Bump pytorch-lightning[extra] from 1.9.4 to 2.0.0 in /requirements

.pre-commit-config.yaml

@@ -34,5 +34,5 @@ repos:
         hooks:
         -   id: mypy
             args: [--strict, --ignore-missing-imports, --show-error-codes]
-            additional_dependencies: [torch>=1.13, torchmetrics>=0.10, pytorch-lightning>=1.7, pytest>=6, pyvista>=0.20, omegaconf>=2.1, kornia>=0.6, numpy>=1.22.0]
+            additional_dependencies: [torch>=1.13, torchmetrics>=0.10, lightning>=1.8, pytest>=6, pyvista>=0.20, omegaconf>=2.1, kornia>=0.6, numpy>=1.22.0]
             exclude: (build|data|dist|logo|logs|output)/

README.md

@@ -39,7 +39,7 @@ The following sections give basic examples of what you can do with TorchGeo.
 First we'll import various classes and functions used in the following sections:
 
 ```python
-from pytorch_lightning import Trainer
+from lightning import Trainer
 from torch.utils.data import DataLoader
 from torchgeo.datamodules import InriaAerialImageLabelingDataModule
 from torchgeo.datasets import CDL, Landsat7, Landsat8, VHR10, stack_samples
@@ -118,9 +118,9 @@ for batch in dataloader:
 
 All TorchGeo datasets are compatible with PyTorch data loaders, making them easy to integrate into existing training workflows. The only difference between a benchmark dataset in TorchGeo and a similar dataset in torchvision is that each dataset returns a dictionary with keys for each PyTorch `Tensor`.
 
-### Reproducibility with PyTorch Lightning
+### Reproducibility with Lightning
 
-In order to facilitate direct comparisons between results published in the literature and further reduce the boilerplate code needed to run experiments with datasets in TorchGeo, we have created PyTorch Lightning [*datamodules*](https://torchgeo.readthedocs.io/en/stable/api/datamodules.html) with well-defined train-val-test splits and [*trainers*](https://torchgeo.readthedocs.io/en/stable/api/trainers.html) for various tasks like classification, regression, and semantic segmentation. These datamodules show how to incorporate augmentations from the kornia library, include preprocessing transforms (with pre-calculated channel statistics), and let users easily experiment with hyperparameters related to the data itself (as opposed to the modeling process). Training a semantic segmentation model on the [Inria Aerial Image Labeling](https://project.inria.fr/aerialimagelabeling/) dataset is as easy as a few imports and four lines of code.
+In order to facilitate direct comparisons between results published in the literature and further reduce the boilerplate code needed to run experiments with datasets in TorchGeo, we have created Lightning [*datamodules*](https://torchgeo.readthedocs.io/en/stable/api/datamodules.html) with well-defined train-val-test splits and [*trainers*](https://torchgeo.readthedocs.io/en/stable/api/trainers.html) for various tasks like classification, regression, and semantic segmentation. These datamodules show how to incorporate augmentations from the kornia library, include preprocessing transforms (with pre-calculated channel statistics), and let users easily experiment with hyperparameters related to the data itself (as opposed to the modeling process). Training a semantic segmentation model on the [Inria Aerial Image Labeling](https://project.inria.fr/aerialimagelabeling/) dataset is as easy as a few imports and four lines of code.
 
 ```python
 datamodule = InriaAerialImageLabelingDataModule(root="...", batch_size=64, num_workers=6)

benchmark.py

@@ -7,7 +7,7 @@
 import os
 import time
 
-import pytorch_lightning as pl
+import lightning as pl
 import torch
 import torch.nn as nn
 import torch.optim as optim

docs/conf.py

@@ -109,7 +109,7 @@
     "matplotlib": ("https://matplotlib.org/stable/", None),
     "numpy": ("https://numpy.org/doc/stable/", None),
     "python": ("https://docs.python.org/3", None),
-    "pytorch-lightning": ("https://pytorch-lightning.readthedocs.io/en/stable/", None),
+    "lightning": ("https://lightning.ai/docs/pytorch/stable/", None),
     "pyvista": ("https://docs.pyvista.org/", None),
     "rasterio": ("https://rasterio.readthedocs.io/en/stable/", None),
     "rtree": ("https://rtree.readthedocs.io/en/stable/", None),

