Bump torch minimum to mitigate CVE-2024-31580 & CVE-2024-31583 and enable numpy 2 compatibility

.github/workflows/cron.yml

@@ -13,17 +13,13 @@ jobs:
     strategy:
       matrix:
         environment:
-          - "PT113+CUDA118"
-          - "PT210+CUDA121"
+          - "PT230+CUDA121"
           - "PT240+CUDA126"
           - "PTLATEST+CUDA126"
         include:
           # https://docs.nvidia.com/deeplearning/frameworks/pytorch-release-notes
-          - environment: PT113+CUDA118
-            pytorch: "torch==1.13.1 torchvision==0.14.1 --extra-index-url https://download.pytorch.org/whl/cu121"
-            base: "nvcr.io/nvidia/pytorch:22.10-py3"  # CUDA 11.8
-          - environment: PT210+CUDA121
-            pytorch: "pytorch==2.1.0 torchvision==0.16.0 --extra-index-url https://download.pytorch.org/whl/cu121"
+          - environment: PT230+CUDA121
+            pytorch: "pytorch==2.3.0 torchvision==0.18.0 --extra-index-url https://download.pytorch.org/whl/cu121"
             base: "nvcr.io/nvidia/pytorch:23.08-py3"  # CUDA 12.1
           - environment: PT240+CUDA126
             pytorch: "pytorch==2.4.0 torchvision==0.19.0 --extra-index-url https://download.pytorch.org/whl/cu121"

.github/workflows/pythonapp-gpu.yml

@@ -22,19 +22,21 @@ jobs:
     strategy:
       matrix:
         environment:
-          - "PT113+CUDA116"
-          - "PT210+CUDA121DOCKER"
+          - "PT230+CUDA124DOCKER"
+          - "PT240+CUDA125DOCKER"
+          - "PT250+CUDA126DOCKER"
         include:
           # https://docs.nvidia.com/deeplearning/frameworks/pytorch-release-notes
-          - environment: PT113+CUDA116
-            pytorch: "torch==1.13.1 torchvision==0.14.1"
-            base: "nvcr.io/nvidia/cuda:11.6.1-devel-ubuntu18.04"
-          - environment: PT210+CUDA121DOCKER
-            # 23.08: 2.1.0a0+29c30b1
+          - environment: PT230+CUDA124DOCKER
+            # 24.04: 2.3.0a0+6ddf5cf85e
             pytorch: "-h"  # we explicitly set pytorch to -h to avoid pip install error
-            base: "nvcr.io/nvidia/pytorch:23.08-py3"
-          - environment: PT210+CUDA121DOCKER
-            # 24.08: 2.3.0a0+40ec155e58.nv24.3
+            base: "nvcr.io/nvidia/pytorch:24.04-py3"
+          - environment: PT240+CUDA125DOCKER
+            # 24.06: 2.4.0a0+f70bd71a48
+            pytorch: "-h"  # we explicitly set pytorch to -h to avoid pip install error
+            base: "nvcr.io/nvidia/pytorch:24.06-py3"
+          - environment: PT250+CUDA126DOCKER
+            # 24.08: 2.5.0a0+872d972e41
             pytorch: "-h"  # we explicitly set pytorch to -h to avoid pip install error
             base: "nvcr.io/nvidia/pytorch:24.08-py3"
     container:
@@ -49,7 +51,7 @@ jobs:
         apt-get update
         apt-get install -y wget
 
-        if [ ${{ matrix.environment }} = "PT113+CUDA116" ]
+        if [ ${{ matrix.environment }} = "PT230+CUDA124" ]
         then
         PYVER=3.9 PYSFX=3 DISTUTILS=python3-distutils && \
         apt-get update && apt-get install -y --no-install-recommends \
@@ -114,7 +116,7 @@ jobs:
         # build for the current self-hosted CI Tesla V100
         BUILD_MONAI=1 TORCH_CUDA_ARCH_LIST="7.0" ./runtests.sh --build --disttests
         ./runtests.sh --quick --unittests
-        if [ ${{ matrix.environment }} = "PT113+CUDA116" ]; then
+        if [ ${{ matrix.environment }} = "PT230+CUDA124" ]; then
           # test the clang-format tool downloading once
           coverage run -m tests.clang_format_utils
         fi

