Skip to content

Commit 4f4d58d

Browse files
GregoryComerGregoryComer
committed
Update
[ghstack-poisoned]
2 parents 8b11366 + cc3332e commit 4f4d58d

File tree

11 files changed

+984
-478
lines changed

11 files changed

+984
-478
lines changed

‎backends/test/suite/operators/__init__.py‎

Lines changed: 4 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -133,7 +133,9 @@ def _create_test_for_backend(
133133

134134

135135
class OperatorTest(unittest.TestCase):
136-
def _test_op(self, model, inputs, flow: TestFlow):
136+
def _test_op(
137+
self, model, inputs, flow: TestFlow, generate_random_test_inputs: bool = True
138+
):
137139
context = get_active_test_context()
138140

139141
# This should be set in the wrapped test. See _make_wrapped_test above.
@@ -145,6 +147,7 @@ def _test_op(self, model, inputs, flow: TestFlow):
145147
flow,
146148
context.test_name,
147149
context.params,
150+
generate_random_test_inputs=generate_random_test_inputs,
148151
)
149152

150153
log_test_summary(run_summary)

‎backends/test/suite/operators/test_conv1d.py‎

Lines changed: 96 additions & 48 deletions
Original file line numberDiff line numberDiff line change
@@ -1,17 +1,22 @@
1-
# (c) Meta Platforms, Inc. and affiliates. Confidential and proprietary.
1+
# Copyright (c) Meta Platforms, Inc. and affiliates.
2+
# All rights reserved.
3+
#
4+
# This source code is licensed under the BSD-style license found in the
5+
# LICENSE file in the root directory of this source tree.
26

3-
# pyre-strict
7+
# pyre-unsafe
48

5-
from typing import Callable
69

710
import torch
11+
from executorch.backends.test.suite.flow import TestFlow
812

9-
from executorch.backends.test.compliance_suite import (
13+
from executorch.backends.test.suite.operators import (
1014
dtype_test,
1115
operator_test,
1216
OperatorTest,
1317
)
1418

19+
1520
class Model(torch.nn.Module):
1621
def __init__(
1722
self,
@@ -37,53 +42,96 @@ def __init__(
3742
bias=bias,
3843
padding_mode=padding_mode,
3944
)
40-
45+
4146
def forward(self, x):
4247
return self.conv(x)
4348

