Remove torchdata==0.9.0 pin and torchtext as deprecated (#4849)

Signed-off-by: Anatoly Myachev <[email protected]>

Author: anmyachev

.github/workflows/e2e-reusable.yml

@@ -90,11 +90,6 @@ jobs:
           # cmake<4.0.0 is required as a workaround for CMake Error at third_party/double-conversion/CMakeLists.txt:1 (cmake_minimum_required)
           pip install wheel 'cmake<4.0.0'
 
-      # https://github.com/pytorch/data/blob/e316c5ca1ab2a4f69dd6d48e8fc9c6f8d0c7c468/README.md?plain=1#L6-L15
-      - name: Install pinned torchdata
-        run: |
-          pip install torchdata==0.9.0
-
       - name: Setup PyTorch
         uses: ./.github/actions/setup-pytorch
         with:
@@ -105,7 +100,6 @@ jobs:
           cd pytorch
           echo "BENCHMARK_COMMIT_ID=$(<.github/ci_commit_pins/torchbench.txt)" | tee -a "$GITHUB_ENV"
           echo "TORCHVISION_COMMIT_ID=$(<.github/ci_commit_pins/vision.txt)" | tee -a "$GITHUB_ENV"
-          echo "TORCHTEXT_COMMIT_ID=$(<.github/ci_commit_pins/text.txt)" | tee -a "$GITHUB_ENV"
           echo "TORCHAUDIO_COMMIT_ID=$(<.github/ci_commit_pins/audio.txt)" | tee -a "$GITHUB_ENV"
           echo "TRANSFORMERS_VERSION=$(<.ci/docker/ci_commit_pins/huggingface.txt)" | tee -a "$GITHUB_ENV"
           echo "TIMM_COMMIT_ID=$(<.ci/docker/ci_commit_pins/timm.txt)" | tee -a "$GITHUB_ENV"
@@ -163,15 +157,6 @@ jobs:
           ref: ${{ env.TORCHVISION_COMMIT_ID }}
           extra-cache-key: ${{ env.PYTORCH_VERSION }}
 
-      - name: Install torchtext package
-        if: ${{ inputs.suite == 'torchbench' }}
-        uses: ./.github/actions/install-dependency
-        with:
-          package: torchtext
-          repository: pytorch/text
-          ref: ${{ env.TORCHTEXT_COMMIT_ID }}
-          extra-cache-key: ${{ env.PYTORCH_VERSION }}
-
       - name: Install torchaudio package
         if: ${{ inputs.suite == 'torchbench' }}
         uses: ./.github/actions/install-dependency
@@ -278,7 +263,6 @@ jobs:
           TRITON_REPO=$GITHUB_REPOSITORY
           TRITON_COMMIT_ID=$GITHUB_SHA
           TORCHVISION_COMMIT_ID=$TORCHVISION_COMMIT_ID
-          TORCHTEXT_COMMIT_ID=$TORCHTEXT_COMMIT_ID
           TORCHAUDIO_COMMIT_ID=$TORCHAUDIO_COMMIT_ID
           TRANSFORMERS_VERSION=$TRANSFORMERS_VERSION
           TIMM_COMMIT_ID=$TIMM_COMMIT_ID

.github/workflows/e2e-windows.yml

@@ -159,17 +159,14 @@ jobs:
           cd /c/pytorch
           echo "BENCHMARK_COMMIT_ID=$(<.github/ci_commit_pins/torchbench.txt)" | tee -a "$GITHUB_ENV"
           echo "TORCHVISION_COMMIT_ID=$(<.github/ci_commit_pins/vision.txt)" | tee -a "$GITHUB_ENV"
-          echo "TORCHTEXT_COMMIT_ID=$(<.github/ci_commit_pins/text.txt)" | tee -a "$GITHUB_ENV"
           echo "TORCHAUDIO_COMMIT_ID=$(<.github/ci_commit_pins/audio.txt)" | tee -a "$GITHUB_ENV"
           echo "TRANSFORMERS_VERSION=$(<.ci/docker/ci_commit_pins/huggingface.txt)" | tee -a "$GITHUB_ENV"
           echo "TIMM_COMMIT_ID=$(<.ci/docker/ci_commit_pins/timm.txt)" | tee -a "$GITHUB_ENV"
 
-      - name: Install pinned torchdata
+      - name: Install python test dependencies
         run: |
           .venv\Scripts\activate.ps1
-          # Pinned torchdata
-          # https://github.com/pytorch/data/blob/e316c5ca1ab2a4f69dd6d48e8fc9c6f8d0c7c468/README.md?plain=1#L6-L15
-          pip install pyyaml pandas scipy numpy psutil pyre_extensions torchrec torchdata==0.9.0
+          pip install pyyaml pandas scipy numpy psutil pyre_extensions torchrec
 
       - name: Install transformers package
         if: inputs.suite == 'all' || inputs.suite == 'huggingface'
@@ -194,16 +191,6 @@ jobs:
           extra-cache-key: ${{ env.PYTORCH_VERSION }}
           workspace: /c/gh${{ github.run_id }}
 
