Merge branch 'master' into feature/13324_validation-interval

Sohaib-Ahmed21 · web-flow · commit 44e88a42154e · 2025-08-26T12:05:51.000-07:00
diff --git a/.github/workflows/ci-tests-fabric.yml b/.github/workflows/ci-tests-fabric.yml
@@ -96,7 +96,7 @@ jobs:
         run: |
           cd requirements/fabric
           pip install -U "lightning-utilities[cli]"
-          python -m lightning_utilities.cli requirements set-oldest --req_files "['base.txt', 'strategies.txt']"
+          python -m lightning_utilities.cli requirements set-oldest --req_files "['base.txt', 'strategies.txt', 'test.txt']"
           pip install "cython<3.0" wheel
           pip install "pyyaml==5.4" --no-build-isolation
 
@@ -142,7 +142,8 @@ jobs:
         run: |
           pip install -e ".[${EXTRA_PREFIX}test,${EXTRA_PREFIX}strategies]" \
             -U --upgrade-strategy=eager --prefer-binary \
-            --extra-index-url="${TORCH_URL}" --find-links="${PYPI_CACHE_DIR}"
+            --extra-index-url="${TORCH_URL}" \
+            --find-links="${PYPI_CACHE_DIR}"
           pip list
       - name: Dump handy wheels
         if: github.event_name == 'push' && github.ref == 'refs/heads/master'
diff --git a/.github/workflows/ci-tests-pytorch.yml b/.github/workflows/ci-tests-pytorch.yml
@@ -101,7 +101,7 @@ jobs:
         run: |
           cd requirements/pytorch
           pip install -U "lightning-utilities[cli]"
-          python -m lightning_utilities.cli requirements set-oldest --req_files "['base.txt', 'extra.txt', 'strategies.txt', 'examples.txt']"
+          python -m lightning_utilities.cli requirements set-oldest --req_files "['base.txt', 'extra.txt', 'strategies.txt', 'examples.txt', 'test.txt']"
           pip install "cython<3.0" wheel
           pip install "pyyaml==5.4" --no-build-isolation
 
@@ -141,7 +141,9 @@ jobs:
           pip install ".[${EXTRA_PREFIX}extra,${EXTRA_PREFIX}test,${EXTRA_PREFIX}strategies]" \
             -U --upgrade-strategy=eager --prefer-binary \
             -r requirements/_integrations/accelerators.txt \
-            --extra-index-url="${TORCH_URL}" --find-links="${PYPI_CACHE_DIR}" --find-links="https://download.pytorch.org/whl/torch-tensorrt"
+            --extra-index-url="${TORCH_URL}" \
+            --find-links="${PYPI_CACHE_DIR}" \
+            --find-links="https://download.pytorch.org/whl/torch-tensorrt"
           pip list
       - name: Drop LAI from extensions
         if: ${{ matrix.pkg-name != 'lightning' }}
diff --git a/docs/source-pytorch/conf.py b/docs/source-pytorch/conf.py
@@ -127,7 +127,7 @@ def _load_py_module(name: str, location: str) -> ModuleType:
         "https://pytorch.org/docs/stable/", "https://pytorch.org/docs/{torch.__version__}/", _PATH_ROOT
     )
     adjust_linked_external_docs(
-        "https://lightning.ai/docs/torchmetrics", "https://lightning.ai/docs/torchmetrics/v{torchmetrics.__version__}/", _PATH_ROOT, version_digits=3
+        "https://lightning.ai/docs/torchmetrics/stable/", "https://lightning.ai/docs/torchmetrics/v{torchmetrics.__version__}/", _PATH_ROOT, version_digits=3
     )
     adjust_linked_external_docs(
         "https://lightning.ai/docs/fabric/stable/", "https://lightning.ai/docs/fabric/{lightning_fabric.__version__}/", _PATH_ROOT, version_digits=3
diff --git a/docs/source-pytorch/extensions/logging.rst b/docs/source-pytorch/extensions/logging.rst
@@ -120,6 +120,10 @@ methods to log from anywhere in a :doc:`LightningModule <../common/lightning_mod
 .. note::
     Everything explained below applies to both :meth:`~lightning.pytorch.core.LightningModule.log` or :meth:`~lightning.pytorch.core.LightningModule.log_dict` methods.
 
