remove nn.Upsample deprecation warnings from tests (#17352)

Differential Revision: D14168481

Pulled By: soumith

fbshipit-source-id: 63c37c5f04d2529abd4f42558a3d5e81993eecec

diff --git a/test/common_nn.py b/test/common_nn.py
index 97d2e0a..4879c63 100644 (file)
--- a/test/common_nn.py
+++ b/test/common_nn.py
@@ -1668,206 +1668,206 @@ new_module_tests = [
         input_size=(1, 9, 4, 4),
     ),
     dict(
-        module_name='Upsample',
-        constructor_args=(12, None, 'nearest'),
+        constructor=wrap_functional(F.interpolate, size=12, scale_factor=None, mode='nearest'),
         input_size=(1, 2, 4),
-        desc='nearest_1d',
+        fullname='interpolate_nearest_1d',
+        pickle=False,
     ),
     dict(
-        module_name='Upsample',
-        constructor_args=((12, ), None, 'nearest'),
+        constructor=wrap_functional(F.interpolate, size=(12, ), scale_factor=None, mode='nearest'),
         input_size=(1, 2, 3),
-        desc='nearest_tuple_1d',
+        fullname='interpolate_nearest_tuple_1d',
+        pickle=False,
     ),
     dict(
-        module_name='Upsample',
-        constructor_args=(None, 4., 'nearest'),
+        constructor=wrap_functional(F.interpolate, size=None, scale_factor=4., mode='nearest'),
         input_size=(1, 2, 4),
-        desc='nearest_scale_1d',
+        fullname='interpolate_nearest_scale_1d',
+        pickle=False,
     ),
     dict(
-        module_name='Upsample',
-        constructor_args=(12, None, 'linear', False),
+        constructor=wrap_functional(F.interpolate, size=12, scale_factor=None, mode='linear', align_corners=False),
         input_size=(1, 2, 4),
-        desc='linear_1d',
+        fullname='interpolate_linear_1d',
+        pickle=False,
     ),
     dict(
-        module_name='Upsample',
-        constructor_args=((4, ), None, 'linear', False),
+        constructor=wrap_functional(F.interpolate, size=(4, ), scale_factor=None, mode='linear', align_corners=False),
         input_size=(1, 2, 3),
-        desc='linear_tuple_1d',
+        fullname='interpolate_linear_tuple_1d',
+        pickle=False,
     ),
     dict(
-        module_name='Upsample',
-        constructor_args=(None, 4., 'linear', False),
+        constructor=wrap_functional(F.interpolate, size=None, scale_factor=4., mode='linear', align_corners=False),
         input_size=(1, 2, 4),
-        desc='linear_scale_1d',
+        fullname='interpolate_linear_scale_1d',
+        pickle=False,
     ),
     dict(
-        module_name='Upsample',
-        constructor_args=(12, None, 'linear', True),
+        constructor=wrap_functional(F.interpolate, size=12, scale_factor=None, mode='linear', align_corners=True),
         input_size=(1, 2, 4),
-        desc='linear_1d_align_corners',
+        fullname='interpolate_linear_1d_align_corners',
+        pickle=False,
     ),
     dict(
-        module_name='Upsample',
-        constructor_args=(None, 4., 'linear', True),
+        constructor=wrap_functional(F.interpolate, size=None, scale_factor=4., mode='linear', align_corners=True),
         input_size=(1, 2, 4),
-        desc='linear_scale_1d_align_corners',
+        fullname='interpolate_linear_scale_1d_align_corners',
+        pickle=False,
     ),
     dict(
-        module_name='Upsample',
-        constructor_args=(12, None, 'nearest'),
+        constructor=wrap_functional(F.interpolate, size=12, scale_factor=None, mode='nearest'),
         input_size=(1, 2, 4, 4),
-        desc='nearest_2d',
+        fullname='interpolate_nearest_2d',
+        pickle=False,
     ),
     dict(
-        module_name='Upsample',
-        constructor_args=((12, 16), None, 'nearest'),
+        constructor=wrap_functional(F.interpolate, size=(12, 16), scale_factor=None, mode='nearest'),
         input_size=(1, 2, 3, 4),
-        desc='nearest_tuple_2d',
+        fullname='interpolate_nearest_tuple_2d',
+        pickle=False,
     ),
     dict(
-        module_name='Upsample',
-        constructor_args=(None, 4., 'nearest'),
+        constructor=wrap_functional(F.interpolate, size=None, scale_factor=4., mode='nearest'),
         input_size=(1, 2, 4, 4),
-        desc='nearest_scale_2d',
+        fullname='interpolate_nearest_scale_2d',
+        pickle=False,
     ),
     dict(
-        module_name='Upsample',
