Merge branch 'master' into dependabot-pip-requirements-rich-gte-12.3.0-and-lt-14.3.0

Author: Borda

.github/workflows/_legacy-checkpoints.yml

@@ -58,7 +58,7 @@ jobs:
       - uses: actions/checkout@v5
 
       - name: Install uv and set Python version
-        uses: astral-sh/setup-uv@v6
+        uses: astral-sh/setup-uv@v7
         with:
           python-version: "3.9"
           # TODO: Avoid activating environment like this

.github/workflows/ci-tests-fabric.yml

@@ -71,7 +71,7 @@ jobs:
       - uses: actions/checkout@v5
 
       - name: Install uv and set Python version
-        uses: astral-sh/setup-uv@v6
+        uses: astral-sh/setup-uv@v7
         with:
           python-version: ${{ matrix.config.python-version || '3.9' }}
           # TODO: Avoid activating environment like this

.github/workflows/ci-tests-pytorch.yml

@@ -76,7 +76,7 @@ jobs:
       - uses: actions/checkout@v5
 
       - name: Install uv and set Python version
-        uses: astral-sh/setup-uv@v6
+        uses: astral-sh/setup-uv@v7
         with:
           python-version: ${{ matrix.config.python-version || '3.9' }}
           # TODO: Avoid activating environment like this

.github/workflows/code-checks.yml

@@ -31,7 +31,7 @@ jobs:
       - uses: actions/checkout@v5
 
       - name: Install uv and set Python version
-        uses: astral-sh/setup-uv@v6
+        uses: astral-sh/setup-uv@v7
         with:
           python-version: "3.11"
           # TODO: Avoid activating environment like this

.github/workflows/docs-build.yml

@@ -73,7 +73,7 @@ jobs:
           lfs: ${{ matrix.pkg-name == 'pytorch' }}
 
       - name: Install uv and set Python version
-        uses: astral-sh/setup-uv@v6
+        uses: astral-sh/setup-uv@v7
         with:
           python-version: "3.10"
           # TODO: Avoid activating environment like this

.pre-commit-config.yaml

@@ -23,7 +23,7 @@ ci:
 
 repos:
   - repo: https://github.com/pre-commit/pre-commit-hooks
-    rev: v5.0.0
+    rev: v6.0.0
     hooks:
       - id: end-of-file-fixer
       - id: trailing-whitespace
@@ -70,7 +70,7 @@ repos:
       - id: sphinx-lint
 
   - repo: https://github.com/astral-sh/ruff-pre-commit
-    rev: v0.12.2
+    rev: v0.13.3
     hooks:
       # try to fix what is possible
       - id: ruff
@@ -95,8 +95,8 @@ repos:
               README.md
           )$
 
-  - repo: https://github.com/pre-commit/mirrors-prettier
-    rev: v3.1.0
+  - repo: https://github.com/JoC0de/pre-commit-prettier
+    rev: v3.6.2
     hooks:
       - id: prettier
         # https://prettier.io/docs/en/options.html#print-width

Makefile

@@ -7,7 +7,11 @@ export SPHINX_MOCK_REQUIREMENTS=1
 # install only Lightning Trainer packages
 export PACKAGE_NAME=pytorch
 
+
+# In Lightning Studio, the `lightning` package comes pre-installed.
+# Uninstall it first to ensure the editable install works correctly.
 setup:
+	uv pip uninstall lightning pytorch-lightning lightning-fabric || true
 	uv pip install -r requirements.txt \
 	    -r requirements/pytorch/base.txt \
 	    -r requirements/pytorch/test.txt \

docs/source-pytorch/deploy/production_advanced_2.rst

@@ -7,15 +7,20 @@ Deploy models into production (advanced)
 
 ----
 
-*********************************
-Compile your model to TorchScript
-*********************************
-`TorchScript <https://pytorch.org/docs/stable/jit.html>`_ allows you to serialize your models in a way that it can be loaded in non-Python environments.
-The ``LightningModule`` has a handy method :meth:`~lightning.pytorch.core.LightningModule.to_torchscript` that returns a scripted module which you
-can save or directly use.
+************************************
+Export your model with torch.export
+************************************
+
+`torch.export <https://pytorch.org/docs/stable/export.html>`_ is the recommended way to capture PyTorch models for
+deployment in production environments. It produces a clean intermediate representation with strong soundness guarantees,
+making models suitable for inference optimization and cross-platform deployment.
+You can export any ``LightningModule`` using the ``torch.export.export()`` API.
 
 .. testcode:: python
 
+    import torch
+    from torch.export import export
+
     class SimpleModel(LightningModule):
         def __init__(self):
             super().__init__()
@@ -25,25 +30,27 @@ can save or directly use.
             return torch.relu(self.l1(x.view(x.size(0), -1)))
 
 
-    # create the model
+    # create the model and example input
     model = SimpleModel()
-    script = model.to_torchscript()
+    example_input = torch.randn(1, 64)
 
-    # save for use in production environment
-    torch.jit.save(script, "model.pt")
+    # export the model
+    exported_program = export(model, (example_input,))
 
-It is recommended that you install the latest supported version of PyTorch to use this feature without limitations.
+    # save for use in production environment
+    torch.export.save(exported_program, "model.pt2")
 
-Once you have the exported model, you can run it in PyTorch or C++ runtime:
+It is recommended that you install the latest supported version of PyTorch to use this feature without
+limitations. Once you have the exported model, you can load and run it:
 
 .. code-block:: python
 
     inp = torch.rand(1, 64)
-    scripted_module = torch.jit.load("model.pt")
-    output = scripted_module(inp)
+    loaded_program = torch.export.load("model.pt2")
+    output = loaded_program.module()(inp)
 
 
-If you want to script a different method, you can decorate the method with :func:`torch.jit.export`:
+For more complex models, you can also export specific methods by creating a wrapper:
 
 .. code-block:: python
 
@@ -54,7 +61,6 @@ If you want to script a different method, you can decorate the method with :func
             self.dropout = nn.Dropout()
             self.mc_iteration = mc_iteration
 
-        @torch.jit.export
         def predict_step(self, batch, batch_idx):
             # enable Monte Carlo Dropout
             self.dropout.train()
@@ -66,4 +72,11 @@ If you want to script a different method, you can decorate the method with :func
 
 
     model = LitMCdropoutModel(...)
-    script = model.to_torchscript(file_path="model.pt", method="script")
+    example_batch = torch.randn(32, 10)  # example input
+
+    # Export the predict_step method
+    exported_program = torch.export.export(
+        lambda batch, idx: model.predict_step(batch, idx),
+        (example_batch, 0)
+    )
+    torch.export.save(exported_program, "mc_dropout_model.pt2")

requirements/doctests.txt

@@ -1,2 +1,2 @@
 pytest ==8.4.2
-pytest-doctestplus ==1.4.0
+pytest-doctestplus ==1.5.0

requirements/fabric/strategies.txt

@@ -5,5 +5,5 @@
 
 # note: is a bug around 0.10 with `MPS_Accelerator must implement all abstract methods`
 #  shall be resolved by https://github.com/microsoft/DeepSpeed/issues/4372
-deepspeed >=0.14.1,<=0.15.0; platform_system != "Windows" and platform_system != "Darwin"  # strict
+deepspeed >=0.15.0,<0.17.0; platform_system != "Windows" and platform_system != "Darwin"  # strict
 bitsandbytes >=0.45.2,<0.47.0; platform_system != "Darwin"