+.. note::
+
+    When using TorchMetrics with Lightning, we recommend referring to the `TorchMetrics Lightning integration documentation <https://lightning.ai/docs/torchmetrics/stable/pages/lightning.html>`_ for logging best practices, common pitfalls, and proper usage patterns.
+
 Depending on where the :meth:`~lightning.pytorch.core.LightningModule.log` method is called, Lightning auto-determines
 the correct logging mode for you. Of course you can override the default behavior by manually setting the
 :meth:`~lightning.pytorch.core.LightningModule.log` parameters.
diff --git a/requirements/docs.txt b/requirements/docs.txt
@@ -1,5 +1,5 @@
 sphinx >5.0, <6.0
-myst-parser >=0.18.1, <4.0.0
+myst-parser >=0.18.1, <5.0.0
 nbsphinx >=0.8.5, <=0.9.7
 nbconvert >7.14, <7.17
 pandoc >=1.0, <=2.4
diff --git a/requirements/fabric/test.txt b/requirements/fabric/test.txt
@@ -1,10 +1,10 @@
-coverage ==7.10.4
-numpy >=1.17.2, <1.27.0
+coverage ==7.10.5
+numpy >=1.21.0, <1.27.0
 pytest ==8.4.1
 pytest-cov ==6.2.1
 pytest-timeout ==2.4.0
 pytest-rerunfailures ==15.1
 pytest-random-order ==1.2.0
 click ==8.1.8; python_version < "3.11"
 click ==8.2.1; python_version > "3.10"
-tensorboardX >=2.2, <2.7.0  # min version is set by torch.onnx missing attribute
+tensorboardX >=2.6, <2.7.0  # todo: relax it back to `>=2.2` after fixing tests
diff --git a/requirements/pytorch/docs.txt b/requirements/pytorch/docs.txt
@@ -4,6 +4,6 @@ nbformat  # used for generate empty notebook
 ipython[notebook] <9.5.0
 setuptools<81.0  # workaround for `error in ipython setup command: use_2to3 is invalid.`
 
-onnxscript >= 0.2.2, <0.4.0
+onnxscript >= 0.2.2, < 0.5.0
 
 #-r ../../_notebooks/.actions/requires.txt
diff --git a/requirements/pytorch/loggers.info b/requirements/pytorch/loggers.info
@@ -4,4 +4,4 @@ neptune >=1.0.0
 comet-ml >=3.31.0
 mlflow >=1.0.0
 wandb >=0.12.10
-tensorboard >=2.9.1
+tensorboard >=2.11
diff --git a/requirements/pytorch/test.txt b/requirements/pytorch/test.txt
@@ -1,4 +1,4 @@
-coverage ==7.10.4
+coverage ==7.10.5
 pytest ==8.4.1
 pytest-cov ==6.2.1
 pytest-timeout ==2.4.0
@@ -8,16 +8,16 @@ pytest-random-order ==1.2.0
 # needed in tests
 cloudpickle >=1.3, <3.2.0
 scikit-learn >0.22.1, <1.8.0
-numpy >=1.17.2, <1.27.0
-onnx >=1.12.0, <1.19.0
+numpy >1.20.0, <1.27.0
+onnx >1.12.0, <1.19.0
 onnxruntime >=1.12.0, <1.23.0
-onnxscript >= 0.2.2, <0.4.0
+onnxscript >= 0.1.0, < 0.5.0
 psutil <7.0.1 # for `DeviceStatsMonitor`
 pandas >2.0, <2.4.0  # needed in benchmarks
 fastapi  # for `ServableModuleValidator`  # not setting version as re-defined in App
 uvicorn  # for `ServableModuleValidator`  # not setting version as re-defined in App
 
-tensorboard >=2.9.1, <2.21.0  # for `TensorBoardLogger`
+tensorboard >=2.11, <2.21.0  # for `TensorBoardLogger`
 
