From bc7349bac20566c2f75a1b1c88e2c872fcc3da18 Mon Sep 17 00:00:00 2001
From: Gregory James Comer <gjcomer@meta.com>
Date: Thu, 4 Sep 2025 17:30:16 -0700
Subject: [PATCH 1/9] Update

[ghstack-poisoned]
---
 backends/test/suite/models/__init__.py | 1 -
 1 file changed, 1 deletion(-)

diff --git a/backends/test/suite/models/__init__.py b/backends/test/suite/models/__init__.py
index 76b2d2966f6..65b546b0eb5 100644
--- a/backends/test/suite/models/__init__.py
+++ b/backends/test/suite/models/__init__.py
@@ -22,7 +22,6 @@
 DTYPES: list[torch.dtype] = [
     torch.float16,
     torch.float32,
-    torch.float64,
 ]
 
 

From 223d5d654671810d55dae40cde82503d836cb6fb Mon Sep 17 00:00:00 2001
From: Gregory Comer <gregoryjcomer@gmail.com>
Date: Thu, 4 Sep 2025 22:45:06 -0700
Subject: [PATCH 2/9] Update

[ghstack-poisoned]
---
 backends/test/suite/operators/test_elu.py         | 2 ++
 backends/test/suite/operators/test_hardsigmoid.py | 2 ++
 backends/test/suite/operators/test_hardswish.py   | 2 ++
 backends/test/suite/operators/test_hardtanh.py    | 2 ++
 backends/test/suite/operators/test_leaky_relu.py  | 2 ++
 backends/test/suite/operators/test_relu.py        | 2 ++
 backends/test/suite/operators/test_silu.py        | 2 ++
 backends/test/suite/operators/test_threshold.py   | 2 ++
 8 files changed, 16 insertions(+)

diff --git a/backends/test/suite/operators/test_elu.py b/backends/test/suite/operators/test_elu.py
index f768a426954..1ecc6c5b1ab 100644
--- a/backends/test/suite/operators/test_elu.py
+++ b/backends/test/suite/operators/test_elu.py
@@ -8,6 +8,7 @@
 
 
 import torch
+import unittest
 from executorch.backends.test.suite.flow import TestFlow
 
 from executorch.backends.test.suite.operators import (
@@ -42,5 +43,6 @@ def test_elu_f32_multi_dim(self, flow: TestFlow) -> None:
     def test_elu_f32_alpha(self, flow: TestFlow) -> None:
         self._test_op(Model(alpha=0.5), (torch.randn(3, 4, 5),), flow)
 
+    @unittest.skip("In place activations aren't properly defunctionalized yet.")
     def test_elu_f32_inplace(self, flow: TestFlow) -> None:
         self._test_op(Model(inplace=True), (torch.randn(3, 4, 5),), flow)
diff --git a/backends/test/suite/operators/test_hardsigmoid.py b/backends/test/suite/operators/test_hardsigmoid.py
index 238b18b1e0d..8eff9d75f6a 100644
--- a/backends/test/suite/operators/test_hardsigmoid.py
+++ b/backends/test/suite/operators/test_hardsigmoid.py
@@ -8,6 +8,7 @@
 
 
 import torch
+import unittest
 from executorch.backends.test.suite.flow import TestFlow
 
 from executorch.backends.test.suite.operators import (
@@ -38,6 +39,7 @@ def test_hardsigmoid_f32_single_dim(self, flow: TestFlow) -> None:
     def test_hardsigmoid_f32_multi_dim(self, flow: TestFlow) -> None:
         self._test_op(Model(), (torch.randn(2, 3, 4, 5),), flow)
 
+    @unittest.skip("In place activations aren't properly defunctionalized yet.")
     def test_hardsigmoid_f32_inplace(self, flow: TestFlow) -> None:
         self._test_op(Model(inplace=True), (torch.randn(3, 4, 5),), flow)
 
diff --git a/backends/test/suite/operators/test_hardswish.py b/backends/test/suite/operators/test_hardswish.py
index 66902791c33..09b959be1ef 100644
--- a/backends/test/suite/operators/test_hardswish.py
+++ b/backends/test/suite/operators/test_hardswish.py
@@ -8,6 +8,7 @@
 
 
 import torch
+import unittest
 from executorch.backends.test.suite.flow import TestFlow
 
 from executorch.backends.test.suite.operators import (
@@ -38,6 +39,7 @@ def test_hardswish_f32_single_dim(self, flow: TestFlow) -> None:
     def test_hardswish_f32_multi_dim(self, flow: TestFlow) -> None:
         self._test_op(Model(), (torch.randn(2, 3, 4, 5),), flow)
 
+    @unittest.skip("In place activations aren't properly defunctionalized yet.")
     def test_hardswish_f32_inplace(self, flow: TestFlow) -> None:
         self._test_op(Model(inplace=True), (torch.randn(3, 4, 5),), flow)
 
diff --git a/backends/test/suite/operators/test_hardtanh.py b/backends/test/suite/operators/test_hardtanh.py
index 2fcd1dbf563..09d25550685 100644
--- a/backends/test/suite/operators/test_hardtanh.py
+++ b/backends/test/suite/operators/test_hardtanh.py
@@ -8,6 +8,7 @@
 
 
 import torch
+import unittest
 from executorch.backends.test.suite.flow import TestFlow
 
 from executorch.backends.test.suite.operators import (
@@ -45,6 +46,7 @@ def test_hardtanh_f32_multi_dim(self, flow: TestFlow) -> None:
     def test_hardtanh_f32_custom_range(self, flow: TestFlow) -> None:
         self._test_op(Model(min_val=-2.0, max_val=2.0), (torch.randn(3, 4, 5),), flow)
 
+    @unittest.skip("In place activations aren't properly defunctionalized yet.")
     def test_hardtanh_f32_inplace(self, flow: TestFlow) -> None:
         self._test_op(Model(inplace=True), (torch.randn(3, 4, 5),), flow)
 
diff --git a/backends/test/suite/operators/test_leaky_relu.py b/backends/test/suite/operators/test_leaky_relu.py
index 983da47bba3..def8f8f8492 100644
--- a/backends/test/suite/operators/test_leaky_relu.py
+++ b/backends/test/suite/operators/test_leaky_relu.py
@@ -8,6 +8,7 @@
 
 
 import torch
+import unittest
 from executorch.backends.test.suite.flow import TestFlow
 
 from executorch.backends.test.suite.operators import (
@@ -44,6 +45,7 @@ def test_leaky_relu_f32_multi_dim(self, flow: TestFlow) -> None:
     def test_leaky_relu_f32_custom_slope(self, flow: TestFlow) -> None:
         self._test_op(Model(negative_slope=0.1), (torch.randn(3, 4, 5),), flow)
 
+    @unittest.skip("In place activations aren't properly defunctionalized yet.")
     def test_leaky_relu_f32_inplace(self, flow: TestFlow) -> None:
         self._test_op(Model(inplace=True), (torch.randn(3, 4, 5),), flow)
 
diff --git a/backends/test/suite/operators/test_relu.py b/backends/test/suite/operators/test_relu.py
index c9f416f090f..fc03bfedfc3 100644
--- a/backends/test/suite/operators/test_relu.py
+++ b/backends/test/suite/operators/test_relu.py
@@ -8,6 +8,7 @@
 
 
 import torch
+import unittest
 from executorch.backends.test.suite.flow import TestFlow
 
 from executorch.backends.test.suite.operators import (
@@ -38,5 +39,6 @@ def test_relu_f32_single_dim(self, flow: TestFlow) -> None:
     def test_relu_f32_multi_dim(self, flow: TestFlow) -> None:
         self._test_op(Model(), (torch.randn(2, 3, 4, 5),), flow)
 
