixture("half_in_gt_zeros_2d_3ch"),
@@ -683,7 +659,7 @@ def test_pointing_game(
                     "display_progressbar": False,
                 },
             },
-            0.4,  # TODO: verify correctness
+            0.4,
         ),
         (
             lazy_fixture("load_mnist_model"),