49+
4450
@operator_test
45-
class TestConv1d(OperatorTest):
51+
class Conv1d(OperatorTest):
4652
@dtype_test
47-
def test_conv1d_dtype(self, dtype, tester_factory: Callable) -> None:
48-
# Input shape: (batch_size, in_channels, length)
49-
self._test_op(Model().to(dtype), ((torch.rand(2, 3, 10) * 10).to(dtype),), tester_factory)
50-
51-
def test_conv1d_basic(self, tester_factory: Callable) -> None:
52-
# Basic test with default parameters
53-
self._test_op(Model(), (torch.randn(2, 3, 10),), tester_factory)
54-
55-
def test_conv1d_kernel_size(self, tester_factory: Callable) -> None:
56-
# Test with different kernel sizes
57-
self._test_op(Model(kernel_size=1), (torch.randn(2, 3, 10),), tester_factory)
58-
self._test_op(Model(kernel_size=5), (torch.randn(2, 3, 10),), tester_factory)
59-
60-
def test_conv1d_stride(self, tester_factory: Callable) -> None:
61-
# Test with different stride values
62-
self._test_op(Model(stride=2), (torch.randn(2, 3, 10),), tester_factory)
63-
64-
def test_conv1d_padding(self, tester_factory: Callable) -> None:
65-
# Test with different padding values
66-
self._test_op(Model(padding=1), (torch.randn(2, 3, 10),), tester_factory)
67-
self._test_op(Model(padding=2), (torch.randn(2, 3, 10),), tester_factory)
68-
69-
def test_conv1d_dilation(self, tester_factory: Callable) -> None:
70-
# Test with different dilation values
71-
self._test_op(Model(dilation=2), (torch.randn(2, 3, 10),), tester_factory)
72-
73-
def test_conv1d_groups(self, tester_factory: Callable) -> None:
74-
# Test with groups=3 (in_channels must be divisible by groups)
75-
self._test_op(Model(in_channels=6, out_channels=6, groups=3), (torch.randn(2, 6, 10),), tester_factory)
76-
77-
def test_conv1d_no_bias(self, tester_factory: Callable) -> None:
78-
# Test without bias
79-
self._test_op(Model(bias=False), (torch.randn(2, 3, 10),), tester_factory)
80-
81-
def test_conv1d_padding_modes(self, tester_factory: Callable) -> None:
82-
# Test different padding modes
53+
def test_conv1d_dtype(self, flow: TestFlow, dtype) -> None:
54+
self._test_op(
55+
Model().to(dtype),
56+
((torch.rand(2, 3, 10) * 10).to(dtype),),
57+
flow,
58+
)
59+
60+
def test_conv1d_basic(self, flow: TestFlow) -> None:
61+
self._test_op(
62+
Model(),
63+
(torch.randn(2, 3, 10),),
64+
flow,
65+
)
66+
67+
def test_conv1d_kernel_size(self, flow: TestFlow) -> None:
68+
self._test_op(
69+
Model(kernel_size=1),
70+
(torch.randn(2, 3, 10),),
71+
flow,
72+
)
73+
self._test_op(
74+
Model(kernel_size=5),
75+
(torch.randn(2, 3, 10),),
76+
flow,
77+
)
78+
79+
def test_conv1d_stride(self, flow: TestFlow) -> None:
80+
self._test_op(
81+
Model(stride=2),
82+
(torch.randn(2, 3, 10),),
83+
flow,
84+
)
85+
86+
def test_conv1d_padding(self, flow: TestFlow) -> None:
87+
self._test_op(
88+
Model(padding=1),
89+
(torch.randn(2, 3, 10),),
90+
flow,
91+
)
92+
self._test_op(
93+
Model(padding=2),
94+
(torch.randn(2, 3, 10),),
95+
flow,
96+
)
97+
98+
def test_conv1d_dilation(self, flow: TestFlow) -> None:
99+
self._test_op(
100+
Model(dilation=2),
101+
(torch.randn(2, 3, 10),),
102+
flow,
103+
)
104+
105+
def test_conv1d_groups(self, flow: TestFlow) -> None:
106+
self._test_op(
107+
Model(in_channels=6, out_channels=6, groups=3),
108+
(torch.randn(2, 6, 10),),
109+
flow,
110+
)
111+
112+
def test_conv1d_no_bias(self, flow: TestFlow) -> None:
113+
self._test_op(
114+
Model(bias=False),
115+
(torch.randn(2, 3, 10),),
116+
flow,
117+
)
118+
119+
def test_conv1d_padding_modes(self, flow: TestFlow) -> None:
83120
for mode in ["zeros", "reflect", "replicate", "circular"]:
84-
self._test_op(Model(padding=1, padding_mode=mode), (torch.randn(2, 3, 10),), tester_factory)
85-
86-
def test_conv1d_channels(self, tester_factory: Callable) -> None:
87-
# Test with different channel configurations
88-
self._test_op(Model(in_channels=1, out_channels=1), (torch.randn(2, 1, 10),), tester_factory)
89-
self._test_op(Model(in_channels=5, out_channels=10), (torch.randn(2, 5, 10),), tester_factory)
121+
self._test_op(
122+
Model(padding=1, padding_mode=mode),
123+
(torch.randn(2, 3, 10),),
124+
flow,
125+
)
126+
127+
def test_conv1d_channels(self, flow: TestFlow) -> None:
128+
self._test_op(
129+
Model(in_channels=1, out_channels=1),
130+
(torch.randn(2, 1, 10),),
131+
flow,
132+
)
133+
self._test_op(
134+
Model(in_channels=5, out_channels=10),
135+
(torch.randn(2, 5, 10),),
136+
flow,
137+
)

‎backends/test/suite/operators/test_conv2d.py‎

Lines changed: 123 additions & 59 deletions
Original file line numberDiff line numberDiff line change
@@ -1,26 +1,32 @@
1-
# (c) Meta Platforms, Inc. and affiliates. Confidential and proprietary.
1+
# Copyright (c) Meta Platforms, Inc. and affiliates.
2+
# All rights reserved.
3+
#
4+
# This source code is licensed under the BSD-style license found in the
5+
# LICENSE file in the root directory of this source tree.
26