+    @unittest.skip("In place activations aren't properly defunctionalized yet.")
     def test_relu_f32_inplace(self, flow: TestFlow) -> None:
         self._test_op(Model(inplace=True), (torch.randn(3, 4, 5),), flow)
diff --git a/backends/test/suite/operators/test_silu.py b/backends/test/suite/operators/test_silu.py
index 69b6576734f..46b7c55489e 100644
--- a/backends/test/suite/operators/test_silu.py
+++ b/backends/test/suite/operators/test_silu.py
@@ -8,6 +8,7 @@
 
 
 import torch
+import unittest
 from executorch.backends.test.suite.flow import TestFlow
 
 from executorch.backends.test.suite.operators import (
@@ -38,6 +39,7 @@ def test_silu_f32_single_dim(self, flow: TestFlow) -> None:
     def test_silu_f32_multi_dim(self, flow: TestFlow) -> None:
         self._test_op(Model(), (torch.randn(2, 3, 4, 5),), flow)
 
+    @unittest.skip("In place activations aren't properly defunctionalized yet.")
     def test_silu_f32_inplace(self, flow: TestFlow) -> None:
         self._test_op(Model(inplace=True), (torch.randn(3, 4, 5),), flow)
 
diff --git a/backends/test/suite/operators/test_threshold.py b/backends/test/suite/operators/test_threshold.py
index 42b6fb801e5..85cf65ab8f2 100644
--- a/backends/test/suite/operators/test_threshold.py
+++ b/backends/test/suite/operators/test_threshold.py
@@ -8,6 +8,7 @@
 
 
 import torch
+import unittest
 from executorch.backends.test.suite.flow import TestFlow
 
 from executorch.backends.test.suite.operators import (
@@ -51,6 +52,7 @@ def test_threshold_f32_custom_value(self, flow: TestFlow) -> None:
     def test_threshold_f32_custom_threshold_value(self, flow: TestFlow) -> None:
         self._test_op(Model(threshold=0.5, value=1.0), (torch.randn(3, 4, 5),), flow)
 
+    @unittest.skip("In place activations aren't properly defunctionalized yet.")
     def test_threshold_f32_inplace(self, flow: TestFlow) -> None:
         self._test_op(Model(inplace=True), (torch.randn(3, 4, 5),), flow)
 

From ed8a85995a342b82e6174e9db2b9a08f4a430e64 Mon Sep 17 00:00:00 2001
From: Gregory James Comer <gjcomer@meta.com>
Date: Thu, 4 Sep 2025 23:05:28 -0700
Subject: [PATCH 3/9] Update

[ghstack-poisoned]
---
 backends/test/suite/operators/test_abs.py          | 2 ++
 backends/test/suite/operators/test_amax.py         | 2 ++
 backends/test/suite/operators/test_amin.py         | 2 ++
 backends/test/suite/operators/test_argmax.py       | 2 ++
 backends/test/suite/operators/test_argmin.py       | 2 ++
 backends/test/suite/operators/test_ceil.py         | 2 ++
 backends/test/suite/operators/test_clamp.py        | 2 ++
 backends/test/suite/operators/test_exp.py          | 2 ++
 backends/test/suite/operators/test_floor.py        | 2 ++
 backends/test/suite/operators/test_floor_divide.py | 2 ++
 backends/test/suite/operators/test_log.py          | 2 ++
 backends/test/suite/operators/test_log10.py        | 2 ++
 backends/test/suite/operators/test_log1p.py        | 2 ++
 backends/test/suite/operators/test_log2.py         | 2 ++
 backends/test/suite/operators/test_mean.py         | 2 ++
 backends/test/suite/operators/test_median.py       | 2 ++
 backends/test/suite/operators/test_neg.py          | 2 ++
 backends/test/suite/operators/test_pow.py          | 2 ++
 backends/test/suite/operators/test_round.py        | 3 +++
 backends/test/suite/operators/test_rsqrt.py        | 2 ++
 backends/test/suite/operators/test_sqrt.py         | 2 ++
 backends/test/suite/operators/test_square.py       | 2 ++
 backends/test/suite/operators/test_trunc.py        | 2 ++
 23 files changed, 47 insertions(+)

diff --git a/backends/test/suite/operators/test_abs.py b/backends/test/suite/operators/test_abs.py
index fdfc6be671e..5ab4eb92173 100644
--- a/backends/test/suite/operators/test_abs.py
+++ b/backends/test/suite/operators/test_abs.py
@@ -8,6 +8,7 @@
 
 
 import torch
+import unittest
 from executorch.backends.test.suite.flow import TestFlow
 
 from executorch.backends.test.suite.operators import (
@@ -45,6 +46,7 @@ def test_abs_shapes(self, flow: TestFlow) -> None:
         # 3D tensor
         self._test_op(AbsModel(), (torch.randn(3, 4, 5),), flow)
 
+    @unittest.skip("NaN and Inf are not enforced for backends.")
     def test_abs_edge_cases(self, flow: TestFlow) -> None:
         # Test edge cases
 
diff --git a/backends/test/suite/operators/test_amax.py b/backends/test/suite/operators/test_amax.py
index 0c9a8c06f0d..ab2b1a310ad 100644
--- a/backends/test/suite/operators/test_amax.py
+++ b/backends/test/suite/operators/test_amax.py
@@ -9,6 +9,7 @@
 from typing import List, Optional, Tuple, Union
 
 import torch
+import unittest
 from executorch.backends.test.suite.flow import TestFlow
 
 from executorch.backends.test.suite.operators import (
@@ -201,6 +202,7 @@ def test_amax_shapes(self, flow: TestFlow) -> None:
             flow,
         )
 
+    @unittest.skip("NaN and Inf are not enforced for backends.")
     def test_amax_edge_cases(self, flow: TestFlow) -> None:
         x = torch.tensor([[1.0, float("inf"), 3.0], [4.0, 5.0, float("inf")]])
         self._test_op(
diff --git a/backends/test/suite/operators/test_amin.py b/backends/test/suite/operators/test_amin.py
index f4b88b1dade..6787567b2dc 100644
--- a/backends/test/suite/operators/test_amin.py
+++ b/backends/test/suite/operators/test_amin.py
@@ -9,6 +9,7 @@
 from typing import List, Optional, Tuple, Union
 
 import torch
+import unittest
 from executorch.backends.test.suite.flow import TestFlow
 
 from executorch.backends.test.suite.operators import (
@@ -203,6 +204,7 @@ def test_amin_shapes(self, flow: TestFlow) -> None:
             flow,
         )
 
+    @unittest.skip("NaN and Inf are not enforced for backends.")
     def test_amin_edge_cases(self, flow: TestFlow) -> None:
         x = torch.tensor([[1.0, float("-inf"), 3.0], [4.0, 5.0, float("-inf")]])
         self._test_op(
diff --git a/backends/test/suite/operators/test_argmax.py b/backends/test/suite/operators/test_argmax.py
index dc8b57fc214..75d34072e77 100644
--- a/backends/test/suite/operators/test_argmax.py
+++ b/backends/test/suite/operators/test_argmax.py
@@ -9,6 +9,7 @@
 from typing import Optional
 
 import torch
+import unittest
 from executorch.backends.test.suite.flow import TestFlow
 
 from executorch.backends.test.suite.operators import (
@@ -143,6 +144,7 @@ def test_argmax_shapes(self, flow: TestFlow) -> None:
             flow,
         )
 
