build(deps): bump torch & torchvision from 2.8.0 to 2.9.0 in /requirements

.lightning/workflows/pytorch.yml

@@ -121,7 +121,11 @@ run: |
 
   echo "Install package"
   extra=$(python -c "print({'lightning': 'pytorch-'}.get('${PACKAGE_NAME}', ''))")
-  uv pip install -e ".[${extra}dev]" --upgrade
+
+  # Use find-links to prefer CUDA-specific packages from PyTorch index
+  uv pip install -e ".[${extra}dev]" --upgrade \
+    --find-links="https://download.pytorch.org/whl/${UV_TORCH_BACKEND}"
+  uv pip list
 
   echo "Ensure only a single package is installed"
   if [ "${PACKAGE_NAME}" == "pytorch" ]; then

requirements/fabric/base.txt

@@ -1,7 +1,7 @@
 # NOTE: the upper bound for the package version is only set for CI stability, and it is dropped while installing this package
 #  in case you want to preserve/enforce restrictions on the latest compatible version, add "strict" as an in-line comment
 
-torch >=2.1.0, <2.9.0
+torch >=2.1.0, <2.10.0
 fsspec[http] >=2022.5.0, <2025.11.0
 packaging >=20.0, <=25.0
 typing-extensions >4.5.0, <4.16.0

requirements/pytorch/base.txt

@@ -1,7 +1,7 @@
 # NOTE: the upper bound for the package version is only set for CI stability, and it is dropped while installing this package
 #  in case you want to preserve/enforce restrictions on the latest compatible version, add "strict" as an in-line comment
 
-torch >=2.1.0, <2.9.0
+torch >=2.1.0, <2.10.0
 tqdm >=4.57.0, <4.68.0
 PyYAML >5.4, <6.1.0
 fsspec[http] >=2022.5.0, <2025.11.0

requirements/pytorch/test.txt

@@ -21,5 +21,6 @@ uvicorn  # for `ServableModuleValidator`  # not setting version as re-defined in
 
 tensorboard >=2.11, <2.21.0  # for `TensorBoardLogger`
 
-torch-tensorrt; platform_system == "Linux" and python_version >= "3.12"
+# TODO: resolve GPU test failures for TensorRT due to defaulting to cu13 installations
+torch-tensorrt<2.9.0; platform_system == "Linux" and python_version >= "3.12"
 huggingface-hub

requirements/typing.txt

@@ -1,5 +1,5 @@
 mypy==1.18.2
-torch==2.8.0
+torch==2.9.0
 
 types-Markdown
 types-PyYAML

src/lightning/fabric/utilities/spike.py

@@ -126,10 +126,10 @@ def _handle_spike(self, fabric: "Fabric", batch_idx: int) -> None:
         raise TrainingSpikeException(batch_idx=batch_idx)
 
     def _check_atol(self, val_a: Union[float, torch.Tensor], val_b: Union[float, torch.Tensor]) -> bool:
-        return (self.atol is None) or bool(abs(val_a - val_b) >= abs(self.atol))
+        return (self.atol is None) or bool(abs(val_a - val_b) >= abs(self.atol))  # type: ignore
 
     def _check_rtol(self, val_a: Union[float, torch.Tensor], val_b: Union[float, torch.Tensor]) -> bool:
-        return (self.rtol is None) or bool(abs(val_a - val_b) >= abs(self.rtol * val_b))
+        return (self.rtol is None) or bool(abs(val_a - val_b) >= abs(self.rtol * val_b))  # type: ignore
 
     def _is_better(self, diff_val: torch.Tensor) -> bool:
         if self.mode == "min":

tests/tests_fabric/conftest.py

@@ -111,6 +111,7 @@ def thread_police_duuu_daaa_duuu_daaa():
             sys.version_info >= (3, 9)
             and isinstance(thread, _ExecutorManagerThread)
             or "ThreadPoolExecutor-" in thread.name
+            or thread.name == "InductorSubproc"  # torch.compile
         ):
             # probably `torch.compile`, can't narrow it down further
             continue

tests/tests_pytorch/conftest.py

@@ -170,6 +170,7 @@ def thread_police_duuu_daaa_duuu_daaa():
             sys.version_info >= (3, 9)
             and isinstance(thread, _ExecutorManagerThread)
             or "ThreadPoolExecutor-" in thread.name
+            or thread.name == "InductorSubproc"  # torch.compile
         ):
             # probably `torch.compile`, can't narrow it down further
             continue