.github/workflows/pythonapp-min.yml

@@ -124,7 +124,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        pytorch-version: ['1.13.1', '2.0.1', '2.2.2', '2.3.1', '2.4.1', 'latest']
+        pytorch-version: ['2.3.1', '2.4.1', '2.5.1', 'latest']
     timeout-minutes: 40
     steps:
     - uses: actions/checkout@v4

.github/workflows/pythonapp.yml

@@ -94,7 +94,7 @@ jobs:
     - if: runner.os == 'windows'
       name: Install torch cpu from pytorch.org (Windows only)
       run: |
-        python -m pip install torch==1.13.1+cpu torchvision==0.14.1+cpu -f https://download.pytorch.org/whl/torch_stable.html
+        python -m pip install torch==2.4.1 torchvision==0.19.1+cpu --index-url https://download.pytorch.org/whl/cpu
     - if: runner.os == 'Linux'
       name: Install itk pre-release (Linux only)
       run: |
@@ -103,7 +103,7 @@ jobs:
     - name: Install the dependencies
       run: |
         python -m pip install --user --upgrade pip wheel
-        python -m pip install torch==1.13.1 torchvision==0.14.1
+        python -m pip install torch==2.4.1 torchvision==0.19.1
         cat "requirements-dev.txt"
         python -m pip install -r requirements-dev.txt
         python -m pip list
@@ -155,7 +155,7 @@ jobs:
         # install the latest pytorch for testing
         # however, "pip install monai*.tar.gz" will build cpp/cuda with an isolated
         # fresh torch installation according to pyproject.toml
-        python -m pip install torch>=1.13.1 torchvision
+        python -m pip install torch>=2.3.0 torchvision
     - name: Check packages
       run: |
         pip uninstall monai

docs/requirements.txt

@@ -1,5 +1,5 @@
--f https://download.pytorch.org/whl/cpu/torch-1.13.1%2Bcpu-cp39-cp39-linux_x86_64.whl
-torch>=1.13.1
+-f https://download.pytorch.org/whl/cpu/torch-2.3.0%2Bcpu-cp39-cp39-linux_x86_64.whl
+torch>=2.3.0
 pytorch-ignite==0.4.11
 numpy>=1.20
 itk>=5.2

environment-dev.yml

@@ -5,8 +5,8 @@ channels:
   - nvidia
   - conda-forge
 dependencies:
-  - numpy>=1.24,<2.0
-  - pytorch>=1.13.1
+  - numpy>=1.24,<3.0
+  - pytorch>=2.3.0
   - torchio
   - torchvision
   - pytorch-cuda>=11.6

monai/engines/evaluator.py

@@ -28,7 +28,7 @@
 from monai.utils import ForwardMode, IgniteInfo, ensure_tuple, min_version, optional_import
 from monai.utils.enums import CommonKeys as Keys
 from monai.utils.enums import EngineStatsKeys as ESKeys
-from monai.utils.module import look_up_option, pytorch_after
+from monai.utils.module import look_up_option
 
 if TYPE_CHECKING:
     from ignite.engine import Engine, EventEnum
@@ -269,13 +269,8 @@ def __init__(
             amp_kwargs=amp_kwargs,
         )
         if compile:
-            if pytorch_after(2, 1):
-                compile_kwargs = {} if compile_kwargs is None else compile_kwargs
-                network = torch.compile(network, **compile_kwargs)  # type: ignore[assignment]
-            else:
-                warnings.warn(
-                    "Network compilation (compile=True) not supported for Pytorch versions before 2.1, no compilation done"
-                )
+            compile_kwargs = {} if compile_kwargs is None else compile_kwargs
+            network = torch.compile(network, **compile_kwargs)  # type: ignore[assignment]
         self.network = network
         self.compile = compile
         self.inferer = SimpleInferer() if inferer is None else inferer

monai/engines/trainer.py

@@ -27,7 +27,6 @@
 from monai.utils import AdversarialIterationEvents, AdversarialKeys, GanKeys, IgniteInfo, min_version, optional_import
 from monai.utils.enums import CommonKeys as Keys
 from monai.utils.enums import EngineStatsKeys as ESKeys
-from monai.utils.module import pytorch_after
 