+    @unittest.skip("NaN and Inf are not enforced for backends.")
     def test_argmax_edge_cases(self, flow: TestFlow) -> None:
         x = torch.tensor([[1.0, float("inf"), 3.0], [4.0, 5.0, float("inf")]])
         self._test_op(
diff --git a/backends/test/suite/operators/test_argmin.py b/backends/test/suite/operators/test_argmin.py
index d7a24e24f5a..e85e918ca0d 100644
--- a/backends/test/suite/operators/test_argmin.py
+++ b/backends/test/suite/operators/test_argmin.py
@@ -9,6 +9,7 @@
 from typing import Optional
 
 import torch
+import unittest
 from executorch.backends.test.suite.flow import TestFlow
 
 from executorch.backends.test.suite.operators import (
@@ -143,6 +144,7 @@ def test_argmin_shapes(self, flow: TestFlow) -> None:
             flow,
         )
 
+    @unittest.skip("NaN and Inf are not enforced for backends.")
     def test_argmin_edge_cases(self, flow: TestFlow) -> None:
         x = torch.tensor([[1.0, float("-inf"), 3.0], [4.0, 5.0, float("-inf")]])
         self._test_op(
diff --git a/backends/test/suite/operators/test_ceil.py b/backends/test/suite/operators/test_ceil.py
index 198c9e9fe16..6ca82afce2d 100644
--- a/backends/test/suite/operators/test_ceil.py
+++ b/backends/test/suite/operators/test_ceil.py
@@ -8,6 +8,7 @@
 
 
 import torch
+import unittest
 from executorch.backends.test.suite.flow import TestFlow
 
 from executorch.backends.test.suite.operators import (
@@ -45,6 +46,7 @@ def test_ceil_shapes(self, flow: TestFlow) -> None:
         # 3D tensor
         self._test_op(CeilModel(), (torch.randn(3, 4, 5),), flow)
 
+    @unittest.skip("NaN and Inf are not enforced for backends.")
     def test_ceil_edge_cases(self, flow: TestFlow) -> None:
         # Test edge cases
 
diff --git a/backends/test/suite/operators/test_clamp.py b/backends/test/suite/operators/test_clamp.py
index 67c61c67caa..8193e62c022 100644
--- a/backends/test/suite/operators/test_clamp.py
+++ b/backends/test/suite/operators/test_clamp.py
@@ -8,6 +8,7 @@
 
 
 import torch
+import unittest
 from executorch.backends.test.suite.flow import TestFlow
 
 from executorch.backends.test.suite.operators import (
@@ -56,6 +57,7 @@ def test_clamp_shapes(self, flow: TestFlow) -> None:
         # 3D tensor
         self._test_op(model, (torch.randn(3, 4, 5),), flow)
 
+    @unittest.skip("NaN and Inf are not enforced for backends.")
     def test_clamp_edge_cases(self, flow: TestFlow) -> None:
         # Test edge cases
 
diff --git a/backends/test/suite/operators/test_exp.py b/backends/test/suite/operators/test_exp.py
index bdae5c6a5e6..0322ee46eaf 100644
--- a/backends/test/suite/operators/test_exp.py
+++ b/backends/test/suite/operators/test_exp.py
@@ -8,6 +8,7 @@
 
 
 import torch
+import unittest
 from executorch.backends.test.suite.flow import TestFlow
 
 from executorch.backends.test.suite.operators import (
@@ -46,6 +47,7 @@ def test_exp_shapes(self, flow: TestFlow) -> None:
         # 3D tensor
         self._test_op(ExpModel(), (torch.randn(3, 4, 5),), flow)
 
+    @unittest.skip("NaN and Inf are not enforced for backends.")
     def test_exp_edge_cases(self, flow: TestFlow) -> None:
         # Test edge cases
 
diff --git a/backends/test/suite/operators/test_floor.py b/backends/test/suite/operators/test_floor.py
index fcc834afa16..c02c5995123 100644
--- a/backends/test/suite/operators/test_floor.py
+++ b/backends/test/suite/operators/test_floor.py
@@ -8,6 +8,7 @@
 
 
 import torch
+import unittest
 from executorch.backends.test.suite.flow import TestFlow
 
 from executorch.backends.test.suite.operators import (
@@ -42,6 +43,7 @@ def test_floor_shapes(self, flow: TestFlow) -> None:
         # 3D tensor
         self._test_op(FloorModel(), (torch.randn(3, 4, 5),), flow)
 
+    @unittest.skip("NaN and Inf are not enforced for backends.")
     def test_floor_edge_cases(self, flow: TestFlow) -> None:
         # Test edge cases
 
diff --git a/backends/test/suite/operators/test_floor_divide.py b/backends/test/suite/operators/test_floor_divide.py
index 87104af11dc..8afb2bb72d0 100644
--- a/backends/test/suite/operators/test_floor_divide.py
+++ b/backends/test/suite/operators/test_floor_divide.py
@@ -7,6 +7,7 @@
 # pyre-unsafe
 
 import torch
+import unittest
 from executorch.backends.test.suite.flow import TestFlow
 
 from executorch.backends.test.suite.operators import (
@@ -178,6 +179,7 @@ def test_floor_divide_values(self, flow: TestFlow) -> None:
         y = torch.tensor([-2.0]).expand_as(x).clone()
         self._test_op(model, (x, y), flow, generate_random_test_inputs=False)
 
+    @unittest.skip("NaN and Inf are not enforced for backends.")
     def test_floor_divide_edge_cases(self, flow: TestFlow) -> None:
         # Test edge cases
         model = FloorDivideModel()
diff --git a/backends/test/suite/operators/test_log.py b/backends/test/suite/operators/test_log.py
index 96ba8da1292..783cae4bf37 100644
--- a/backends/test/suite/operators/test_log.py
+++ b/backends/test/suite/operators/test_log.py
@@ -8,6 +8,7 @@
 
 
 import torch
+import unittest
 from executorch.backends.test.suite.flow import TestFlow
 
 from executorch.backends.test.suite.operators import (
@@ -46,6 +47,7 @@ def test_log_shapes(self, flow: TestFlow) -> None:
         # 3D tensor
         self._test_op(LogModel(), (torch.rand(3, 4, 5) + 0.01,), flow)
 
+    @unittest.skip("NaN and Inf are not enforced for backends.")
     def test_log_edge_cases(self, flow: TestFlow) -> None:
         # Test edge cases
         # Tensor with infinity
diff --git a/backends/test/suite/operators/test_log10.py b/backends/test/suite/operators/test_log10.py
index 7d0e2e111d6..fcbbb7d69a9 100644
--- a/backends/test/suite/operators/test_log10.py
+++ b/backends/test/suite/operators/test_log10.py
@@ -8,6 +8,7 @@
 
 
 import torch
+import unittest
 from executorch.backends.test.suite.flow import TestFlow
 
 from executorch.backends.test.suite.operators import (
@@ -46,6 +47,7 @@ def test_log10_shapes(self, flow: TestFlow) -> None:
         # 3D tensor
         self._test_op(Log10Model(), (torch.rand(3, 4, 5) + 0.01,), flow)
 