-        constructor_args=(12, None, 'bilinear', False),
+        constructor=wrap_functional(F.interpolate, size=12, scale_factor=None, mode='bilinear', align_corners=False),
         input_size=(1, 2, 4, 4),
-        desc='bilinear_2d',
+        fullname='interpolate_bilinear_2d',
+        pickle=False,
     ),
     dict(
-        module_name='Upsample',
-        constructor_args=((4, 6), None, 'bilinear', False),
+        constructor=wrap_functional(F.interpolate, size=(4, 6), scale_factor=None, mode='bilinear', align_corners=False),
         input_size=(1, 2, 2, 3),
-        desc='bilinear_tuple_2d',
+        fullname='interpolate_bilinear_tuple_2d',
+        pickle=False,
     ),
     dict(
-        module_name='Upsample',
-        constructor_args=(None, 4., 'bilinear', False),
+        constructor=wrap_functional(F.interpolate, size=None, scale_factor=4., mode='bilinear', align_corners=False),
         input_size=(1, 2, 4, 4),
-        desc='bilinear_scale_2d',
+        fullname='interpolate_bilinear_scale_2d',
+        pickle=False,
     ),
     dict(
-        module_name='Upsample',
-        constructor_args=(None, (2., 2.), 'bilinear', False),
+        constructor=wrap_functional(F.interpolate, size=None, scale_factor=(2., 2.), mode='bilinear', align_corners=False),
         input_size=(1, 2, 4, 4),
-        desc='bilinear_scale_tuple_shared_2d',
+        fullname='interpolate_bilinear_scale_tuple_shared_2d',
+        pickle=False,
     ),
     dict(
-        module_name='Upsample',
-        constructor_args=(None, (2., 1.), 'bilinear', False),
+        constructor=wrap_functional(F.interpolate, size=None, scale_factor=(2., 1.), mode='bilinear', align_corners=False),
         input_size=(1, 2, 4, 4),
-        desc='bilinear_scale_tuple_skewed_2d',
+        fullname='interpolate_bilinear_scale_tuple_skewed_2d',
+        pickle=False,
     ),
     dict(
-        module_name='Upsample',
-        constructor_args=((4, 6), None, 'bilinear', True),
+        constructor=wrap_functional(F.interpolate, size=(4, 6), scale_factor=None, mode='bilinear', align_corners=True),
         input_size=(1, 2, 4, 4),
-        desc='bilinear_tuple_2d_align_corners',
+        fullname='interpolate_bilinear_tuple_2d_align_corners',
+        pickle=False,
     ),
     dict(
-        module_name='Upsample',
-        constructor_args=(None, (2., 1.), 'bilinear', True),
+        constructor=wrap_functional(F.interpolate, size=None, scale_factor=(2., 1.), mode='bilinear', align_corners=True),
         input_size=(1, 2, 4, 4),
-        desc='bilinear_scale_tuple_skewed_2d_align_corners',
+        fullname='interpolate_bilinear_scale_tuple_skewed_2d_align_corners',
+        pickle=False,
     ),
     dict(
-        module_name='Upsample',
-        constructor_args=(12, None, 'bicubic', False),
+        constructor=wrap_functional(F.interpolate, size=12, scale_factor=None, mode='bicubic', align_corners=False),
         input_size=(1, 2, 4, 4),
-        desc='bicubic_2d',
+        fullname='interpolate_bicubic_2d',
+        pickle=False,
     ),
     dict(
-        module_name='Upsample',
-        constructor_args=((4, 6), None, 'bicubic', False),
+        constructor=wrap_functional(F.interpolate, size=(4, 6), scale_factor=None, mode='bicubic', align_corners=False),
         input_size=(1, 2, 2, 3),
-        desc='bicubic_tuple_2d',
+        fullname='interpolate_bicubic_tuple_2d',
+        pickle=False,
     ),
     dict(
-        module_name='Upsample',
-        constructor_args=(None, 4., 'bicubic', False),
+        constructor=wrap_functional(F.interpolate, size=None, scale_factor=4., mode='bicubic', align_corners=False),
         input_size=(1, 2, 4, 4),
-        desc='bicubic_scale_2d',
+        fullname='interpolate_bicubic_scale_2d',
+        pickle=False,
     ),
     dict(
-        module_name='Upsample',
-        constructor_args=(None, (2., 2.), 'bicubic', False),
+        constructor=wrap_functional(F.interpolate, size=None, scale_factor=(2., 2.), mode='bicubic', align_corners=False),
         input_size=(1, 2, 4, 4),
-        desc='bicubic_scale_tuple_shared_2d',
+        fullname='interpolate_bicubic_scale_tuple_shared_2d',
+        pickle=False,
     ),
     dict(
-        module_name='Upsample',
-        constructor_args=(None, (2., 1.), 'bicubic', False),
