replace uses of np.ndarray with npt.NDArray (pytorch#1387)

Summary:
X-link: pytorch/audio#3846

X-link: pytorch/opacus#680
X-link: pytorch/botorch#2584
X-link: pytorch/audio#3845

This replaces uses of `numpy.ndarray` in type annotations with `numpy.typing.NDArray`. In Numpy-1.24.0+ `numpy.ndarray` is annotated as generic type. Without template parameters it triggers static analysis errors: 
```counterexample
Generic type `ndarray` expects 2 type parameters.
```
`numpy.typing.NDArray` is an alias that provides default template parameters.

Differential Revision: D64619891

captum/attr/_utils/visualization.py

@@ -8,6 +8,7 @@
 import matplotlib
 
 import numpy as np
+import numpy.typing as npt
 from matplotlib import cm, colors, pyplot as plt
 from matplotlib.axes import Axes
 from matplotlib.collections import LineCollection
@@ -47,11 +48,11 @@ class VisualizeSign(Enum):
     all = 4
 
 
-def _prepare_image(attr_visual: ndarray) -> ndarray:
+def _prepare_image(attr_visual: npt.NDArray) -> npt.NDArray:
     return np.clip(attr_visual.astype(int), 0, 255)
 
 
-def _normalize_scale(attr: ndarray, scale_factor: float) -> ndarray:
+def _normalize_scale(attr: npt.NDArray, scale_factor: float) -> npt.NDArray:
     assert scale_factor != 0, "Cannot normalize by scale factor = 0"
     if abs(scale_factor) < 1e-5:
         warnings.warn(
@@ -64,7 +65,9 @@ def _normalize_scale(attr: ndarray, scale_factor: float) -> ndarray:
     return np.clip(attr_norm, -1, 1)
 
 
-def _cumulative_sum_threshold(values: ndarray, percentile: Union[int, float]) -> float:
+def _cumulative_sum_threshold(
+    values: npt.NDArray, percentile: Union[int, float]
+) -> float:
     # given values should be non-negative
     assert percentile >= 0 and percentile <= 100, (
         "Percentile for thresholding must be " "between 0 and 100 inclusive."
@@ -76,11 +79,11 @@ def _cumulative_sum_threshold(values: ndarray, percentile: Union[int, float]) ->
 
 
 def _normalize_attr(
-    attr: ndarray,
+    attr: npt.NDArray,
     sign: str,
     outlier_perc: Union[int, float] = 2,
     reduction_axis: Optional[int] = None,
-) -> ndarray:
+) -> npt.NDArray:
     attr_combined = attr
     if reduction_axis is not None:
         attr_combined = np.sum(attr, axis=reduction_axis)
@@ -130,7 +133,7 @@ def _initialize_cmap_and_vmin_vmax(
 
 def _visualize_original_image(
     plt_axis: Axes,
-    original_image: Optional[ndarray],
+    original_image: Optional[npt.NDArray],
     **kwargs: Any,
 ) -> None:
     assert (
@@ -143,7 +146,7 @@ def _visualize_original_image(
 
 def _visualize_heat_map(
     plt_axis: Axes,
-    norm_attr: ndarray,
+    norm_attr: npt.NDArray,
     cmap: Union[str, Colormap],
     vmin: float,
     vmax: float,
@@ -155,8 +158,8 @@ def _visualize_heat_map(
 
 def _visualize_blended_heat_map(
     plt_axis: Axes,
-    original_image: ndarray,
-    norm_attr: ndarray,
+    original_image: npt.NDArray,
+    norm_attr: npt.NDArray,
     cmap: Union[str, Colormap],
     vmin: float,
     vmax: float,
@@ -176,8 +179,8 @@ def _visualize_blended_heat_map(
 def _visualize_masked_image(
     plt_axis: Axes,
     sign: str,
-    original_image: ndarray,
-    norm_attr: ndarray,
+    original_image: npt.NDArray,
+    norm_attr: npt.NDArray,
     **kwargs: Any,
 ) -> None:
     assert VisualizeSign[sign].value != VisualizeSign.all.value, (
@@ -190,8 +193,8 @@ def _visualize_masked_image(
 def _visualize_alpha_scaling(
     plt_axis: Axes,
     sign: str,
-    original_image: ndarray,
-    norm_attr: ndarray,
+    original_image: npt.NDArray,
+    norm_attr: npt.NDArray,
     **kwargs: Any,
 ) -> None:
     assert VisualizeSign[sign].value != VisualizeSign.all.value, (
@@ -210,8 +213,8 @@ def _visualize_alpha_scaling(
 
 
 def visualize_image_attr(
-    attr: ndarray,
-    original_image: Optional[ndarray] = None,
+    attr: npt.NDArray,
+    original_image: Optional[npt.NDArray] = None,
     method: str = "heat_map",
     sign: str = "absolute_value",
     plt_fig_axis: Optional[Tuple[Figure, Axes]] = None,
@@ -417,8 +420,8 @@ def visualize_image_attr(
 
 
 def visualize_image_attr_multiple(
-    attr: ndarray,
-    original_image: Union[None, ndarray],
+    attr: npt.NDArray,
+    original_image: Union[None, npt.NDArray],
     methods: List[str],
     signs: List[str],
     titles: Optional[List[str]] = None,
@@ -526,9 +529,9 @@ def visualize_image_attr_multiple(
 
 
 def visualize_timeseries_attr(
-    attr: ndarray,
-    data: ndarray,
-    x_values: Optional[ndarray] = None,
+    attr: npt.NDArray,
+    data: npt.NDArray,
+    x_values: Optional[npt.NDArray] = None,
     method: str = "overlay_individual",
     sign: str = "absolute_value",
     channel_labels: Optional[List[str]] = None,

tests/attr/test_gradient_shap.py

@@ -5,6 +5,7 @@
 from typing import cast, Tuple
 
 import numpy as np
+import numpy.typing as npt
 import torch
 from captum._utils.typing import Tensor
 from captum.attr._core.gradient_shap import GradientShap
@@ -132,7 +133,7 @@ def generate_baselines_with_inputs(inputs: Tensor) -> Tensor:
             inp_shape = cast(Tuple[int, ...], inputs.shape)
             return torch.arange(0.0, inp_shape[1] * 2.0).reshape(2, inp_shape[1])
 
-        def generate_baselines_returns_array() -> ndarray:
+        def generate_baselines_returns_array() -> npt.NDArray:
             return np.arange(0.0, num_in * 4.0).reshape(4, num_in)
 
         # 10-class classification model

tests/utils/models/linear_models/_test_linear_classifier.py

@@ -5,6 +5,7 @@
 
 import captum._utils.models.linear_model.model as pytorch_model_module
 import numpy as np
+import numpy.typing as npt
 import sklearn.datasets as datasets
 import torch
 from tests.helpers.evaluate_linear_model import evaluate
@@ -107,7 +108,7 @@ def compare_to_sk_learn(
         o_sklearn["l1_reg"] = alpha * sklearn_h.norm(p=1, dim=-1)
 
     rel_diff = cast(
-        np.ndarray,
+        npt.NDArray,
         # pyre-fixme[6]: For 1st argument expected `int` but got `Union[int, Tensor]`.
         (sum(o_sklearn.values()) - sum(o_pytorch.values())),
     ) / abs(sum(o_sklearn.values()))