+    @unittest.skip("NaN and Inf are not enforced for backends.")
     def test_log10_edge_cases(self, flow: TestFlow) -> None:
         # Test edge cases
         # Tensor with infinity
diff --git a/backends/test/suite/operators/test_log1p.py b/backends/test/suite/operators/test_log1p.py
index 383e3116b32..2a00202097a 100644
--- a/backends/test/suite/operators/test_log1p.py
+++ b/backends/test/suite/operators/test_log1p.py
@@ -8,6 +8,7 @@
 
 
 import torch
+import unittest
 from executorch.backends.test.suite.flow import TestFlow
 
 from executorch.backends.test.suite.operators import (
@@ -46,6 +47,7 @@ def test_log1p_shapes(self, flow: TestFlow) -> None:
         # 3D tensor
         self._test_op(Log1pModel(), (torch.rand(3, 4, 5) * 2 - 0.5,), flow)
 
+    @unittest.skip("NaN and Inf are not enforced for backends.")
     def test_log1p_edge_cases(self, flow: TestFlow) -> None:
         # Test edge cases
         # Tensor with infinity
diff --git a/backends/test/suite/operators/test_log2.py b/backends/test/suite/operators/test_log2.py
index ddcafaf08d2..36acee8edf2 100644
--- a/backends/test/suite/operators/test_log2.py
+++ b/backends/test/suite/operators/test_log2.py
@@ -8,6 +8,7 @@
 
 
 import torch
+import unittest
 from executorch.backends.test.suite.flow import TestFlow
 
 from executorch.backends.test.suite.operators import (
@@ -46,6 +47,7 @@ def test_log2_shapes(self, flow: TestFlow) -> None:
         # 3D tensor
         self._test_op(Log2Model(), (torch.rand(3, 4, 5) + 0.01,), flow)
 
+    @unittest.skip("NaN and Inf are not enforced for backends.")
     def test_log2_edge_cases(self, flow: TestFlow) -> None:
         # Test edge cases
         # Tensor with infinity
diff --git a/backends/test/suite/operators/test_mean.py b/backends/test/suite/operators/test_mean.py
index 746a4b16d9f..b53144ed18a 100644
--- a/backends/test/suite/operators/test_mean.py
+++ b/backends/test/suite/operators/test_mean.py
@@ -9,6 +9,7 @@
 from typing import List, Optional, Tuple, Union
 
 import torch
+import unittest
 from executorch.backends.test.suite.flow import TestFlow
 
 from executorch.backends.test.suite.operators import (
@@ -229,6 +230,7 @@ def test_mean_shapes(self, flow: TestFlow) -> None:
             flow,
         )
 
+    @unittest.skip("NaN and Inf are not enforced for backends.")
     def test_mean_edge_cases(self, flow: TestFlow) -> None:
         x = torch.tensor([[1.0, float("inf"), 3.0], [4.0, 5.0, float("inf")]])
         self._test_op(
diff --git a/backends/test/suite/operators/test_median.py b/backends/test/suite/operators/test_median.py
index 93823b812ca..89d1fdb0069 100644
--- a/backends/test/suite/operators/test_median.py
+++ b/backends/test/suite/operators/test_median.py
@@ -9,6 +9,7 @@
 from typing import Optional
 
 import torch
+import unittest
 from executorch.backends.test.suite.flow import TestFlow
 
 from executorch.backends.test.suite.operators import (
@@ -167,6 +168,7 @@ def test_median_shapes(self, flow: TestFlow) -> None:
         # 5D tensor
         self._test_op(MedianValueOnlyModel(), (torch.randn(2, 2, 3, 4, 5),), flow)
 
+    @unittest.skip("NaN and Inf are not enforced for backends.")
     def test_median_edge_cases(self, flow: TestFlow) -> None:
         # Tensor with NaN (NaN should be propagated)
         x = torch.tensor([[1.0, float("nan"), 3.0], [4.0, 5.0, float("nan")]])
diff --git a/backends/test/suite/operators/test_neg.py b/backends/test/suite/operators/test_neg.py
index 35c9d851817..a2b4cf942f8 100644
--- a/backends/test/suite/operators/test_neg.py
+++ b/backends/test/suite/operators/test_neg.py
@@ -7,6 +7,7 @@
 # pyre-unsafe
 
 import torch
+import unittest
 from executorch.backends.test.suite.flow import TestFlow
 
 from executorch.backends.test.suite.operators import (
@@ -55,6 +56,7 @@ def test_neg_shapes(self, flow: TestFlow) -> None:
             NegModel(), (torch.randn(3, 4, 5),), flow, generate_random_test_inputs=False
         )
 
+    @unittest.skip("NaN and Inf are not enforced for backends.")
     def test_neg_edge_cases(self, flow: TestFlow) -> None:
         # Test edge cases
 
diff --git a/backends/test/suite/operators/test_pow.py b/backends/test/suite/operators/test_pow.py
index 334038d73d3..b613683372e 100644
--- a/backends/test/suite/operators/test_pow.py
+++ b/backends/test/suite/operators/test_pow.py
@@ -7,6 +7,7 @@
 # pyre-unsafe
 
 import torch
+import unittest
 from executorch.backends.test.suite.flow import TestFlow
 
 from executorch.backends.test.suite.operators import (
@@ -127,6 +128,7 @@ def test_pow_shapes(self, flow: TestFlow) -> None:
             model, (torch.rand(3, 4, 5) + 0.1,), flow, generate_random_test_inputs=False
         )
 
+    @unittest.skip("NaN and Inf are not enforced for backends.")
     def test_pow_edge_cases(self, flow: TestFlow) -> None:
         # Test edge cases
 
diff --git a/backends/test/suite/operators/test_round.py b/backends/test/suite/operators/test_round.py
index ca8e6368d48..1794da838f9 100644
--- a/backends/test/suite/operators/test_round.py
+++ b/backends/test/suite/operators/test_round.py
@@ -7,6 +7,7 @@
 # pyre-unsafe
 
 import torch
+import unittest
 from executorch.backends.test.suite.flow import TestFlow
 
 from executorch.backends.test.suite.operators import (
@@ -52,6 +53,7 @@ def test_round_values(self, flow: TestFlow) -> None:
         x = torch.arange(-5, 5, 0.5)  # [-5.0, -4.5, -4.0, ..., 4.0, 4.5]
         self._test_op(RoundModel(), (x,), flow, generate_random_test_inputs=False)
 
+    @unittest.skip("NaN and Inf are not enforced for backends.")
     def test_round_edge_cases(self, flow: TestFlow) -> None:
         # Test edge cases
 
@@ -98,6 +100,7 @@ def test_round_decimals(self, flow: TestFlow) -> None:
             RoundModel(decimals=-2), (x,), flow, generate_random_test_inputs=False
         )
 
+    @unittest.skip("NaN and Inf are not enforced for backends.")
     def test_round_decimals_edge_cases(self, flow: TestFlow) -> None:
         # Test edge cases with decimal places
 
diff --git a/backends/test/suite/operators/test_rsqrt.py b/backends/test/suite/operators/test_rsqrt.py
index 175bbcdb2cc..cd4d5db89f5 100644
--- a/backends/test/suite/operators/test_rsqrt.py
+++ b/backends/test/suite/operators/test_rsqrt.py
@@ -7,6 +7,7 @@
 # pyre-unsafe
 
 import torch
+import unittest
 from executorch.backends.test.suite.flow import TestFlow
 
 from executorch.backends.test.suite.operators import (
@@ -45,6 +46,7 @@ def test_rsqrt_shapes(self, flow: TestFlow) -> None:
         # 3D tensor
         self._test_op(RsqrtModel(), (torch.rand(3, 4, 5) + 0.01,), flow)
 