docs/tutorials/pretrained_weights.ipynb

@@ -64,10 +64,10 @@
     "\n",
     "import matplotlib.pyplot as plt\n",
     "import numpy as np\n",
-    "import pytorch_lightning as pl\n",
     "import timm\n",
-    "from pytorch_lightning.callbacks import EarlyStopping, ModelCheckpoint\n",
-    "from pytorch_lightning.loggers import CSVLogger\n",
+    "from lightning import Trainer\n",
+    "from lightning.callbacks import EarlyStopping, ModelCheckpoint\n",
+    "from lightning.loggers import CSVLogger\n",
     "\n",
     "from torchgeo.datamodules import EuroSATDataModule\n",
     "from torchgeo.trainers import ClassificationTask\n",
@@ -92,7 +92,7 @@
    "source": [
     "## Datamodule\n",
     "\n",
-    "We will utilize TorchGeo's datamodules from [PyTorch Lightning](https://pytorch-lightning.readthedocs.io/) to organize the dataloader setup."
+    "We will utilize TorchGeo's [Lightning](https://lightning.ai/docs/pytorch/stable/) datamodules to organize the dataloader setup."
    ]
   },
   {
@@ -178,7 +178,7 @@
    "source": [
     "## Training\n",
     "\n",
-    "To train our pretrained model on the EuroSAT dataset we will make use of PyTorch Lightning's [Trainer](https://pytorch-lightning.readthedocs.io/en/latest/common/trainer.html). For a more elaborate explanation of how TorchGeo uses PyTorch Lightning, check out [this tutorial](https://torchgeo.readthedocs.io/en/stable/tutorials/trainers.html)."
+    "To train our pretrained model on the EuroSAT dataset we will make use of Lightning's [Trainer](https://lightning.ai/docs/pytorch/stable/common/trainer.html). For a more elaborate explanation of how TorchGeo uses Lightning, check out [this tutorial](https://torchgeo.readthedocs.io/en/stable/tutorials/trainers.html)."
    ]
   },
   {
@@ -204,7 +204,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "trainer = pl.Trainer(\n",
+    "trainer = Trainer(\n",
     "    callbacks=[checkpoint_callback, early_stopping_callback],\n",
     "    logger=[csv_logger],\n",
     "    default_root_dir=experiment_dir,\n",

docs/tutorials/trainers.ipynb

@@ -17,7 +17,7 @@
     "id": "NdrXRgjU7Zih"
    },
    "source": [
-    "# PyTorch Lightning Trainers\n",
+    "# Lightning Trainers\n",
     "\n",
     "In this tutorial, we demonstrate TorchGeo trainers to train and test a model. Specifically, we use the [Tropical Cyclone dataset](https://torchgeo.readthedocs.io/en/latest/api/datasets.html#tropical-cyclone) and train models to predict cyclone wind speed given imagery of the cyclone.\n",
     "\n",
@@ -74,10 +74,10 @@
     "\n",
     "import numpy as np\n",
     "import matplotlib.pyplot as plt\n",
+    "from lightning import Trainer\n",
+    "from lightning.callbacks import EarlyStopping, ModelCheckpoint\n",
+    "from lightning.loggers import CSVLogger\n",
     "\n",
-    "import pytorch_lightning as pl\n",
-    "from pytorch_lightning.callbacks import EarlyStopping, ModelCheckpoint\n",
-    "from pytorch_lightning.loggers import CSVLogger\n",
     "from torchgeo.datamodules import TropicalCycloneDataModule\n",
     "from torchgeo.trainers import RegressionTask"
    ]
@@ -103,7 +103,7 @@
    "source": [
     "## Lightning modules\n",
     "\n",
-    "Our trainers use [PyTorch Lightning](https://pytorch-lightning.readthedocs.io/) to organize both the training code, and the dataloader setup code. This makes it easy to create and share reproducible experiments and results.\n",
+    "Our trainers use [Lightning](https://lightning.ai/docs/pytorch/stable/) to organize both the training code, and the dataloader setup code. This makes it easy to create and share reproducible experiments and results.\n",
     "\n",
     "First we'll create a `TropicalCycloneDataModule` object which is simply a wrapper around the [TropicalCyclone](https://torchgeo.readthedocs.io/en/latest/api/datasets.html#tropical-cyclone) dataset. This object 1.) ensures that the data is downloaded*, 2.) sets up PyTorch `DataLoader` objects for the train, validation, and test splits, and 3.) ensures that data from the same cyclone **is not** shared between the training and validation sets so that you can properly evaluate the generalization performance of your model.\n",
     "\n",
@@ -178,7 +178,7 @@
    "source": [
     "## Training\n",
     "\n",
-    "Now that we have the Lightning modules set up, we can use a PyTorch Lightning [Trainer](https://pytorch-lightning.readthedocs.io/en/latest/common/trainer.html) to run the training and evaluation loops. There are many useful pieces of configuration that can be set in the `Trainer` -- below we set up model checkpointing based on the validation loss, early stopping based on the validation loss, and a CSV based logger. We encourage you to see the [PyTorch Lightning docs](https://pytorch-lightning.readthedocs.io/) for other options that can be set here, e.g. Tensorboard logging, automatically selecting your optimizer's learning rate, and easy multi-GPU training."
+    "Now that we have the Lightning modules set up, we can use a Lightning [Trainer](https://lightning.ai/docs/pytorch/stable/common/trainer.html) to run the training and evaluation loops. There are many useful pieces of configuration that can be set in the `Trainer` -- below we set up model checkpointing based on the validation loss, early stopping based on the validation loss, and a CSV based logger. We encourage you to see the [Lightning docs](https://lightning.ai/docs/pytorch/stable/) for other options that can be set here, e.g. Tensorboard logging, automatically selecting your optimizer's learning rate, and easy multi-GPU training."
    ]
   },
   {
@@ -226,7 +226,7 @@
     }
    ],
    "source": [
-    "trainer = pl.Trainer(\n",
+    "trainer = Trainer(\n",
     "    callbacks=[checkpoint_callback, early_stopping_callback],\n",
     "    logger=[csv_logger],\n",
     "    default_root_dir=experiment_dir,\n",