 if TYPE_CHECKING:
     from ignite.engine import Engine, EventEnum
@@ -183,13 +182,8 @@ def __init__(
             amp_kwargs=amp_kwargs,
         )
         if compile:
-            if pytorch_after(2, 1):
-                compile_kwargs = {} if compile_kwargs is None else compile_kwargs
-                network = torch.compile(network, **compile_kwargs)  # type: ignore[assignment]
-            else:
-                warnings.warn(
-                    "Network compilation (compile=True) not supported for Pytorch versions before 2.1, no compilation done"
-                )
+            compile_kwargs = {} if compile_kwargs is None else compile_kwargs
+            network = torch.compile(network, **compile_kwargs)  # type: ignore[assignment]
         self.network = network
         self.compile = compile
         self.optimizer = optimizer

monai/networks/blocks/crossattention.py

@@ -17,7 +17,7 @@
 import torch.nn as nn
 
 from monai.networks.layers.utils import get_rel_pos_embedding_layer
-from monai.utils import optional_import, pytorch_after
+from monai.utils import optional_import
 
 Rearrange, _ = optional_import("einops.layers.torch", name="Rearrange")
 
@@ -84,11 +84,6 @@ def __init__(
         if causal and sequence_length is None:
             raise ValueError("sequence_length is necessary for causal attention.")
 
-        if use_flash_attention and not pytorch_after(minor=13, major=1, patch=0):
-            raise ValueError(
-                "use_flash_attention is only supported for PyTorch versions >= 2.0."
-                "Upgrade your PyTorch or set the flag to False."
-            )
         if use_flash_attention and save_attn:
             raise ValueError(
                 "save_attn has been set to True, but use_flash_attention is also set"

monai/networks/blocks/selfattention.py

@@ -18,7 +18,7 @@
 import torch.nn.functional as F
 
 from monai.networks.layers.utils import get_rel_pos_embedding_layer
-from monai.utils import optional_import, pytorch_after
+from monai.utils import optional_import
 
 Rearrange, _ = optional_import("einops.layers.torch", name="Rearrange")
 
@@ -90,11 +90,6 @@ def __init__(
         if causal and sequence_length is None:
             raise ValueError("sequence_length is necessary for causal attention.")
 
-        if use_flash_attention and not pytorch_after(minor=13, major=1, patch=0):
-            raise ValueError(
-                "use_flash_attention is only supported for PyTorch versions >= 2.0."
-                "Upgrade your PyTorch or set the flag to False."
-            )
         if use_flash_attention and save_attn:
             raise ValueError(
                 "save_attn has been set to True, but use_flash_attention is also set"

monai/networks/blocks/upsample.py

@@ -17,8 +17,8 @@
 import torch.nn as nn
 
 from monai.networks.layers.factories import Conv, Pad, Pool
-from monai.networks.utils import CastTempType, icnr_init, pixelshuffle
-from monai.utils import InterpolateMode, UpsampleMode, ensure_tuple_rep, look_up_option, pytorch_after
+from monai.networks.utils import icnr_init, pixelshuffle
+from monai.utils import InterpolateMode, UpsampleMode, ensure_tuple_rep, look_up_option
 
 __all__ = ["Upsample", "UpSample", "SubpixelUpsample", "Subpixelupsample", "SubpixelUpSample"]
 
@@ -164,15 +164,7 @@ def __init__(
                 align_corners=align_corners,
             )
 
-            # Cast to float32 as 'upsample_nearest2d_out_frame' op does not support bfloat16
-            # https://github.com/pytorch/pytorch/issues/86679. This issue is solved in PyTorch 2.1
-            if pytorch_after(major=2, minor=1):
-                self.add_module("upsample_non_trainable", upsample)
-            else:
-                self.add_module(
-                    "upsample_non_trainable",
-                    CastTempType(initial_type=torch.bfloat16, temporary_type=torch.float32, submodule=upsample),
-                )
+            self.add_module("upsample_non_trainable", upsample)
             if post_conv:
                 self.add_module("postconv", post_conv)
         elif up_mode == UpsampleMode.PIXELSHUFFLE:

pyproject.toml

@@ -2,7 +2,7 @@
 requires = [
   "wheel",
   "setuptools",
-  "torch>=1.13.1",
+  "torch>=2.3.0",
   "ninja",
   "packaging"
 ]

requirements.txt

@@ -1,2 +1,3 @@
-torch>=1.13.1,<2.6
-numpy>=1.24,<2.0
+torch>=2.3.0,<2.6; sys_platform != 'win32'
+torch>=2.4.1,<2.6; sys_platform == 'win32'
+numpy>=1.24,<3.0

setup.cfg

@@ -42,8 +42,9 @@ setup_requires =
     ninja
     packaging
 install_requires =
-    torch>=1.13.1
-    numpy>=1.24,<2.0
+    torch>=2.3.0; sys_platform != 'win32'
+    torch>=2.4.1; sys_platform == 'win32'
+    numpy>=1.24,<3.0
 