+        constructor=wrap_functional(F.interpolate, size=None, scale_factor=(2., 1.), mode='bicubic', align_corners=False),
         input_size=(1, 2, 4, 4),
-        desc='bicubic_scale_tuple_skewed_2d',
+        fullname='interpolate_bicubic_scale_tuple_skewed_2d',
+        pickle=False,
     ),
     dict(
-        module_name='Upsample',
-        constructor_args=((4, 6), None, 'bicubic', True),
+        constructor=wrap_functional(F.interpolate, size=(4, 6), scale_factor=None, mode='bicubic', align_corners=True),
         input_size=(1, 2, 4, 4),
-        desc='bicubic_tuple_2d_align_corners',
+        fullname='interpolate_bicubic_tuple_2d_align_corners',
+        pickle=False,
     ),
     dict(
-        module_name='Upsample',
-        constructor_args=(None, (2., 1.), 'bicubic', True),
+        constructor=wrap_functional(F.interpolate, size=None, scale_factor=(2., 1.), mode='bicubic', align_corners=True),
         input_size=(1, 2, 4, 4),
-        desc='bicubic_scale_tuple_skewed_2d_align_corners',
+        fullname='interpolate_bicubic_scale_tuple_skewed_2d_align_corners',
+        pickle=False,
     ),
     dict(
-        module_name='Upsample',
-        constructor_args=(12, None, 'nearest'),
+        constructor=wrap_functional(F.interpolate, size=12, scale_factor=None, mode='nearest'),
         input_size=(1, 2, 4, 4, 4),
-        desc='nearest_3d',
+        fullname='interpolate_nearest_3d',
+        pickle=False,
     ),
     dict(
-        module_name='Upsample',
-        constructor_args=((12, 16, 16), None, 'nearest'),
+        constructor=wrap_functional(F.interpolate, size=(12, 16, 16), scale_factor=None, mode='nearest'),
         input_size=(1, 2, 3, 4, 4),
-        desc='nearest_tuple_3d',
+        fullname='interpolate_nearest_tuple_3d',
+        pickle=False,
     ),
     dict(
-        module_name='Upsample',
-        constructor_args=(None, 4., 'nearest'),
+        constructor=wrap_functional(F.interpolate, size=None, scale_factor=4., mode='nearest'),
         input_size=(1, 2, 4, 4, 4),
-        desc='nearest_scale_3d',
+        fullname='interpolate_nearest_scale_3d',
+        pickle=False,
     ),
     dict(
-        module_name='Upsample',
-        constructor_args=(12, None, 'trilinear', False),
+        constructor=wrap_functional(F.interpolate, size=12, scale_factor=None, mode='trilinear', align_corners=False),
         input_size=(1, 2, 4, 4, 4),
-        desc='trilinear_3d',
+        fullname='interpolate_trilinear_3d',
+        pickle=False,
     ),
     dict(
-        module_name='Upsample',
-        constructor_args=((4, 6, 6), None, 'trilinear', False),
+        constructor=wrap_functional(F.interpolate, size=(4, 6, 6), scale_factor=None, mode='trilinear', align_corners=False),
         input_size=(1, 2, 2, 3, 3),
-        desc='trilinear_tuple_3d',
+        fullname='interpolate_trilinear_tuple_3d',
+        pickle=False,
     ),
     dict(
-        module_name='Upsample',
-        constructor_args=(None, 3., 'trilinear', False),
+        constructor=wrap_functional(F.interpolate, size=None, scale_factor=3., mode='trilinear', align_corners=False),
         input_size=(1, 2, 3, 4, 4),
-        desc='trilinear_scale_3d',
+        fullname='interpolate_trilinear_scale_3d',
         # See https://github.com/pytorch/pytorch/issues/5006
         precision=3e-4,
+        pickle=False,
     ),
     dict(
-        module_name='Upsample',
-        constructor_args=((4, 6, 6), None, 'trilinear', True),
+        constructor=wrap_functional(F.interpolate, size=(4, 6, 6), scale_factor=None, mode='trilinear', align_corners=True),
         input_size=(1, 2, 2, 3, 3),
-        desc='trilinear_tuple_3d_align_corners',
+        fullname='interpolate_trilinear_tuple_3d_align_corners',
+        pickle=False,
     ),
     dict(
-        module_name='Upsample',
-        constructor_args=(None, 3., 'trilinear', True),
+        constructor=wrap_functional(F.interpolate, size=None, scale_factor=3., mode='trilinear', align_corners=True),
         input_size=(1, 2, 3, 4, 4),
-        desc='trilinear_scale_3d_align_corners',
+        fullname='interpolate_trilinear_scale_3d_align_corners',
         # See https://github.com/pytorch/pytorch/issues/5006
         precision=3e-4,
+        pickle=False,
     ),
     dict(
         module_name='AdaptiveMaxPool1d',

