[release/2.13] [LNL] test_transformers_xpu.py::TestTransformersXPU::test_transformerencoder_fastpath_use_torchscript_False_enable_nested_tensor_True_use_autocast_False_d_model AssertionError: Tensor-likes are not close!

@jmamzax is already working on this.

Since Aug 18, 2026.

### 🐛 Describe the bug

test_transformerencoder_fastpath_use_torchscript_False_enable_nested_tensor_True_use_autocast_False_d_model_12_xpu and test_transformerencoder_fastpath_use_torchscript_False_enable_nested_tensor_True_use_autocast_False_d_model_256_xpu fail on LNL (Windows). Tests passed on BMG PT2.13 and LNL PT2.12

**Affected Test Cases**

```
test_transformers_xpu.py::TestTransformersXPU::test_transformerencoder_fastpath_use_torchscript_False_enable_nested_tensor_True_use_autocast_False_d_model_12_xpu
test_transformers_xpu.py::TestTransformersXPU::test_transformerencoder_fastpath_use_torchscript_False_enable_nested_tensor_True_use_autocast_False_d_model_256_xpu
```

**Error Message**

- test_transformerencoder_fastpath_use_torchscript_False_enable_nested_tensor_True_use_autocast_False_d_model_12_xpu

```
_ TestTransformersXPU.test_transformerencoder_fastpath_use_torchscript_False_enable_nested_tensor_True_use_autocast_False_d_model_12_xpu _
[gw0] win32 -- Python 3.12.13 C:\Users\gta\miniforge3\envs\pytorch_2.13\python.exe
Traceback (most recent call last):
File "C:\Users\gta\repositories\pytorch\pytorch\third_party\torch-xpu-ops\test\xpu\test_transformers_xpu.py", line 709, in test_transformerencoder_fastpath
self.assertEqual(fastpath_output_expanded, slowpath_output)
File "C:\Users\gta\miniforge3\envs\pytorch_2.13\Lib\site-packages\torch\testing\_internal\common_utils.py", line 4571, in assertEqual
raise error_metas.pop()[0].to_error( # type: ignore[index]
AssertionError: Tensor-likes are not close!

Mismatched elements: 24 / 72 (33.3%)
Greatest absolute difference: 1.5866498947143555 at index (1, 0, 8) (up to 1e-05 allowed)
Greatest relative difference: 21.598501205444336 at index (1, 0, 9) (up to 1.3e-06 allowed)

To execute this test, run the following from the base repo dir:
PYTORCH_TEST_WITH_SLOW=1 python test\xpu\test_transformers_xpu.py TestTransformersXPU.test_transformerencoder_fastpath_use_torchscript_False_enable_nested_tensor_True_use_autocast_False_d_model_12_xpu

```

- test_transformerencoder_fastpath_use_torchscript_False_enable_nested_tensor_True_use_autocast_False_d_model_256_xpu

```
_ TestTransformersXPU.test_transformerencoder_fastpath_use_torchscript_False_enable_nested_tensor_True_use_autocast_False_d_model_256_xpu _
[gw0] win32 -- Python 3.12.13 C:\Users\gta\miniforge3\envs\pytorch_2.13\python.exe
Traceback (most recent call last):
File "C:\Users\gta\repositories\pytorch\pytorch\third_party\torch-xpu-ops\test\xpu\test_transformers_xpu.py", line 709, in test_transformerencoder_fastpath
self.assertEqual(fastpath_output_expanded, slowpath_output)
File "C:\Users\gta\miniforge3\envs\pytorch_2.13\Lib\site-packages\torch\testing\_internal\common_utils.py", line 4571, in assertEqual
raise error_metas.pop()[0].to_error( # type: ignore[index]
AssertionError: Tensor-likes are not close!

Mismatched elements: 512 / 1536 (33.3%)
Greatest absolute difference: 2.053421974182129 at index (1, 0, 39) (up to 1e-05 allowed)
Greatest relative difference: 1790.7701416015625 at index (0, 0, 23) (up to 1.3e-06 allowed)

To execute this test, run the following from the base repo dir:
PYTORCH_TEST_WITH_SLOW=1 python test\xpu\test_transformers_xpu.py TestTransformersXPU.test_transformerencoder_fastpath_use_torchscript_False_enable_nested_tensor_True_use_autocast_False_d_model_256_xpu

```

### Versions

PyTorch version: 2.13.0+xpu
Is debug build: False
CUDA used to build PyTorch: None
ROCM used to build PyTorch: N/A

OS: Microsoft Windows 11 Pro (10.0.26100 64-bit)
GCC version: Could not collect
Clang version: Could not collect
CMake version: version 3.31.6
Libc version: N/A