environment.yml

@@ -25,6 +25,7 @@ dependencies:
     - isort[colors]>=5.8
     - kornia>=0.6.5
     - laspy>=2
+    - lightning>=1.8
     - mypy>=0.900
     - nbmake>=0.1
     - nbsphinx>=0.8.5
@@ -35,10 +36,9 @@ dependencies:
     - pydocstyle[toml]>=6.1
     - pytest>=6.1.2
     - pytest-cov>=2.4
-    - pytorch-lightning>=1.5.1
     - git+https://github.com/pytorch/pytorch_sphinx_theme
     - pyupgrade>=2.4
-    - radiant-mlhub>=0.2.1
+    - radiant-mlhub>=0.3
     - rtree>=1
     - scikit-image>=0.18
     - scikit-learn>=0.22

evaluate.py

@@ -10,7 +10,7 @@
 import os
 from typing import Any, Dict, Union, cast
 
-import pytorch_lightning as pl
+import lightning as pl
 import torch
 from torchmetrics import MetricCollection
 from torchmetrics.classification import BinaryAccuracy, BinaryJaccardIndex
@@ -146,7 +146,7 @@ def main(args: argparse.Namespace) -> None:
     model.freeze()
     model.eval()
 
-    dm = DATAMODULE(  # type: ignore[call-arg]
+    dm = DATAMODULE(
         seed=args.seed,
         root=args.root,
         num_workers=args.num_workers,

experiments/run_chesapeakecvpr_models.py

@@ -8,8 +8,7 @@
 import csv
 import os
 
-import pytorch_lightning as pl
-import torch
+from lightning import Trainer
 
 from torchgeo.datamodules import ChesapeakeCVPRDataModule
 from torchgeo.trainers.chesapeake import SemanticSegmentationTask
@@ -86,8 +85,9 @@ def main(args: argparse.Namespace) -> None:
         writer.writeheader()
 
     # Test loop
-    trainer = pl.Trainer(
-        gpus=[args.device] if torch.cuda.is_available() else None,
+    trainer = Trainer(
+        accelerator="auto",
+        devices=[args.device],
         logger=False,
         enable_progress_bar=False,
         enable_checkpointing=False,

pyproject.toml

@@ -73,9 +73,9 @@ filterwarnings = [
     # https://github.com/Lightning-AI/lightning/pull/13261
     "ignore:torch.distributed._sharded_tensor will be deprecated:DeprecationWarning:torch.distributed._sharded_tensor",
     # https://github.com/Lightning-AI/lightning/issues/13989
-    "ignore:SelectableGroups dict interface is deprecated. Use select.:DeprecationWarning:pytorch_lightning.trainer.connectors.callback_connector",
+    "ignore:SelectableGroups dict interface is deprecated. Use select.:DeprecationWarning:lightning.pytorch.trainer.connectors.callback_connector",
     # https://github.com/Lightning-AI/lightning/issues/14594
-    "ignore:To copy construct from a tensor, it is recommended to use:UserWarning:pytorch_lightning.core.module",
+    "ignore:To copy construct from a tensor, it is recommended to use:UserWarning:lightning.pytorch.core.module",
     # https://github.com/rasterio/rasterio/issues/1742
     # https://github.com/rasterio/rasterio/pull/1753
     "ignore:Using or importing the ABCs from 'collections' instead of from 'collections.abc' is deprecated:DeprecationWarning:rasterio.crs",
@@ -94,21 +94,20 @@ filterwarnings = [
     # https://github.com/lanpa/tensorboardX/pull/677
     "ignore:ANTIALIAS is deprecated and will be removed in Pillow 10:DeprecationWarning:tensorboardX.summary",
     # https://github.com/Lightning-AI/lightning/issues/16756
-    "ignore:Deprecated call to `pkg_resources.declare_namespace:DeprecationWarning:lightning_fabric",
-    "ignore:Deprecated call to `pkg_resources.declare_namespace:DeprecationWarning:pytorch_lightning",
+    "ignore:Deprecated call to `pkg_resources.declare_namespace:DeprecationWarning",
 