+    @unittest.skip("NaN and Inf are not enforced for backends.")
     def test_rsqrt_edge_cases(self, flow: TestFlow) -> None:
         # Tensor with infinity
         x = torch.tensor([float("inf"), 1.0, 4.0])
diff --git a/backends/test/suite/operators/test_sqrt.py b/backends/test/suite/operators/test_sqrt.py
index c3874dcb209..0c8498981d8 100644
--- a/backends/test/suite/operators/test_sqrt.py
+++ b/backends/test/suite/operators/test_sqrt.py
@@ -7,6 +7,7 @@
 # pyre-unsafe
 
 import torch
+import unittest
 from executorch.backends.test.suite.flow import TestFlow
 
 from executorch.backends.test.suite.operators import (
@@ -45,6 +46,7 @@ def test_sqrt_shapes(self, flow: TestFlow) -> None:
         # 3D tensor
         self._test_op(SqrtModel(), (torch.rand(3, 4, 5),), flow)
 
+    @unittest.skip("NaN and Inf are not enforced for backends.")
     def test_sqrt_edge_cases(self, flow: TestFlow) -> None:
         # Test edge cases
 
diff --git a/backends/test/suite/operators/test_square.py b/backends/test/suite/operators/test_square.py
index 52cd739bf9f..489b48520d5 100644
--- a/backends/test/suite/operators/test_square.py
+++ b/backends/test/suite/operators/test_square.py
@@ -7,6 +7,7 @@
 # pyre-unsafe
 
 import torch
+import unittest
 from executorch.backends.test.suite.flow import TestFlow
 
 from executorch.backends.test.suite.operators import (
@@ -44,6 +45,7 @@ def test_square_shapes(self, flow: TestFlow) -> None:
         # 3D tensor
         self._test_op(SquareModel(), (torch.randn(3, 4, 5),), flow)
 
+    @unittest.skip("NaN and Inf are not enforced for backends.")
     def test_square_edge_cases(self, flow: TestFlow) -> None:
         # Test edge cases
 
diff --git a/backends/test/suite/operators/test_trunc.py b/backends/test/suite/operators/test_trunc.py
index 1d6d18817bd..d864bad6f68 100644
--- a/backends/test/suite/operators/test_trunc.py
+++ b/backends/test/suite/operators/test_trunc.py
@@ -7,6 +7,7 @@
 # pyre-unsafe
 
 import torch
+import unittest
 from executorch.backends.test.suite.flow import TestFlow
 
 from executorch.backends.test.suite.operators import (
@@ -44,6 +45,7 @@ def test_trunc_shapes(self, flow: TestFlow) -> None:
         # 3D tensor
         self._test_op(TruncModel(), (torch.randn(3, 4, 5) * 5,), flow)
 
+    @unittest.skip("NaN and Inf are not enforced for backends.")
     def test_trunc_edge_cases(self, flow: TestFlow) -> None:
         # Test edge cases
 

From 774f59ff7d6614fe1f97ad9b4b83fcc6bea91475 Mon Sep 17 00:00:00 2001
From: Gregory James Comer <gjcomer@meta.com>
Date: Thu, 4 Sep 2025 23:27:31 -0700
Subject: [PATCH 4/9] Update

[ghstack-poisoned]
---
 backends/test/suite/reporting.py | 12 ++++++++++++
 1 file changed, 12 insertions(+)

diff --git a/backends/test/suite/reporting.py b/backends/test/suite/reporting.py
index ce8a48dcc12..cdf2ce870e1 100644
--- a/backends/test/suite/reporting.py
+++ b/backends/test/suite/reporting.py
@@ -45,6 +45,8 @@
         ]
     )
 
+CSV_FIELD_NAMES.append("Error")
+
 
 # Operators that are excluded from the counts returned by count_ops. These are used to
 # exclude operatations that are not logically relevant or delegatable to backends.
@@ -365,6 +367,15 @@ def write_csv_header(output: TextIO):
 def write_csv_row(record: TestCaseSummary, output: TextIO):
     writer = csv.DictWriter(output, CSV_FIELD_NAMES)
 
+    # Truncate error message if it's too long, keeping first and last 200 characters
+    error_message = ""
+    if record.error is not None:
+        error_str = str(record.error)
+        if len(error_str) > 400:
+            error_message = error_str[:200] + "..." + error_str[-200:]
+        else:
+            error_message = error_str
+
     row = {
         "Test ID": record.name,
         "Test Case": record.base_name,
@@ -373,6 +384,7 @@ def write_csv_row(record: TestCaseSummary, output: TextIO):
         "Params": _serialize_params(record.params),
         "Result": record.result.to_short_str(),
         "Result Detail": record.result.to_detail_str(),
+        "Error": error_message,
         "Delegated": "True" if record.is_delegated() else "False",
         "Quantize Time (s)": (
             f"{record.quantize_time.total_seconds():.3f}"

From 51137cd007ac58ba36861fd3ae858b495778fd73 Mon Sep 17 00:00:00 2001
From: Gregory James Comer <gjcomer@meta.com>
Date: Thu, 4 Sep 2025 23:59:07 -0700
Subject: [PATCH 5/9] Update

[ghstack-poisoned]
---
 backends/test/suite/flow.py               | 8 +++++++-
 backends/test/suite/flows/coreml.py       | 1 +
 backends/test/suite/models/__init__.py    | 3 +++
 backends/test/suite/operators/__init__.py | 3 +++
 4 files changed, 14 insertions(+), 1 deletion(-)

diff --git a/backends/test/suite/flow.py b/backends/test/suite/flow.py
index b7a126eaf35..9df3805444a 100644
--- a/backends/test/suite/flow.py
+++ b/backends/test/suite/flow.py
@@ -1,6 +1,6 @@
 import logging
 
-from dataclasses import dataclass
+from dataclasses import dataclass, field
 from typing import Callable
 
 from executorch.backends.test.harness import Tester
@@ -35,6 +35,12 @@ class TestFlow:
     is_delegated: bool = True
     """ Indicates whether the flow is expected to generate CALL_DELEGATE nodes. """
 
+    skip_patterns: list[str] = field(default_factory=lambda: [])
+    """ Tests with names containing any substrings in this list are skipped. """
+
+    def should_skip_test(self, test_name: str) -> bool:
+        return any(pattern in test_name for pattern in self.skip_patterns)
+
 
 def all_flows() -> dict[str, TestFlow]:
     flows = []
diff --git a/backends/test/suite/flows/coreml.py b/backends/test/suite/flows/coreml.py
index fd956b64f05..8a532ff0003 100644
--- a/backends/test/suite/flows/coreml.py
+++ b/backends/test/suite/flows/coreml.py
@@ -19,6 +19,7 @@ def _create_coreml_flow(
             CoreMLTester, minimum_deployment_target=minimum_deployment_target
         ),
         quantize=quantize,
+        skip_patterns=["test_argmin", "test_argmax"],
     )
 
 
diff --git a/backends/test/suite/models/__init__.py b/backends/test/suite/models/__init__.py
index 65b546b0eb5..a6775dd8246 100644
--- a/backends/test/suite/models/__init__.py
+++ b/backends/test/suite/models/__init__.py
@@ -52,6 +52,9 @@ def wrapped_test(self):
             "use_dynamic_shapes": use_dynamic_shapes,
         }
         with TestContext(test_name, test_func.__name__, flow.name, params):
+            if flow.should_skip_test(test_name):
+                raise unittest.SkipTest(f"Skipping test due to matching flow {flow.name} skip patterns")
+
             test_func(self, flow, dtype, use_dynamic_shapes)
 
     wrapped_test._name = test_func.__name__  # type: ignore
diff --git a/backends/test/suite/operators/__init__.py b/backends/test/suite/operators/__init__.py
index 6ceb9086f71..464eb1aaf55 100644
--- a/backends/test/suite/operators/__init__.py
+++ b/backends/test/suite/operators/__init__.py
@@ -97,6 +97,9 @@ def _make_wrapped_test(
 ):
     def wrapped_test(self):
         with TestContext(test_name, test_base_name, flow.name, params):
+            if flow.should_skip_test(test_name):
+                raise unittest.SkipTest(f"Skipping test due to matching flow {flow.name} skip patterns")
+
             test_kwargs = copy.copy(params) or {}
             test_kwargs["flow"] = flow
 