Python version: 3.12.13 | packaged by conda-forge | (main, Mar 5 2026, 16:36:12) [MSC v.1944 64 bit (AMD64)] (64-bit runtime)
Python platform: Windows-11-10.0.26100-SP0
Is CUDA available: False
CUDA runtime version: No CUDA
CUDA_MODULE_LOADING set to: N/A
GPU models and configuration: No CUDA
Nvidia driver version: No CUDA
cuDNN version: No CUDA
Is XPU available: True
XPU used to build PyTorch: 20260000
Intel GPU driver version:
* 32.0.101.8826 (20260529000000.******+***)
Intel GPU models onboard:
* Intel(R) Arc(TM) 140V GPU (16GB)
Intel GPU models detected:
* [0] _XpuDeviceProperties(name='Intel(R) Arc(TM) 140V GPU (16GB)', platform_name='Intel(R) oneAPI Unified Runtime over Level-Zero V2', type='gpu', device_id=0x64A0, uuid=8680a064-0400-0000-0002-000000000000, driver_version='1.15.37858', total_memory=16870MB, local_mem_size=128KB, last_level_cache_size=8192KB, max_compute_units=64, memory_clock_rate=0MHz, memory_bus_width=64-bit, gpu_eu_count=64, gpu_subslice_count=8, max_work_group_size=1024, max_num_sub_groups=64, sub_group_sizes=[16 32], has_fp16=1, has_fp64=1, has_atomic64=1, is_integrated_gpu=1)
HIP runtime version: N/A
MIOpen runtime version: N/A
Is XNNPACK available: False
Caching allocator config: N/A

CPU:
Name: Intel(R) Core(TM) Ultra 7 258V
Manufacturer: GenuineIntel
Family: 774
Architecture: 9
ProcessorType: 3
DeviceID: CPU0
CurrentClockSpeed: 2200
MaxClockSpeed: 2200
L2CacheSize: 14336
L2CacheSpeed: None
Revision: None

Versions of relevant libraries:
[pip3] dpcpp-cpp-rt==2026.0.0
[pip3] intel-cmplr-lib-rt==2026.0.0
[pip3] intel-cmplr-lib-ur==2026.0.0
[pip3] intel-cmplr-lic-rt==2026.0.0
[pip3] intel-opencl-rt==2026.0.0
[pip3] intel-openmp==2026.0.0
[pip3] intel-pti==0.17.0
[pip3] intel-sycl-rt==2026.0.0
[pip3] mkl==2026.0.0
[pip3] mkl-include==2024.2.0
[pip3] mkl-static==2024.2.0
[pip3] mypy_extensions==1.1.0
[pip3] numpy==1.26.2
[pip3] onemkl-license==2026.0.0
[pip3] onemkl-sycl-blas==2026.0.0
[pip3] onemkl-sycl-dft==2026.0.0
[pip3] onemkl-sycl-lapack==2026.0.0
[pip3] onemkl-sycl-rng==2026.0.0
[pip3] onemkl-sycl-sparse==2026.0.0
[pip3] onnx==1.21.0
[pip3] onnx-ir==0.1.16
[pip3] onnxscript==0.6.2
[pip3] optree==0.13.0
[pip3] tbb==2023.0.0
[pip3] tbb-devel==2021.13.1
[pip3] tcmlib==1.5.0
[pip3] torch==2.13.0+xpu
[pip3] torchaudio==2.11.0+xpu
[pip3] torchvision==0.28.0+xpu
[pip3] triton-xpu==3.7.2
[pip3] umf==1.1.0
[conda] dpcpp-cpp-rt 2026.0.0 pypi_0 pypi
[conda] intel-cmplr-lib-rt 2026.0.0 pypi_0 pypi
[conda] intel-cmplr-lib-ur 2026.0.0 pypi_0 pypi
[conda] intel-cmplr-lic-rt 2026.0.0 pypi_0 pypi
[conda] intel-opencl-rt 2026.0.0 pypi_0 pypi
[conda] intel-openmp 2026.0.0 pypi_0 pypi
[conda] intel-pti 0.17.0 pypi_0 pypi
[conda] intel-sycl-rt 2026.0.0 pypi_0 pypi
[conda] mkl 2026.0.0 pypi_0 pypi
[conda] mkl-include 2024.2.0 pypi_0 pypi
[conda] mkl-static 2024.2.0 pypi_0 pypi
[conda] numpy 1.26.2 pypi_0 pypi
[conda] onemkl-license 2026.0.0 pypi_0 pypi
[conda] onemkl-sycl-blas 2026.0.0 pypi_0 pypi
[conda] onemkl-sycl-dft 2026.0.0 pypi_0 pypi
[conda] onemkl-sycl-lapack 2026.0.0 pypi_0 pypi
[conda] onemkl-sycl-rng 2026.0.0 pypi_0 pypi
[conda] onemkl-sycl-sparse 2026.0.0 pypi_0 pypi
[conda] optree 0.13.0 pypi_0 pypi
[conda] tbb 2023.0.0 pypi_0 pypi
[conda] tbb-devel 2021.13.1 pypi_0 pypi
[conda] tcmlib 1.5.0 pypi_0 pypi
[conda] torch 2.13.0+xpu pypi_0 pypi
[conda] torchaudio 2.11.0+xpu pypi_0 pypi
[conda] torchvision 0.28.0+xpu pypi_0 pypi
[conda] triton-xpu 3.7.2 pypi_0 pypi
[conda] umf 1.1.0 pypi_0 pypi