     # Expected warnings
-    # pytorch-lightning warns us about using num_workers=0, but it's faster on macOS
+    # Lightning warns us about using num_workers=0, but it's faster on macOS
     "ignore:The dataloader, .*, does not have many workers which may be a bottleneck:UserWarning",
-    # pytorch-lightning warns us about using the CPU when a GPU is available
+    # Lightning warns us about using the CPU when a GPU is available
     "ignore:GPU available but not used.:UserWarning",
     # https://github.com/kornia/kornia/pull/1611
     "ignore:`ColorJitter` is now following Torchvision implementation.:DeprecationWarning:kornia.augmentation._2d.intensity.color_jitter",
     # https://github.com/kornia/kornia/pull/1663
     "ignore:`RandomGaussianBlur` has changed its behavior and now randomly sample sigma for both axes.:DeprecationWarning:kornia.augmentation._2d.intensity.gaussian_blur",
 
     # Unexpected warnings, worth investigating
-    # pytorch-lightning is having trouble inferring the batch size for ChesapeakeCVPRDataModule and CycloneDataModule for some reason
+    # Lightning is having trouble inferring the batch size for ChesapeakeCVPRDataModule and CycloneDataModule for some reason
     "ignore:Trying to infer the `batch_size` from an ambiguous collection:UserWarning",
 ]
 markers = [

requirements/min-reqs.old

@@ -5,12 +5,12 @@ setuptools==42.0.0
 einops==0.3.0
 fiona==1.8.12
 kornia==0.6.5
+lightning==1.8.0
 matplotlib==3.3.0
 numpy==1.17.3
 omegaconf==2.1.0
 pillow==6.2.1
 pyproj==2.4.1
-pytorch-lightning==1.5.1
 rasterio==1.1.1
 rtree==1.0.0
 scikit-learn==0.22
@@ -28,7 +28,7 @@ opencv-python==4.1.2.30
 pandas==0.25.2
 pycocotools==2.0.1
 pyvista==0.25.2
-radiant-mlhub==0.2.1
+radiant-mlhub==0.3.0
 rarfile==4.0
 scikit-image==0.18.0
 scipy==1.6.2

requirements/required.txt

@@ -5,12 +5,12 @@ setuptools==67.6.0
 einops==0.6.0
 fiona==1.9.1
 kornia==0.6.10
+lightning[extra]==2.0.0
 matplotlib==3.7.1
 numpy==1.24.2
 omegaconf==2.3.0
 pillow==9.4.0
 pyproj==3.4.1
-pytorch-lightning[extra]==1.9.4
 rasterio==1.3.6
 rtree==1.0.1
 scikit-learn==1.2.2

setup.cfg

@@ -30,6 +30,8 @@ install_requires =
     fiona>=1.8.12,<2
     # kornia 0.6.5+ required due to change in kornia.augmentation API
     kornia>=0.6.5,<0.7
+    # lightning 1.8+ is first release
+    lightning[extra]>=1.8,<2
     # matplotlib 3.3+ required for (H, W, 1) image support in plt.imshow
     matplotlib>=3.3,<4
     # numpy 1.17.3+ required by Python 3.8 wheels
@@ -40,8 +42,6 @@ install_requires =
     pillow>=6.2.1,<10
     # pyproj 2.4.1+ required for Python 3.8 wheels
     pyproj>=2.4.1,<4
-    # pytorch-lightning 1.5.1+ required for apply_to_collection bugfix
-    pytorch-lightning[extra]>=1.5.1,<2
     # rasterio 1.1.1+ required for Python 3.8 wheels
     rasterio>=1.1.1,<2
     # rtree 1+ required for len(index), index & index, index | index
@@ -83,9 +83,8 @@ datasets =
     pycocotools>=2.0.1,<3
     # pyvista 0.25.2 required for wheels
     pyvista>=0.25.2,<0.39
-    # radiant-mlhub 0.2.1+ required for api_key bugfix:
-    # https://github.com/radiantearth/radiant-mlhub/pull/48
-    radiant-mlhub>=0.2.1,<0.6
+    # radiant-mlhub 0.3+ required for newer tqdm support required by lightning
+    radiant-mlhub>=0.3,<0.6
     # rarfile 4+ required for wheels
     rarfile>=4,<5
     # scikit-image 0.18+ required for numpy 1.17+ compatibility