From 4410ca125a8b7850c36140eeba305267016b8022 Mon Sep 17 00:00:00 2001
From: Gregory James Comer <gjcomer@meta.com>
Date: Fri, 5 Sep 2025 00:35:52 -0700
Subject: [PATCH 6/9] Update

[ghstack-poisoned]
---
 .../test/suite/generate_markdown_summary.py   | 24 +++++++++++++++++--
 1 file changed, 22 insertions(+), 2 deletions(-)

diff --git a/backends/test/suite/generate_markdown_summary.py b/backends/test/suite/generate_markdown_summary.py
index 37bf758fed0..fac3c2471f6 100644
--- a/backends/test/suite/generate_markdown_summary.py
+++ b/backends/test/suite/generate_markdown_summary.py
@@ -12,6 +12,25 @@
 #
 
 
+def escape_for_markdown(text: str) -> str:
+    """
+    Modify a string to properly display in a markdown table cell.
+    """
+    if not text:
+        return text
+    
+    # Replace newlines with <br /> tags
+    escaped = text.replace('\n', '<br />')
+
+    # Escape backslashes.
+    escaped = escaped.replace('\\', '\\\\')
+    
+    # Escape pipe characters that would break table structure
+    escaped = escaped.replace('|', '\\|')
+    
+    return escaped
+
+
 def generate_markdown(csv_path: str, exit_code: int = 0):  # noqa (C901)
     # Print warning if exit code is non-zero
     if exit_code != 0:
@@ -46,7 +65,7 @@ def generate_markdown(csv_path: str, exit_code: int = 0):  # noqa (C901)
 
     for row in data_rows:
         # Make a copy of the row to avoid modifying the original
-        processed_row = row.copy()
+        processed_row = [escape_for_markdown(cell) for cell in row]
 
         # Count results and collect failed tests
         if result_column_index is not None and result_column_index < len(row):
@@ -96,7 +115,8 @@ def generate_markdown(csv_path: str, exit_code: int = 0):  # noqa (C901)
     # Generate Failed Tests section
     print("# Failed Tests\n")
     if failed_tests:
-        print("| " + " | ".join(header) + " |")
+        escaped_header = [escape_for_markdown(col) for col in header]
+        print("| " + " | ".join(escaped_header) + " |")
         print("|" + "|".join(["---"] * len(header)) + "|")
         for row in failed_tests:
             print("| " + " | ".join(row) + " |")

From db1e1765dd0fa9bb87bbe3ef2f388f7f61b2e66e Mon Sep 17 00:00:00 2001
From: Gregory James Comer <gjcomer@meta.com>
Date: Tue, 9 Sep 2025 17:28:51 -0700
Subject: [PATCH 7/9] Update

[ghstack-poisoned]
---
 .github/workflows/_test_backend.yml        | 63 ++++++++++++++++++++++
 .github/workflows/nightly.yml              | 48 -----------------
 .github/workflows/test-backend-coreml.yml  | 28 ++++++++++
 .github/workflows/test-backend-qnn.yml     | 28 ++++++++++
 .github/workflows/test-backend-vulkan.yml  | 28 ++++++++++
 .github/workflows/test-backend-xnnpack.yml | 28 ++++++++++
 6 files changed, 175 insertions(+), 48 deletions(-)
 create mode 100644 .github/workflows/_test_backend.yml
 create mode 100644 .github/workflows/test-backend-coreml.yml
 create mode 100644 .github/workflows/test-backend-qnn.yml
 create mode 100644 .github/workflows/test-backend-vulkan.yml
 create mode 100644 .github/workflows/test-backend-xnnpack.yml