 --find-links https://download.pytorch.org/whl/torch-tensorrt
 torch-tensorrt; platform_system == "Linux" and python_version >= "3.12"
diff --git a/src/lightning/pytorch/CHANGELOG.md b/src/lightning/pytorch/CHANGELOG.md
@@ -37,6 +37,11 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 - Fixed learning rate not being correctly set after using `LearningRateFinder` callback ([#21068](https://github.com/Lightning-AI/pytorch-lightning/pull/21068))
 
 
+- Fixed misalignment column while using rich model summary in `DeepSpeedstrategy` ([#21100](https://github.com/Lightning-AI/pytorch-lightning/pull/21100))
+
+
+- Fixed `RichProgressBar` crashing when sanity checking using val dataloader with 0 len ([#21108](https://github.com/Lightning-AI/pytorch-lightning/pull/21108))
+
 ---
 
 ## [2.5.3] - 2025-08-13
diff --git a/src/lightning/pytorch/callbacks/progress/rich_progress.py b/src/lightning/pytorch/callbacks/progress/rich_progress.py
@@ -390,8 +390,7 @@ def on_sanity_check_start(self, trainer: "pl.Trainer", pl_module: "pl.LightningM
 
     @override
     def on_sanity_check_end(self, trainer: "pl.Trainer", pl_module: "pl.LightningModule") -> None:
-        if self.progress is not None:
-            assert self.val_sanity_progress_bar_id is not None
+        if self.progress is not None and self.val_sanity_progress_bar_id is not None:
             self.progress.update(self.val_sanity_progress_bar_id, advance=0, visible=False)
         self.refresh()
 
diff --git a/src/lightning/pytorch/callbacks/rich_model_summary.py b/src/lightning/pytorch/callbacks/rich_model_summary.py
@@ -79,18 +79,21 @@ def summarize(
         from rich.table import Table
 
         console = get_console()
+        column_names = list(zip(*summary_data))[0]
 
         header_style: str = summarize_kwargs.get("header_style", "bold magenta")
         table = Table(header_style=header_style)
         table.add_column(" ", style="dim")
         table.add_column("Name", justify="left", no_wrap=True)
         table.add_column("Type")
         table.add_column("Params", justify="right")
+
+        if "Params per Device" in column_names:
+            table.add_column("Params per Device", justify="right")
+
         table.add_column("Mode")
         table.add_column("FLOPs", justify="right")
 
-        column_names = list(zip(*summary_data))[0]
-
         for column_name in ["In sizes", "Out sizes"]:
             if column_name in column_names:
                 table.add_column(column_name, justify="right", style="white")
diff --git a/src/lightning/pytorch/trainer/connectors/logger_connector/logger_connector.py b/src/lightning/pytorch/trainer/connectors/logger_connector/logger_connector.py
@@ -93,8 +93,9 @@ def log_metrics(self, metrics: _OUT_DICT, step: Optional[int] = None) -> None:
 
         Args:
             metrics: Metric values
-            step: Step for which metrics should be logged. Default value is `self.global_step` during training or
-                the total validation / test log step count during validation and testing.
+            step: Step for which metrics should be logged. If a `step` metric is logged, this value will
+                be used else will default to `self.global_step` during training or the total log step count
+                during validation and testing.
 
