apply style

isl-org · Sep 4, 2024 · 170b650 · 170b650
1 parent 35ee51e
commit 170b650
Show file tree

Hide file tree

Showing 2 changed files with 38 additions and 14 deletions.
diff --git a/python/open3d/ml/tf/python/layers/neighbor_search.py b/python/open3d/ml/tf/python/layers/neighbor_search.py
@@ -122,9 +122,9 @@ def call(self,
             queries = queries.values
 
         if isinstance(radius, tf.Tensor):
-          radius_ = tf.cast(radius, points.dtype)
+            radius_ = tf.cast(radius, points.dtype)
         else:
-          radius_ = radius
+            radius_ = radius
 
         if points_row_splits is None:
             points_row_splits = tf.cast(tf.stack([0, tf.shape(points)[0]]),

diff --git a/python/test/ml_ops/test_sparseconv.py b/python/test/ml_ops/test_sparseconv.py
@@ -89,9 +89,15 @@ def bias_initializer(a):
     if ml.module.__name__ == 'torch':
         sparse_conv.to(ml.device)
 
-    y = mltest.run_op(ml, ml.device, True, sparse_conv, inp_features,
-                      inp_positions * voxel_size, out_positions * voxel_size,
-                      voxel_size=voxel_size, inp_importance=inp_importance)
+    y = mltest.run_op(ml,
+                      ml.device,
+                      True,
+                      sparse_conv,
+                      inp_features,
+                      inp_positions * voxel_size,
+                      out_positions * voxel_size,
+                      voxel_size=voxel_size,
+                      inp_importance=inp_importance)
 
     # Compare the output to a standard 3d conv
     # store features in a volume to use standard 3d convs
@@ -210,9 +216,15 @@ def test_compare_to_conv3d_batches(ml, dtype, kernel_size, out_channels,
     inp_features = tf.RaggedTensor.from_row_splits(
         values=inp_features, row_splits=inp_positions_row_splits)
 
-    y = mltest.run_op(ml, ml.device, True, sparse_conv, inp_features,
-                      inp_positions * voxel_size, out_positions * voxel_size,
-                      voxel_size=voxel_size, inp_importance=inp_importance)
+    y = mltest.run_op(ml,
+                      ml.device,
+                      True,
+                      sparse_conv,
+                      inp_features,
+                      inp_positions * voxel_size,
+                      out_positions * voxel_size,
+                      voxel_size=voxel_size,
+                      inp_importance=inp_importance)
     for idx in range(batch_size):
         inp_pos = inp_positions[idx].numpy()
         inp_feat = inp_features[idx].numpy()
@@ -336,9 +348,15 @@ def bias_initializer(a):
     if ml.module.__name__ == 'torch':
         sparse_conv_transpose.to(ml.device)
 
-    y = mltest.run_op(ml, ml.device, True, sparse_conv_transpose, inp_features,
-                      inp_positions * voxel_size, out_positions * voxel_size,
-                      voxel_size=voxel_size, out_importance=out_importance)
+    y = mltest.run_op(ml,
+                      ml.device,
+                      True,
+                      sparse_conv_transpose,
+                      inp_features,
+                      inp_positions * voxel_size,
+                      out_positions * voxel_size,
+                      voxel_size=voxel_size,
+                      out_importance=out_importance)
 
     # Compare the output to a standard 3d conv
     # store features in a volume to use standard 3d convs
@@ -463,9 +481,15 @@ def test_compare_to_conv3dtranspose_batches(ml, dtype, kernel_size,
     inp_features = tf.RaggedTensor.from_row_splits(
         values=inp_features, row_splits=inp_positions_row_splits)
 
-    y = mltest.run_op(ml, ml.device, True, sparse_conv_transpose, inp_features,
-                      inp_positions * voxel_size, out_positions * voxel_size,
-                      voxel_size=voxel_size, out_importance=out_importance)
+    y = mltest.run_op(ml,
+                      ml.device,
+                      True,
+                      sparse_conv_transpose,
+                      inp_features,
+                      inp_positions * voxel_size,
+                      out_positions * voxel_size,
+                      voxel_size=voxel_size,
+                      out_importance=out_importance)
     for idx in range(batch_size):
         inp_pos = inp_positions[idx].numpy()
         inp_feat = inp_features[idx].numpy()