tests/datamodules/test_oscd.py

@@ -5,7 +5,7 @@
 
 import pytest
 from _pytest.fixtures import SubRequest
-from pytorch_lightning import Trainer
+from lightning import Trainer
 
 from torchgeo.datamodules import OSCDDataModule
 
@@ -30,7 +30,7 @@ def datamodule(self, request: SubRequest) -> OSCDDataModule:
 
     def test_train_dataloader(self, datamodule: OSCDDataModule) -> None:
         datamodule.setup("fit")
-        datamodule.trainer.training = True  # type: ignore[union-attr]
+        datamodule.trainer.training = True
         batch = next(iter(datamodule.train_dataloader()))
         batch = datamodule.on_after_batch_transfer(batch, 0)
         assert batch["image"].shape[-2:] == batch["mask"].shape[-2:] == (2, 2)
@@ -42,7 +42,7 @@ def test_train_dataloader(self, datamodule: OSCDDataModule) -> None:
 
     def test_val_dataloader(self, datamodule: OSCDDataModule) -> None:
         datamodule.setup("validate")
-        datamodule.trainer.validating = True  # type: ignore[union-attr]
+        datamodule.trainer.validating = True
         batch = next(iter(datamodule.val_dataloader()))
         batch = datamodule.on_after_batch_transfer(batch, 0)
         if datamodule.val_split_pct > 0.0:
@@ -55,7 +55,7 @@ def test_val_dataloader(self, datamodule: OSCDDataModule) -> None:
 
     def test_test_dataloader(self, datamodule: OSCDDataModule) -> None:
         datamodule.setup("test")
-        datamodule.trainer.testing = True  # type: ignore[union-attr]
+        datamodule.trainer.testing = True
         batch = next(iter(datamodule.test_dataloader()))
         batch = datamodule.on_after_batch_transfer(batch, 0)
         assert batch["image"].shape[-2:] == batch["mask"].shape[-2:] == (2, 2)

tests/trainers/test_byol.py

@@ -12,8 +12,8 @@
 import torchvision
 from _pytest.fixtures import SubRequest
 from _pytest.monkeypatch import MonkeyPatch
+from lightning import LightningDataModule, Trainer
 from omegaconf import OmegaConf
-from pytorch_lightning import LightningDataModule, Trainer
 from torchvision.models import resnet18
 from torchvision.models._api import WeightsEnum
 

tests/trainers/test_classification.py

@@ -12,8 +12,8 @@
 import torchvision
 from _pytest.fixtures import SubRequest
 from _pytest.monkeypatch import MonkeyPatch
+from lightning import LightningDataModule, Trainer
 from omegaconf import OmegaConf
-from pytorch_lightning import LightningDataModule, Trainer
 from torch.nn.modules import Module
 from torchvision.models._api import WeightsEnum
 

tests/trainers/test_detection.py

@@ -9,8 +9,8 @@
 import torch.nn as nn
 import torchvision.models.detection
 from _pytest.monkeypatch import MonkeyPatch
+from lightning import LightningDataModule, Trainer
 from omegaconf import OmegaConf
-from pytorch_lightning import LightningDataModule, Trainer
 from torch.nn.modules import Module
 
 from torchgeo.datamodules import MisconfigurationException, NASAMarineDebrisDataModule

tests/trainers/test_regression.py

@@ -11,8 +11,8 @@
 import torchvision
 from _pytest.fixtures import SubRequest
 from _pytest.monkeypatch import MonkeyPatch
+from lightning import LightningDataModule, Trainer
 from omegaconf import OmegaConf
-from pytorch_lightning import LightningDataModule, Trainer
 from torchvision.models._api import WeightsEnum
 
 from torchgeo.datamodules import (

tests/trainers/test_segmentation.py

@@ -9,8 +9,8 @@
 import torch
 import torch.nn as nn
 from _pytest.monkeypatch import MonkeyPatch
+from lightning import LightningDataModule, Trainer
 from omegaconf import OmegaConf
-from pytorch_lightning import LightningDataModule, Trainer
 from torch.nn.modules import Module
 
 from torchgeo.datamodules import (