         """
         if not self.trainer.loggers or not metrics:
diff --git a/src/lightning/pytorch/utilities/model_summary/model_summary_deepspeed.py b/src/lightning/pytorch/utilities/model_summary/model_summary_deepspeed.py
@@ -99,6 +99,7 @@ def _get_summary_data(self) -> list[tuple[str, list[str]]]:
             ("Params", list(map(get_human_readable_count, self.param_nums))),
             ("Params per Device", list(map(get_human_readable_count, self.parameters_per_layer))),
             ("Mode", ["train" if mode else "eval" for mode in self.training_modes]),
+            ("FLOPs", list(map(get_human_readable_count, (sum(x.values()) for x in self.flop_counts.values())))),
         ]
         if self._model.example_input_array is not None:
             arrays.append(("In sizes", [str(x) for x in self.in_sizes]))
diff --git a/tests/tests_pytorch/callbacks/progress/test_rich_progress_bar.py b/tests/tests_pytorch/callbacks/progress/test_rich_progress_bar.py
@@ -577,3 +577,31 @@ def test_rich_progress_bar_metrics_theme_update(*_):
     theme = RichProgressBar(theme=RichProgressBarTheme(metrics_format=".3e", metrics_text_delimiter="\n")).theme
     assert theme.metrics_format == ".3e"
     assert theme.metrics_text_delimiter == "\n"
+
+
+@RunIf(rich=True)
+def test_rich_progress_bar_empty_val_dataloader_model(tmp_path):
+    """Test that RichProgressBar doesn't crash with empty val_dataloader list from model."""
+
+    class EmptyListModel(BoringModel):
+        def train_dataloader(self):
+            return DataLoader(RandomDataset(32, 64), batch_size=2)
+
+        def val_dataloader(self):
+            return []
+
+    model = EmptyListModel()
+    progress_bar = RichProgressBar()
+
+    trainer = Trainer(
+        default_root_dir=tmp_path,
+        max_epochs=1,
+        num_sanity_val_steps=1,
+        callbacks=[progress_bar],
+        limit_train_batches=2,
+        enable_checkpointing=False,
+        logger=False,
+    )
+
+    # This should not raise an AssertionError
+    trainer.fit(model)
diff --git a/tests/tests_pytorch/utilities/test_deepspeed_model_summary.py b/tests/tests_pytorch/utilities/test_deepspeed_model_summary.py
@@ -12,6 +12,10 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
+from unittest import mock
+
+import torch
+
 import lightning.pytorch as pl
 from lightning.pytorch import Callback, Trainer
 from lightning.pytorch.demos.boring_classes import BoringModel
@@ -51,3 +55,38 @@ def on_fit_start(self, trainer: "pl.Trainer", pl_module: "pl.LightningModule") -
     )
 
     trainer.fit(model)
+
+
+@RunIf(min_cuda_gpus=1, deepspeed=True, rich=True)
+@mock.patch("rich.table.Table.add_row", autospec=True)
+def test_deepspeed_summary_with_rich_model_summary(mock_table_add_row, tmp_path):
+    from lightning.pytorch.callbacks import RichModelSummary
+
+    model = BoringModel()
+    model.example_input_array = torch.randn(4, 32)
+
+    trainer = Trainer(
+        strategy=DeepSpeedStrategy(stage=3),
+        default_root_dir=tmp_path,
+        accelerator="gpu",
+        fast_dev_run=True,
+        devices=1,
+        enable_model_summary=True,
+        callbacks=[RichModelSummary()],
+    )
+
+    trainer.fit(model)
+
+    # assert that the input summary data was converted correctly
+    args, _ = mock_table_add_row.call_args_list[0]
+    assert args[1:] == (
+        "0",
+        "layer",
+        "Linear",
+        "66  ",
+        "66  ",
+        "train",
+        "512  ",
+        "[4, 32]",
+        "[4, 2]",
+    )

Original file line number	Diff line number	Diff line change
`@@ -127,7 +127,7 @@ def _load_py_module(name: str, location: str) -> ModuleType:`
`127`	`127`	`"https://pytorch.org/docs/stable/", "https://pytorch.org/docs/{torch.__version__}/", _PATH_ROOT`
`128`	`128`	`)`
`129`	`129`	`adjust_linked_external_docs(`
`130`		`- "https://lightning.ai/docs/torchmetrics", "https://lightning.ai/docs/torchmetrics/v{torchmetrics.__version__}/", _PATH_ROOT, version_digits=3`
	`130`	`+ "https://lightning.ai/docs/torchmetrics/stable/", "https://lightning.ai/docs/torchmetrics/v{torchmetrics.__version__}/", _PATH_ROOT, version_digits=3`
`131`	`131`	`)`
`132`	`132`	`adjust_linked_external_docs(`
`133`	`133`	`"https://lightning.ai/docs/fabric/stable/", "https://lightning.ai/docs/fabric/{lightning_fabric.__version__}/", _PATH_ROOT, version_digits=3`