diff --git a/.github/workflows/_test_backend.yml b/.github/workflows/_test_backend.yml
new file mode 100644
index 00000000000..4a52cedcc84
--- /dev/null
+++ b/.github/workflows/_test_backend.yml
@@ -0,0 +1,63 @@
+name: Test Backend
+
+on:
+  workflow_call:
+    inputs:
+      backend:
+        description: 'Backend to test (xnnpack, coreml, vulkan, qnn)'
+        required: true
+        type: string
+      flows:
+        description: 'JSON array of flows to test'
+        required: true
+        type: string
+      runner:
+        description: 'Runner to use'
+        required: true
+        type: string
+      docker-image:
+        description: 'Docker image to use (Linux only)'
+        required: false
+        type: string
+      python-version:
+        description: 'Python version to use (macOS only)'
+        required: false
+        type: string
+      ref:
+        description: 'Git ref to checkout'
+        required: false
+        type: string
+        default: ${{ github.sha }}
+      timeout:
+        description: 'Job timeout in minutes'
+        required: false
+        type: number
+        default: 120
+
+jobs:
+  test-backend:
+    strategy:
+      fail-fast: false
+      matrix:
+        flow: ${{ fromJSON(inputs.flows) }}
+        suite: [models, operators]
+
+    uses: ${{ contains(inputs.runner, 'macos') && 'pytorch/test-infra/.github/workflows/macos_job.yml@main' || 'pytorch/test-infra/.github/workflows/linux_job_v2.yml@main' }}
+    permissions:
+      id-token: write
+      contents: read
+    with:
+      ref: ${{ inputs.ref }}
+      runner: ${{ inputs.runner }}
+      docker-image: ${{ inputs.docker-image }}
+      python-version: ${{ inputs.python-version }}
+      submodules: recursive
+      timeout: ${{ inputs.timeout }}
+      upload-artifact: test-report-${{ matrix.flow }}-${{ matrix.suite }}
+      script: |
+        set -eux
+
+        ${{ contains(inputs.runner, 'macos') && '# This is needed to get the prebuilt PyTorch wheel from S3
+        ${CONDA_RUN} --no-capture-output pip install awscli==1.37.21
+
+        source .ci/scripts/test_backend_macos.sh "${{ matrix.suite }}" "${{ matrix.flow }}" "${RUNNER_ARTIFACT_DIR}"' || 'source .ci/scripts/test_backend_linux.sh "${{ matrix.suite }}" "${{ matrix.flow }}" "${RUNNER_ARTIFACT_DIR}"' }}
diff --git a/.github/workflows/nightly.yml b/.github/workflows/nightly.yml
index c220b371c0a..4658fdc0d26 100644
--- a/.github/workflows/nightly.yml
+++ b/.github/workflows/nightly.yml
@@ -36,51 +36,3 @@ jobs:
     uses: ./.github/workflows/_link_check.yml
     with:
       ref: ${{ github.sha }}
-
-  backend-test-linux:
-    uses: pytorch/test-infra/.github/workflows/linux_job_v2.yml@main
-    strategy:
-      fail-fast: false
-      matrix:
-        flow: [
-          qnn, qnn_16a16w, qnn_16a8w, qnn_16a4w, qnn_16a4w_block, qnn_8a8w,
-          vulkan, vulkan_static_int8_per_channel,
-          xnnpack, xnnpack_dynamic_int8_per_channel, xnnpack_static_int8_per_channel, xnnpack_static_int8_per_tensor
-        ]
-        suite: [models, operators]
-    with:
-      ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
-      runner: linux.4xlarge.memory
-      docker-image: ci-image:executorch-ubuntu-22.04-clang12
-      submodules: recursive
-      timeout: 120
-      upload-artifact: test-report-${{ matrix.flow }}-${{ matrix.suite }}
-      script: |
-        set -eux
-
-        source .ci/scripts/test_backend_linux.sh "${{ matrix.suite }}" "${{ matrix.flow }}" "${RUNNER_ARTIFACT_DIR}"
-
-  backend-test-macos:
-    uses: pytorch/test-infra/.github/workflows/macos_job.yml@main
-    permissions:
-      id-token: write
-      contents: read
-    strategy:
-      fail-fast: false
-      matrix:
-        flow: [coreml, coreml_static_int8]
-        suite: [models, operators]
-    with:
-      ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
-      runner: macos-m1-stable
-      python-version: 3.12
-      submodules: recursive
-      timeout: 120
-      upload-artifact: test-report-${{ matrix.flow }}-${{ matrix.suite }}
-      script: |
-        set -eux
-
-        # This is needed to get the prebuilt PyTorch wheel from S3
-        ${CONDA_RUN} --no-capture-output pip install awscli==1.37.21
-
-        source .ci/scripts/test_backend_macos.sh "${{ matrix.suite }}" "${{ matrix.flow }}" "${RUNNER_ARTIFACT_DIR}"
diff --git a/.github/workflows/test-backend-coreml.yml b/.github/workflows/test-backend-coreml.yml
new file mode 100644
index 00000000000..4aa90eabfe9
--- /dev/null
+++ b/.github/workflows/test-backend-coreml.yml
@@ -0,0 +1,28 @@
+name: Test CoreML Backend
+
+on:
+  schedule:
+    - cron: 0 2 * * *
+  push:
+    tags:
+      - ciflow/nightly/*
+  pull_request:
+    paths:
+      - .github/workflows/test-backend-coreml.yml
+      - .github/workflows/_test_backend.yml
+  workflow_dispatch:
+
+concurrency:
+  group: ${{ github.workflow }}--${{ github.event.pull_request.number || github.sha }}-${{ github.event_name == 'workflow_dispatch' }}
+  cancel-in-progress: true
+
+jobs:
+  test-coreml:
+    uses: ./.github/workflows/_test_backend.yml
+    with:
+      backend: coreml
+      flows: '["coreml", "coreml_static_int8"]'
+      runner: macos-m1-stable
+      python-version: "3.12"
+      ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
+      timeout: 120
diff --git a/.github/workflows/test-backend-qnn.yml b/.github/workflows/test-backend-qnn.yml
new file mode 100644
index 00000000000..a100b1398fc
--- /dev/null
+++ b/.github/workflows/test-backend-qnn.yml
@@ -0,0 +1,28 @@
+name: Test QNN Backend
+
+on:
+  schedule:
+    - cron: 0 2 * * *
+  push:
+    tags:
+      - ciflow/nightly/*
+  pull_request:
+    paths:
+      - .github/workflows/test-backend-qnn.yml
+      - .github/workflows/_test_backend.yml
+  workflow_dispatch:
+
+concurrency:
+  group: ${{ github.workflow }}--${{ github.event.pull_request.number || github.sha }}-${{ github.event_name == 'workflow_dispatch' }}
+  cancel-in-progress: true
+
+jobs:
+  test-qnn:
+    uses: ./.github/workflows/_test_backend.yml
+    with:
+      backend: qnn
+      flows: '["qnn", "qnn_16a16w", "qnn_16a8w", "qnn_16a4w", "qnn_16a4w_block", "qnn_8a8w"]'
+      runner: linux.4xlarge.memory
+      docker-image: ci-image:executorch-ubuntu-22.04-clang12
+      ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
+      timeout: 120
diff --git a/.github/workflows/test-backend-vulkan.yml b/.github/workflows/test-backend-vulkan.yml
new file mode 100644
index 00000000000..9ecb9e827ee
--- /dev/null
+++ b/.github/workflows/test-backend-vulkan.yml
@@ -0,0 +1,28 @@
+name: Test Vulkan Backend
+
+on:
+  schedule:
+    - cron: 0 2 * * *
+  push:
+    tags:
+      - ciflow/nightly/*
+  pull_request:
+    paths:
+      - .github/workflows/test-backend-vulkan.yml
+      - .github/workflows/_test_backend.yml
+  workflow_dispatch:
+
+concurrency:
+  group: ${{ github.workflow }}--${{ github.event.pull_request.number || github.sha }}-${{ github.event_name == 'workflow_dispatch' }}
+  cancel-in-progress: true
+
+jobs:
+  test-vulkan:
+    uses: ./.github/workflows/_test_backend.yml
+    with:
+      backend: vulkan
+      flows: '["vulkan", "vulkan_static_int8_per_channel"]'
+      runner: linux.4xlarge.memory
+      docker-image: ci-image:executorch-ubuntu-22.04-clang12
+      ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
+      timeout: 120
diff --git a/.github/workflows/test-backend-xnnpack.yml b/.github/workflows/test-backend-xnnpack.yml
new file mode 100644
index 00000000000..da8e277f8d5
--- /dev/null
+++ b/.github/workflows/test-backend-xnnpack.yml
@@ -0,0 +1,28 @@
+name: Test XNNPACK Backend
+
+on:
+  schedule:
+    - cron: 0 2 * * *
+  push:
+    tags:
+      - ciflow/nightly/*
+  pull_request:
+    paths:
+      - .github/workflows/test-backend-xnnpack.yml
+      - .github/workflows/_test_backend.yml
+  workflow_dispatch:
+
+concurrency:
+  group: ${{ github.workflow }}--${{ github.event.pull_request.number || github.sha }}-${{ github.event_name == 'workflow_dispatch' }}
+  cancel-in-progress: true
+
+jobs:
+  test-xnnpack:
+    uses: ./.github/workflows/_test_backend.yml
+    with:
+      backend: xnnpack
+      flows: '["xnnpack", "xnnpack_dynamic_int8_per_channel", "xnnpack_static_int8_per_channel", "xnnpack_static_int8_per_tensor"]'
+      runner: linux.4xlarge.memory
+      docker-image: ci-image:executorch-ubuntu-22.04-clang12
+      ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
+      timeout: 120

From 1d00f658fea77b537d12ac4fc3e619a17aa3c9d2 Mon Sep 17 00:00:00 2001
From: Gregory James Comer <gjcomer@meta.com>
Date: Tue, 9 Sep 2025 17:37:41 -0700
Subject: [PATCH 8/9] Update