diff --git a/test/test_nn.py b/test/test_nn.py
index 9d61732..0a5d070 100644 (file)
--- a/test/test_nn.py
+++ b/test/test_nn.py
@@ -6432,7 +6432,8 @@ class TestNN(NNTestCase):
     def test_upsamplingNearest1d(self):
         m = nn.Upsample(size=4, mode='nearest')
         in_t = torch.ones(1, 1, 2)
-        out_t = m(Variable(in_t))
+        with warnings.catch_warnings(record=True) as w:
+            out_t = m(in_t)
         self.assertEqual(torch.ones(1, 1, 4), out_t.data)
 
         input = torch.randn(1, 1, 2, requires_grad=True)
@@ -6447,7 +6448,8 @@ class TestNN(NNTestCase):
                 m = nn.Upsample(scale_factor=scale_factor, **kwargs)
                 in_t = torch.ones(1, 1, 2)
                 out_size = int(math.floor(in_t.shape[-1] * scale_factor))
-                out_t = m(in_t)
+                with warnings.catch_warnings(record=True) as w:
+                    out_t = m(in_t)
                 self.assertEqual(torch.ones(1, 1, out_size), out_t.data)
 
                 input = torch.randn(1, 1, 2, requires_grad=True)
@@ -6457,14 +6459,16 @@ class TestNN(NNTestCase):
         m = nn.Upsample(scale_factor=3, mode='linear', align_corners=False)
         in_t_9 = torch.zeros(1, 1, 9)
         in_t_9[:, :, :4].normal_()
-        out_t_9 = m(in_t_9)
-        out_t_5 = m(in_t_9[:, :, :5])
+        with warnings.catch_warnings(record=True) as w:
+            out_t_9 = m(in_t_9)
+            out_t_5 = m(in_t_9[:, :, :5])
         self.assertEqual(out_t_9[:, :, :15], out_t_5)
 