-      - name: Install torchtext package
-        if: inputs.suite == 'all' || inputs.suite == 'torchbench'
-        uses: ./.github/actions/install-dependency
-        with:
-          package: torchtext
-          repository: pytorch/text
-          ref: ${{ env.TORCHTEXT_COMMIT_ID }}
-          extra-cache-key: ${{ env.PYTORCH_VERSION }}
-          workspace: /c/gh${{ github.run_id }}
-
       - name: Install torchaudio package
         if: inputs.suite == 'all' || inputs.suite == 'torchbench'
         uses: ./.github/actions/install-dependency
@@ -348,7 +335,6 @@ jobs:
           TRITON_REPO=$GITHUB_REPOSITORY
           TRITON_COMMIT_ID=$GITHUB_SHA
           TORCHVISION_COMMIT_ID=$TORCHVISION_COMMIT_ID
-          TORCHTEXT_COMMIT_ID=$TORCHTEXT_COMMIT_ID
           TORCHAUDIO_COMMIT_ID=$TORCHAUDIO_COMMIT_ID
           TRANSFORMERS_VERSION=$TRANSFORMERS_VERSION
           TIMM_COMMIT_ID=$TIMM_COMMIT_ID

.github/workflows/nightly-wheels.yml

@@ -49,19 +49,13 @@ jobs:
           # cmake 3.22.1 does not work with the recent torchaudio: https://github.com/intel/intel-xpu-backend-for-triton/issues/2079
           pip install wheel cmake
 
-      # https://github.com/pytorch/data/blob/e316c5ca1ab2a4f69dd6d48e8fc9c6f8d0c7c468/README.md?plain=1#L6-L15
-      - name: Install pinned torchdata
-        run: |
-          pip install torchdata==0.9.0
-
       - name: Setup PyTorch
         uses: ./.github/actions/setup-pytorch
 
       - name: Identify pinned versions
         run: |
           cd pytorch
           echo "TORCHVISION_COMMIT_ID=$(<.github/ci_commit_pins/vision.txt)" | tee -a $GITHUB_ENV
-          echo "TORCHTEXT_COMMIT_ID=$(<.github/ci_commit_pins/text.txt)" | tee -a $GITHUB_ENV
           echo "TORCHAUDIO_COMMIT_ID=$(<.github/ci_commit_pins/audio.txt)" | tee -a $GITHUB_ENV
           echo "TRANSFORMERS_VERSION=$(<.ci/docker/ci_commit_pins/huggingface.txt)" | tee -a $GITHUB_ENV
           echo "TIMM_COMMIT_ID=$(<.ci/docker/ci_commit_pins/timm.txt)" | tee -a $GITHUB_ENV
@@ -85,14 +79,6 @@ jobs:
           ref: ${{ env.TORCHVISION_COMMIT_ID }}
           extra-cache-key: ${{ env.PYTORCH_VERSION }}
 
-      - name: Install torchtext package
-        uses: ./.github/actions/install-dependency
-        with:
-          package: torchtext
-          repository: pytorch/text
-          ref: ${{ env.TORCHTEXT_COMMIT_ID }}
-          extra-cache-key: ${{ env.PYTORCH_VERSION }}
-
       - name: Install torchaudio package
         uses: ./.github/actions/install-dependency
         with:
@@ -124,7 +110,6 @@ jobs:
           cp -L pytorch/dist/*.whl wheels/
           cp -L dist/*.whl wheels/
           cp -L torchvision*/dist/*.whl wheels/
-          cp -L torchtext*/dist/*.whl wheels/
           cp -L torchaudio*/dist/*.whl wheels/
           cp -L timm*/dist/*.whl wheels/
           cp -L transformers*/dist/*.whl wheels/
@@ -148,7 +133,6 @@ jobs:
           TRITON_REPO=intel/intel-xpu-backend-for-triton
           TRITON_COMMIT_ID=$TRITON_COMMIT_ID
           TORCHVISION_COMMIT_ID=$TORCHVISION_COMMIT_ID
-          TORCHTEXT_COMMIT_ID=$TORCHTEXT_COMMIT_ID
           TORCHAUDIO_COMMIT_ID=$TORCHAUDIO_COMMIT_ID
           EOF
 

scripts/install-pytorch.sh

@@ -129,7 +129,7 @@ fi
 ############################################################################
 # Check installed torch pinned dependencies
 
-PINNED_TORCH_DEPENDENCIES_REGEX="^torchtext==|^torchaudio==|^torchvision=="
+PINNED_TORCH_DEPENDENCIES_REGEX="^torchaudio==|^torchvision=="
 INSTALLED_PINNED_TORCH_DEPENDENCIES=$(pip list --format=freeze | grep -iE "$PINNED_TORCH_DEPENDENCIES_REGEX" || true)
 
 if [ -n "$INSTALLED_PINNED_TORCH_DEPENDENCIES" ]; then
@@ -143,7 +143,7 @@ if [ -n "$INSTALLED_PINNED_TORCH_DEPENDENCIES" ]; then
     echo "**** INFO: PyTorch pinned dependencies build from source mode is not supported. ****"
     exit 1
   fi
-  pip uninstall -y torchtext torchaudio torchvision
+  pip uninstall -y torchaudio torchvision
 fi
 
 ############################################################################