3-
# pyre-strict
7+
# pyre-unsafe
48

5-
from typing import Callable, Union, Tuple
9+
from typing import Tuple, Union
610

711
import torch
12+
from executorch.backends.test.suite.flow import TestFlow
813

9-
from executorch.backends.test.compliance_suite import (
14+
from executorch.backends.test.suite.operators import (
1015
dtype_test,
1116
operator_test,
1217
OperatorTest,
1318
)
1419

20+
1521
class Model(torch.nn.Module):
1622
def __init__(
1723
self,
1824
in_channels=3,
1925
out_channels=6,
20-
kernel_size=3,
21-
stride=1,
22-
padding=0,
23-
dilation=1,
26+
kernel_size: Union[int, Tuple[int, int]] = 3,
27+
stride: Union[int, Tuple[int, int]] = 1,
28+
padding: Union[int, Tuple[int, int]] = 0,
29+
dilation: Union[int, Tuple[int, int]] = 1,
2430
groups=1,
2531
bias=True,
2632
padding_mode="zeros",
@@ -37,60 +43,118 @@ def __init__(
3743
bias=bias,
3844
padding_mode=padding_mode,
3945
)
40-
46+
4147
def forward(self, x):
4248
return self.conv(x)
4349

50+
4451
@operator_test
45-
class TestConv2d(OperatorTest):
52+
class Conv2d(OperatorTest):
4653
@dtype_test
47-
def test_conv2d_dtype(self, dtype, tester_factory: Callable) -> None:
48-
# Input shape: (batch_size, in_channels, height, width)
49-
self._test_op(Model().to(dtype), ((torch.rand(2, 3, 8, 8) * 10).to(dtype),), tester_factory)
50-
51-
def test_conv2d_basic(self, tester_factory: Callable) -> None:
52-
# Basic test with default parameters
53-
self._test_op(Model(), (torch.randn(2, 3, 8, 8),), tester_factory)
54-
55-
def test_conv2d_kernel_size(self, tester_factory: Callable) -> None:
56-
# Test with different kernel sizes
57-
self._test_op(Model(kernel_size=1), (torch.randn(2, 3, 8, 8),), tester_factory)
58-
self._test_op(Model(kernel_size=5), (torch.randn(2, 3, 8, 8),), tester_factory)
59-
self._test_op(Model(kernel_size=(3, 5)), (torch.randn(2, 3, 8, 8),), tester_factory)
60-
61-
def test_conv2d_stride(self, tester_factory: Callable) -> None:
62-
# Test with different stride values
63-
self._test_op(Model(stride=2), (torch.randn(2, 3, 8, 8),), tester_factory)
64-
self._test_op(Model(stride=(2, 1)), (torch.randn(2, 3, 8, 8),), tester_factory)
65-
66-
def test_conv2d_padding(self, tester_factory: Callable) -> None:
67-
# Test with different padding values
68-
self._test_op(Model(padding=1), (torch.randn(2, 3, 8, 8),), tester_factory)
69-
self._test_op(Model(padding=(1, 2)), (torch.randn(2, 3, 8, 8),), tester_factory)
70-
71-
def test_conv2d_dilation(self, tester_factory: Callable) -> None:
72-
# Test with different dilation values
73-
self._test_op(Model(dilation=2), (torch.randn(2, 3, 8, 8),), tester_factory)
74-
self._test_op(Model(dilation=(2, 1)), (torch.randn(2, 3, 8, 8),), tester_factory)
75-
76-
def test_conv2d_groups(self, tester_factory: Callable) -> None:
77-
# Test with groups=3 (in_channels must be divisible by groups)
78-
self._test_op(Model(in_channels=6, out_channels=6, groups=3), (torch.randn(2, 6, 8, 8),), tester_factory)
79-
80-
def test_conv2d_no_bias(self, tester_factory: Callable) -> None:
81-
# Test without bias
82-
self._test_op(Model(bias=False), (torch.randn(2, 3, 8, 8),), tester_factory)
83-
84-
def test_conv2d_padding_modes(self, tester_factory: Callable) -> None:
85-
# Test different padding modes
54+
def test_conv2d_dtype(self, flow: TestFlow, dtype) -> None:
55+
self._test_op(
56+
Model().to(dtype),
57+
((torch.rand(2, 3, 8, 8) * 10).to(dtype),),
58+
flow,
59+
)
60+
61+
def test_conv2d_basic(self, flow: TestFlow) -> None:
62+
self._test_op(
63+
Model(),
64+
(torch.randn(2, 3, 8, 8),),
65+
flow,
66+
)
67+
68+
def test_conv2d_kernel_size(self, flow: TestFlow) -> None:
69+
self._test_op(
70+
Model(kernel_size=1),
71+
(torch.randn(2, 3, 8, 8),),
72+
flow,
73+
)
74+
self._test_op(
75+
Model(kernel_size=5),
76+
(torch.randn(2, 3, 8, 8),),
77+
flow,
78+
)
79+
self._test_op(
80+
Model(kernel_size=(3, 5)),
81+
(torch.randn(2, 3, 8, 8),),
82+
flow,
83+
)
84+
85+
def test_conv2d_stride(self, flow: TestFlow) -> None:
86+
self._test_op(
87+
Model(stride=2),
88+
(torch.randn(2, 3, 8, 8),),
89+
flow,
90+
)
91+
self._test_op(
92+
Model(stride=(2, 1)),
93+
(torch.randn(2, 3, 8, 8),),
94+
flow,
95+
)
96+
97+
def test_conv2d_padding(self, flow: TestFlow) -> None:
98+
self._test_op(
99+
Model(padding=1),
100+
(torch.randn(2, 3, 8, 8),),
101+
flow,
102+
)
103+
self._test_op(
104+
Model(padding=(1, 2)),
105+
(torch.randn(2, 3, 8, 8),),
106+
flow,
107+
)
108+
109+
def test_conv2d_dilation(self, flow: TestFlow) -> None:
110+
self._test_op(
111+
Model(dilation=2),
112+
(torch.randn(2, 3, 8, 8),),
113+
flow,
114+
)
115+
self._test_op(
116+
Model(dilation=(2, 1)),
117+
(torch.randn(2, 3, 8, 8),),
118+
flow,
119+
)
120+
121+
def test_conv2d_groups(self, flow: TestFlow) -> None:
122+
self._test_op(
123+
Model(in_channels=6, out_channels=6, groups=3),
124+
(torch.randn(2, 6, 8, 8),),
125+
flow,
126+
)
127+
128+
def test_conv2d_no_bias(self, flow: TestFlow) -> None:
129+
self._test_op(
130+
Model(bias=False),
131+
(torch.randn(2, 3, 8, 8),),
132+
flow,
133+
)
134+
135+
def test_conv2d_padding_modes(self, flow: TestFlow) -> None:
86136
for mode in ["zeros", "reflect", "replicate", "circular"]:
87-
self._test_op(Model(padding=1, padding_mode=mode), (torch.randn(2, 3, 8, 8),), tester_factory)
88-
89-
def test_conv2d_channels(self, tester_factory: Callable) -> None:
90-
# Test with different channel configurations
91-
self._test_op(Model(in_channels=1, out_channels=1), (torch.randn(2, 1, 8, 8),), tester_factory)
92-
self._test_op(Model(in_channels=5, out_channels=10), (torch.randn(2, 5, 8, 8),), tester_factory)
93-
94-
def test_conv2d_different_spatial_dims(self, tester_factory: Callable) -> None:
95-
# Test with different height and width
96-
self._test_op(Model(), (torch.randn(2, 3, 10, 8),), tester_factory)
137+
self._test_op(
138+
Model(padding=1, padding_mode=mode),
139+
(torch.randn(2, 3, 8, 8),),
140+
flow,
141+
)
142+
143+
def test_conv2d_channels(self, flow: TestFlow) -> None:
144+
self._test_op(
145+
Model(in_channels=1, out_channels=1),
146+
(torch.randn(2, 1, 8, 8),),
147+
flow,
148+
)
149+
self._test_op(
150+
Model(in_channels=5, out_channels=10),
151+
(torch.randn(2, 5, 8, 8),),
152+
flow,
153+
)
154+
155+
def test_conv2d_different_spatial_dims(self, flow: TestFlow) -> None:
156+
self._test_op(
157+
Model(),
158+
(torch.randn(2, 3, 10, 8),),
159+
flow,
160+
)

0 commit comments

Comments
 (0)