     def test_upsamplingNearest2d(self):
         m = nn.Upsample(size=4, mode='nearest')
         in_t = torch.ones(1, 1, 2, 2)
-        out_t = m(Variable(in_t))
+        with warnings.catch_warnings(record=True) as w:
+            out_t = m(Variable(in_t))
         self.assertEqual(torch.ones(1, 1, 4, 4), out_t.data)
 
         input = torch.randn(1, 1, 2, 2, requires_grad=True)
@@ -6483,7 +6487,8 @@ class TestNN(NNTestCase):
                 m = nn.Upsample(scale_factor=scale_factor, **kwargs)
                 in_t = torch.ones(1, 1, 2, 2)
                 out_size = int(math.floor(in_t.shape[-1] * scale_factor))
-                out_t = m(in_t)
+                with warnings.catch_warnings(record=True) as w:
+                    out_t = m(in_t)
                 self.assertEqual(torch.ones(1, 1, out_size, out_size), out_t.data)
 
                 input = torch.randn(1, 1, 2, 2, requires_grad=True)
@@ -6518,14 +6523,16 @@ class TestNN(NNTestCase):
         m = nn.Upsample(scale_factor=3, mode='bilinear', align_corners=False)
         in_t_9 = torch.zeros(1, 1, 9, 9)
         in_t_9[:, :, :4, :4].normal_()
-        out_t_9 = m(in_t_9)
-        out_t_5 = m(in_t_9[:, :, :5, :5])
+        with warnings.catch_warnings(record=True) as w:
+            out_t_9 = m(in_t_9)
+            out_t_5 = m(in_t_9[:, :, :5, :5])
         self.assertEqual(out_t_9[:, :, :15, :15], out_t_5)
 
     def test_upsamplingNearest3d(self):
         m = nn.Upsample(size=4, mode='nearest')
         in_t = torch.ones(1, 1, 2, 2, 2)
-        out_t = m(Variable(in_t))
+        with warnings.catch_warnings(record=True) as w:
+            out_t = m(Variable(in_t))
         self.assertEqual(torch.ones(1, 1, 4, 4, 4), out_t.data)
 
         input = torch.randn(1, 1, 2, 2, 2, requires_grad=True)
@@ -6540,7 +6547,8 @@ class TestNN(NNTestCase):
                 m = nn.Upsample(scale_factor=scale_factor, **kwargs)
                 in_t = torch.ones(1, 1, 2, 2, 2)
                 out_size = int(math.floor(in_t.shape[-1] * scale_factor))
-                out_t = m(in_t)
+                with warnings.catch_warnings(record=True) as w:
+                    out_t = m(in_t)
                 self.assertEqual(torch.ones(1, 1, out_size, out_size, out_size), out_t.data)
 
                 input = torch.randn(1, 1, 2, 2, 2, requires_grad=True)
@@ -6554,8 +6562,9 @@ class TestNN(NNTestCase):
         m = nn.Upsample(scale_factor=3, mode='trilinear', align_corners=False)
         in_t_9 = torch.zeros(1, 1, 9, 9, 9)
         in_t_9[:, :, :4, :4, :4].normal_()
-        out_t_9 = m(in_t_9)
-        out_t_5 = m(in_t_9[:, :, :5, :5, :5])
+        with warnings.catch_warnings(record=True) as w:
+            out_t_9 = m(in_t_9)
+            out_t_5 = m(in_t_9[:, :, :5, :5, :5])
         self.assertEqual(out_t_9[:, :, :15, :15, :15], out_t_5)
 
     def test_interpolate(self):
@@ -6563,7 +6572,8 @@ class TestNN(NNTestCase):
             out_size = int(math.floor(in_t.shape[-1] * scale_factor))
             dim = len(in_t.shape) - 2
             out_shape = [1, 1] + [out_size] * dim
-            out_t = m(in_t)
+            with warnings.catch_warnings(record=True) as w:
+                out_t = m(in_t)
             self.assertEqual(torch.ones(out_shape), out_t)
 
             self.assertEqual(