[ghstack-poisoned]
---
 .github/workflows/_test_backend.yml        | 60 +++++++++++++++-------
 .github/workflows/test-backend-coreml.yml  |  3 +-
 .github/workflows/test-backend-qnn.yml     |  3 +-
 .github/workflows/test-backend-vulkan.yml  |  3 +-
 .github/workflows/test-backend-xnnpack.yml |  3 +-
 5 files changed, 45 insertions(+), 27 deletions(-)

diff --git a/.github/workflows/_test_backend.yml b/.github/workflows/_test_backend.yml
index 4a52cedcc84..b1c8370bc24 100644
--- a/.github/workflows/_test_backend.yml
+++ b/.github/workflows/_test_backend.yml
@@ -11,18 +11,6 @@ on:
         description: 'JSON array of flows to test'
         required: true
         type: string
-      runner:
-        description: 'Runner to use'
-        required: true
-        type: string
-      docker-image:
-        description: 'Docker image to use (Linux only)'
-        required: false
-        type: string
-      python-version:
-        description: 'Python version to use (macOS only)'
-        required: false
-        type: string
       ref:
         description: 'Git ref to checkout'
         required: false
@@ -33,31 +21,65 @@ on:
         required: false
         type: number
         default: 120
+      run-linux:
+        description: 'Whether to run Linux tests'
+        required: false
+        type: boolean
+        default: false
+      run-macos:
+        description: 'Whether to run macOS tests'
+        required: false
+        type: boolean
+        default: false
 
 jobs:
-  test-backend:
+  test-backend-linux:
+    if: ${{ inputs.run-linux }}
+    strategy:
+      fail-fast: false
+      matrix:
+        flow: ${{ fromJSON(inputs.flows) }}
+        suite: [models, operators]
+
+    uses: pytorch/test-infra/.github/workflows/linux_job_v2.yml@main
+    permissions:
+      id-token: write
+      contents: read
+    with:
+      ref: ${{ inputs.ref }}
+      runner: linux.4xlarge.memory
+      docker-image: ci-image:executorch-ubuntu-22.04-clang12
+      submodules: recursive
+      timeout: ${{ inputs.timeout }}
+      upload-artifact: test-report-${{ matrix.flow }}-${{ matrix.suite }}
+      script: |
+        set -eux
+
+        source .ci/scripts/test_backend_linux.sh "${{ matrix.suite }}" "${{ matrix.flow }}" "${RUNNER_ARTIFACT_DIR}"
+
+  test-backend-macos:
+    if: ${{ inputs.run-macos }}
     strategy:
       fail-fast: false
       matrix:
         flow: ${{ fromJSON(inputs.flows) }}
         suite: [models, operators]
 
-    uses: ${{ contains(inputs.runner, 'macos') && 'pytorch/test-infra/.github/workflows/macos_job.yml@main' || 'pytorch/test-infra/.github/workflows/linux_job_v2.yml@main' }}
+    uses: pytorch/test-infra/.github/workflows/macos_job.yml@main
     permissions:
       id-token: write
       contents: read
     with:
       ref: ${{ inputs.ref }}
-      runner: ${{ inputs.runner }}
-      docker-image: ${{ inputs.docker-image }}
-      python-version: ${{ inputs.python-version }}
+      runner: macos-m1-stable
+      python-version: "3.12"
       submodules: recursive
       timeout: ${{ inputs.timeout }}
       upload-artifact: test-report-${{ matrix.flow }}-${{ matrix.suite }}
       script: |
         set -eux
 
-        ${{ contains(inputs.runner, 'macos') && '# This is needed to get the prebuilt PyTorch wheel from S3
+        # This is needed to get the prebuilt PyTorch wheel from S3
         ${CONDA_RUN} --no-capture-output pip install awscli==1.37.21
 
-        source .ci/scripts/test_backend_macos.sh "${{ matrix.suite }}" "${{ matrix.flow }}" "${RUNNER_ARTIFACT_DIR}"' || 'source .ci/scripts/test_backend_linux.sh "${{ matrix.suite }}" "${{ matrix.flow }}" "${RUNNER_ARTIFACT_DIR}"' }}
+        source .ci/scripts/test_backend_macos.sh "${{ matrix.suite }}" "${{ matrix.flow }}" "${RUNNER_ARTIFACT_DIR}"
diff --git a/.github/workflows/test-backend-coreml.yml b/.github/workflows/test-backend-coreml.yml
index 4aa90eabfe9..c6970ddff61 100644
--- a/.github/workflows/test-backend-coreml.yml
+++ b/.github/workflows/test-backend-coreml.yml
@@ -22,7 +22,6 @@ jobs:
     with:
       backend: coreml
       flows: '["coreml", "coreml_static_int8"]'
-      runner: macos-m1-stable
-      python-version: "3.12"
       ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
       timeout: 120
+      run-macos: true
diff --git a/.github/workflows/test-backend-qnn.yml b/.github/workflows/test-backend-qnn.yml
index a100b1398fc..64dc7cdce36 100644
--- a/.github/workflows/test-backend-qnn.yml
+++ b/.github/workflows/test-backend-qnn.yml
@@ -22,7 +22,6 @@ jobs:
     with:
       backend: qnn
       flows: '["qnn", "qnn_16a16w", "qnn_16a8w", "qnn_16a4w", "qnn_16a4w_block", "qnn_8a8w"]'
-      runner: linux.4xlarge.memory
-      docker-image: ci-image:executorch-ubuntu-22.04-clang12
       ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
       timeout: 120
+      run-linux: true
diff --git a/.github/workflows/test-backend-vulkan.yml b/.github/workflows/test-backend-vulkan.yml
index 9ecb9e827ee..f04fdcdd1f1 100644
--- a/.github/workflows/test-backend-vulkan.yml
+++ b/.github/workflows/test-backend-vulkan.yml
@@ -22,7 +22,6 @@ jobs:
     with:
       backend: vulkan
       flows: '["vulkan", "vulkan_static_int8_per_channel"]'
-      runner: linux.4xlarge.memory
-      docker-image: ci-image:executorch-ubuntu-22.04-clang12
       ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
       timeout: 120
+      run-linux: true
diff --git a/.github/workflows/test-backend-xnnpack.yml b/.github/workflows/test-backend-xnnpack.yml
index da8e277f8d5..2ae423dd99b 100644
--- a/.github/workflows/test-backend-xnnpack.yml
+++ b/.github/workflows/test-backend-xnnpack.yml
@@ -22,7 +22,6 @@ jobs:
     with:
       backend: xnnpack
       flows: '["xnnpack", "xnnpack_dynamic_int8_per_channel", "xnnpack_static_int8_per_channel", "xnnpack_static_int8_per_tensor"]'
-      runner: linux.4xlarge.memory
-      docker-image: ci-image:executorch-ubuntu-22.04-clang12
       ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
       timeout: 120
+      run-linux: true

From 311e536a86d29fca3d8482974e1e2d9bef0a1940 Mon Sep 17 00:00:00 2001
From: Gregory James Comer <gjcomer@meta.com>
Date: Tue, 9 Sep 2025 17:45:50 -0700
Subject: [PATCH 9/9] Update

[ghstack-poisoned]
---
 .github/workflows/_test_backend.yml | 6 ------
 1 file changed, 6 deletions(-)

diff --git a/.github/workflows/_test_backend.yml b/.github/workflows/_test_backend.yml
index b1c8370bc24..8b572825bdf 100644
--- a/.github/workflows/_test_backend.yml
+++ b/.github/workflows/_test_backend.yml
@@ -42,9 +42,6 @@ jobs:
         suite: [models, operators]
 
     uses: pytorch/test-infra/.github/workflows/linux_job_v2.yml@main
-    permissions:
-      id-token: write
-      contents: read
     with:
       ref: ${{ inputs.ref }}
       runner: linux.4xlarge.memory
@@ -66,9 +63,6 @@ jobs:
         suite: [models, operators]
 
     uses: pytorch/test-infra/.github/workflows/macos_job.yml@main
-    permissions:
-      id-token: write
-      contents: read
     with:
       ref: ${{ inputs.ref }}
       runner: macos-m1-stable