[ExecuTorch][WebGPU] squeeze_copy + unsqueeze_copy test suites (cases.py op-test framework)

Pull Request resolved: #20393

Registers `aten.squeeze_copy.dims` and `aten.unsqueeze_copy.default` in the `cases.py` op-test framework: a `_squeeze_suite` of 3 configs (squeeze leading/middle/multiple size-1 dims) and a `_unsqueeze_suite` of 3 configs (insert dim at front/middle/last) that `generate_op_tests` exports via `VulkanPartitioner` and compares to a torch golden on Dawn. Also adds `test/ops/squeeze/test_squeeze.py` (`SqueezeModule` + `CONFIGS` + `_op_delegated` smoke test), `test/ops/unsqueeze/test_unsqueeze.py` (`UnsqueezeModule` + `CONFIGS` + `_op_delegated` smoke test), and the two partitioner-allowlist entries in `tester.py`.
ghstack-source-id: 397026525
@exported-using-ghexport

Differential Revision: [D108793152](https://our.internmc.facebook.com/intern/diff/D108793152/)

Author: JulianCloudNTH

backends/webgpu/test/op_tests/cases.py

@@ -49,6 +49,16 @@
     N as _SIGMOID_N,
     SigmoidModule,
 )
+
+from executorch.backends.webgpu.test.ops.test_squeeze import (
+    CONFIGS as _SQUEEZE_CONFIGS,
+    SqueezeModule,
+)
+
+from executorch.backends.webgpu.test.ops.test_unsqueeze import (
+    CONFIGS as _UNSQUEEZE_CONFIGS,
+    UnsqueezeModule,
+)
 from executorch.backends.webgpu.test.ops.test_view_copy import (
     CONFIGS as _VIEW_CONFIGS,
     ViewModule,
@@ -184,3 +194,29 @@ def _sigmoid_suite() -> WebGPUTestSuite:
         atol=1e-4,
         rtol=1e-4,
     )
+
+
+@register_op_test("squeeze")
+def _squeeze_suite() -> WebGPUTestSuite:
+    # CONFIGS: name -> (shape, dim) where dim is an int or a tuple.
+    return WebGPUTestSuite(
+        module_factory=lambda dim: SqueezeModule(dim),
+        cases=[
+            Case(name=n, construct={"dim": dim}, inputs=(shape,))
+            for n, (shape, dim) in _SQUEEZE_CONFIGS.items()
+        ],
+        golden_dtype="float32",  # reshape copies values; fp64 bit-identical
+    )
+
+
+@register_op_test("unsqueeze")
+def _unsqueeze_suite() -> WebGPUTestSuite:
+    # CONFIGS: name -> (shape, dim).
+    return WebGPUTestSuite(
+        module_factory=lambda dim: UnsqueezeModule(dim),
+        cases=[
+            Case(name=n, construct={"dim": dim}, inputs=(shape,))
+            for n, (shape, dim) in _UNSQUEEZE_CONFIGS.items()
+        ],
+        golden_dtype="float32",  # reshape copies values; fp64 bit-identical
+    )

backends/webgpu/test/ops/test_squeeze.py

@@ -0,0 +1,75 @@
+# Copyright (c) Meta Platforms, Inc. and affiliates.
+# All rights reserved.
+#
+# This source code is licensed under the BSD-style license found in the
+# LICENSE file in the root directory of this source tree.
+
+"""`aten.squeeze_copy.dims` module + configs for the WebGPU op-test framework.
+
+`SqueezeModule` + `CONFIGS` are imported by `cases.py` to drive the declarative
+op-test suite. `SqueezeTest` is the export-delegation smoke
+test.
+"""
+
+import unittest
+
+import torch
+
+from executorch.backends.vulkan.partitioner.vulkan_partitioner import VulkanPartitioner
+from executorch.exir import to_edge_transform_and_lower
+
+# name -> (input_shape, squeeze_dim)
+CONFIGS = {
+    "dim0": ((1, 3, 4), 0),
+    "mid": ((2, 1, 4), 1),
+    "multi": ((1, 3, 1, 4), (0, 2)),
+}
+
+
+class SqueezeModule(torch.nn.Module):
+    def __init__(self, dim):
+        super().__init__()
+        self.dim = dim
+
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        return torch.squeeze(x, self.dim)
+
+
+def _det_input(shape):
+    g = torch.Generator().manual_seed(0)
+    return torch.randn(*shape, generator=g, dtype=torch.float32)
+
+
+def _lower(dim, x: torch.Tensor):
+    ep = torch.export.export(SqueezeModule(dim).eval(), (x,))
+    return to_edge_transform_and_lower(ep, partitioner=[VulkanPartitioner()])
+
+
+def _delegated(et) -> bool:
+    return any(
+        d.id == "VulkanBackend"
+        for plan in et.executorch_program.execution_plan
+        for d in plan.delegates
+    )
+
+
+def _op_delegated(edge, op_substr: str) -> bool:
+    # op must be absorbed into the delegate, not left as a CPU-fallback node.
+    gm = edge.exported_program().graph_module
+    return all(op_substr not in str(getattr(n, "target", "")) for n in gm.graph.nodes)
+
+
+class SqueezeTest(unittest.TestCase):
+    def test_export_delegates(self) -> None:
+        for name, (shape, dim) in CONFIGS.items():
+            with self.subTest(name=name):
+                edge = _lower(dim, _det_input(shape))
+                et = edge.to_executorch()
+                self.assertTrue(
+                    _delegated(et),
+                    f"Expected a VulkanBackend delegate (squeeze {name})",
+                )
+                self.assertTrue(
+                    _op_delegated(edge, "squeeze_copy"),
+                    f"squeeze_copy not delegated (fell back to CPU) for {name}",
+                )