 [options.extras_require]
 all =

tests/integration/test_integration_bundle_run.py

@@ -76,8 +76,7 @@ def test_tiny(self):
             )
         with open(meta_file, "w") as f:
             json.dump(
-                {"version": "0.1.0", "monai_version": "1.1.0", "pytorch_version": "1.13.1", "numpy_version": "1.22.2"},
-                f,
+                {"version": "0.1.0", "monai_version": "1.1.0", "pytorch_version": "2.3.0", "numpy_version": "1.22.2"}, f
             )
         cmd = ["coverage", "run", "-m", "monai.bundle"]
         # test both CLI entry "run" and "run_workflow"
@@ -114,8 +113,7 @@ def test_scripts_fold(self):
             )
         with open(meta_file, "w") as f:
             json.dump(
-                {"version": "0.1.0", "monai_version": "1.1.0", "pytorch_version": "1.13.1", "numpy_version": "1.22.2"},
-                f,
+                {"version": "0.1.0", "monai_version": "1.1.0", "pytorch_version": "2.3.0", "numpy_version": "1.22.2"}, f
             )
 
         os.mkdir(scripts_dir)

tests/metrics/test_surface_dice.py

@@ -82,7 +82,7 @@ def test_tolerance_euclidean_distance_with_spacing(self):
         expected_res0[1, 1] = np.nan
         for b, c in np.ndindex(batch_size, n_class):
             np.testing.assert_allclose(expected_res0[b, c], res0[b, c].cpu())
-        np.testing.assert_array_equal(agg0.cpu(), np.nanmean(np.nanmean(expected_res0, axis=1), axis=0))
+        np.testing.assert_allclose(agg0.cpu(), np.nanmean(np.nanmean(expected_res0, axis=1), axis=0))
         np.testing.assert_equal(not_nans.cpu(), torch.tensor(2))
 
     def test_tolerance_euclidean_distance(self):
@@ -126,7 +126,7 @@ def test_tolerance_euclidean_distance(self):
         expected_res0[1, 1] = np.nan
         for b, c in np.ndindex(batch_size, n_class):
             np.testing.assert_allclose(expected_res0[b, c], res0[b, c].cpu())
-        np.testing.assert_array_equal(agg0.cpu(), np.nanmean(np.nanmean(expected_res0, axis=1), axis=0))
+        np.testing.assert_allclose(agg0.cpu(), np.nanmean(np.nanmean(expected_res0, axis=1), axis=0))
         np.testing.assert_equal(not_nans.cpu(), torch.tensor(2))
 
     def test_tolerance_euclidean_distance_3d(self):
@@ -173,7 +173,7 @@ def test_tolerance_euclidean_distance_3d(self):
         expected_res0[1, 1] = np.nan
         for b, c in np.ndindex(batch_size, n_class):
             np.testing.assert_allclose(expected_res0[b, c], res0[b, c].cpu())
-        np.testing.assert_array_equal(agg0.cpu(), np.nanmean(np.nanmean(expected_res0, axis=1), axis=0))
+        np.testing.assert_allclose(agg0.cpu(), np.nanmean(np.nanmean(expected_res0, axis=1), axis=0))
         np.testing.assert_equal(not_nans.cpu(), torch.tensor(2))
 
     def test_tolerance_all_distances(self):

tests/nonconfig_workflow.py

@@ -65,7 +65,7 @@ def initialize(self):
             self._monai_version = "1.1.0"
 
         if self._pytorch_version is None:
-            self._pytorch_version = "1.13.1"
+            self._pytorch_version = "2.3.0"
 
         if self._numpy_version is None:
             self._numpy_version = "1.22.2"