backends/webgpu/test/ops/test_unsqueeze.py

@@ -0,0 +1,75 @@
+# Copyright (c) Meta Platforms, Inc. and affiliates.
+# All rights reserved.
+#
+# This source code is licensed under the BSD-style license found in the
+# LICENSE file in the root directory of this source tree.
+
+"""`aten.unsqueeze_copy.default` module + configs for the WebGPU op-test framework.
+
+`UnsqueezeModule` + `CONFIGS` are imported by `cases.py` to drive the declarative
+op-test suite. `UnsqueezeTest` is the export-delegation smoke
+test.
+"""
+
+import unittest
+
+import torch
+
+from executorch.backends.vulkan.partitioner.vulkan_partitioner import VulkanPartitioner
+from executorch.exir import to_edge_transform_and_lower
+
+# name -> (input_shape, unsqueeze_dim)
+CONFIGS = {
+    "front": ((3, 4), 0),
+    "mid": ((2, 4), 1),
+    "last": ((3, 4), 2),
+}
+
+
+class UnsqueezeModule(torch.nn.Module):
+    def __init__(self, dim):
+        super().__init__()
+        self.dim = dim
+
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        return torch.unsqueeze(x, self.dim)
+
+
+def _det_input(shape):
+    g = torch.Generator().manual_seed(0)
+    return torch.randn(*shape, generator=g, dtype=torch.float32)
+
+
+def _lower(dim, x: torch.Tensor):
+    ep = torch.export.export(UnsqueezeModule(dim).eval(), (x,))
+    return to_edge_transform_and_lower(ep, partitioner=[VulkanPartitioner()])
+
+
+def _delegated(et) -> bool:
+    return any(
+        d.id == "VulkanBackend"
+        for plan in et.executorch_program.execution_plan
+        for d in plan.delegates
+    )
+
+
+def _op_delegated(edge, op_substr: str) -> bool:
+    # op must be absorbed into the delegate, not left as a top-level CPU-fallback node.
+    gm = edge.exported_program().graph_module
+    return all(op_substr not in str(getattr(n, "target", "")) for n in gm.graph.nodes)
+
+
+class UnsqueezeTest(unittest.TestCase):
+    def test_export_delegates(self) -> None:
+        for name, (shape, dim) in CONFIGS.items():
+            with self.subTest(name=name):
+                edge = _lower(dim, _det_input(shape))
+                et = edge.to_executorch()
+                self.assertTrue(
+                    _delegated(et),
+                    f"Expected a VulkanBackend delegate (unsqueeze {name})",
+                )
+                self.assertTrue(
+                    _op_delegated(edge, "unsqueeze_copy"),
+                    f"unsqueeze_copy not delegated (fell back to CPU) for {name}",
+                )

backends/webgpu/test/tester.py

@@ -25,6 +25,8 @@
     exir_ops.edge.aten.view_copy.default,
     exir_ops.edge.aten.select_copy.int,
     exir_ops.edge.aten.sigmoid.default,
+    exir_ops.edge.aten.squeeze_copy.dims,
+    exir_ops.edge.aten.unsqueeze_copy.default,
 ]