diff --git a/.github/workflows/pytest.yml b/.github/workflows/pytest.yml
index 03a307da140..ce2ca43673b 100644
--- a/.github/workflows/pytest.yml
+++ b/.github/workflows/pytest.yml
@@ -22,36 +22,19 @@ jobs:
       fail-fast: false
       matrix:
         os: [ubuntu-latest]
-        python-version: ["3.8", "3.9", "3.10"]
-        test-markers: ["not distributed", "distributed"]
-        include:
-          - python-version: "3.8"
-            pytorch-version: 2.0.0
-            torchscript-version: 1.10.2
-            ray-version: 2.3.1
-          - python-version: "3.9"
-            pytorch-version: 2.1.1
-            torchscript-version: 1.10.2
-            ray-version: 2.3.1
-          - python-version: "3.10"
-            # pytorch-version: nightly
-            pytorch-version: 2.2.1
-            torchscript-version: 1.10.2
-            ray-version: 2.3.1
+        #python-version: ["3.10", "3.11", "3.12"]
+        python-version: ["3.11"]
+        test-markers: ["not distributed"] #["not distributed", "distributed"]
+
     env:
-      PYTORCH: ${{ matrix.pytorch-version }}
       MARKERS: ${{ matrix.test-markers }}
-      NEUROPOD_BASE_DIR: "/usr/local/lib/neuropod"
-      NEUROPOD_VERISON: "0.3.0-rc6"
-      TORCHSCRIPT_VERSION: ${{ matrix.torchscript-version }}
-      RAY_VERSION: ${{ matrix.ray-version }}
       AWS_ACCESS_KEY_ID: ${{ secrets.LUDWIG_TESTS_AWS_ACCESS_KEY_ID }}
       AWS_SECRET_ACCESS_KEY: ${{ secrets.LUDWIG_TESTS_AWS_SECRET_ACCESS_KEY }}
       KAGGLE_USERNAME: ${{ secrets.KAGGLE_USERNAME }}
       KAGGLE_KEY: ${{ secrets.KAGGLE_KEY }}
       IS_NOT_FORK: ${{ !(github.event.pull_request.base.repo.full_name == 'ludwig-ai/ludwig' && github.event.pull_request.head.repo.fork) }}
 
-    name: py${{ matrix.python-version  }}, torch-${{ matrix.pytorch-version }}, ${{ matrix.test-markers }}, ${{ matrix.os }}, ray ${{ matrix.ray-version }}
+    name: py${{ matrix.python-version  }}, ${{ matrix.test-markers }}, ${{ matrix.os }}
     services:
       minio:
         image: fclairamb/minio-github-actions
@@ -79,7 +62,19 @@ jobs:
       - name: Setup Linux
         if: runner.os == 'linux'
         run: |
-          sudo apt-get update && sudo apt-get install -y cmake libsndfile1 wget libsox-dev
+          sudo apt-get update && sudo apt-get install -y build-essential cmake liblapack-dev gfortran libsndfile1 wget libsox-dev libopenblas-dev
+
+      - name: Download longintrepr.h
+        run: |
+          sudo mkdir -p /usr/include/python3.11
+          sudo curl -o /usr/include/python3.11/longintrepr.h https://raw.githubusercontent.com/python/cpython/refs/heads/main/Include/cpython/longintrepr.h
+          
+
+      # - name: Install GPy
+      #   run: |
+      #     python -m pip install -U pip
+      #     pip install Cython==0.29.35
+      #     pip install GPy==1.10.0
 
       - name: Setup macOS
         if: runner.os == 'macOS'
@@ -91,7 +86,7 @@ jobs:
         uses: actions/cache@v2
         with:
           path: ~/.cache/pip
-          key: ${{ runner.os }}-pip-py${{ matrix.python-version }}-torch${{ matrix.pytorch-version }}-${{ matrix.test-markers }}-${{ hashFiles('requirements*.txt', '.github/workflows/pytest.yml') }}
+          key: ${{ runner.os }}-pip-py${{ matrix.python-version }}-${{ matrix.test-markers }}-${{ hashFiles('.github/workflows/pytest.yml') }}
 
       - name: Debug out of space
         run: |
@@ -103,59 +98,244 @@ jobs:
           python --version
           pip --version
           python -m pip install -U pip
+          pip install .
           cmake --version
-
-          # remove torch and ray from the dependencies so we can add them depending on the matrix args for the job.
-          cat requirements.txt | sed '/^torch[>=<\b]/d' | sed '/^torchtext/d' | sed '/^torchvision/d' | sed '/^torchaudio/d' > requirements-temp && mv requirements-temp requirements.txt
-          cat requirements_distributed.txt | sed '/^ray[\[]/d'
-
-          if [ "$MARKERS" != "distributed" ]; then
-            # Skip distributed and hyperopt requirements to test optional imports
-            echo > requirements-temp && mv requirements-temp requirements_distributed.txt
-            echo > requirements-temp && mv requirements-temp requirements_hyperopt.txt
-
-            # Skip distributed tree requirement (lightgbm-ray)
-            cat requirements_tree.txt | sed '/^lightgbm-ray/d' > requirements-temp && mv requirements-temp requirements_tree.txt
-          else
-            if [ "$RAY_VERSION" == "nightly" ]; then
-              # NOTE: hardcoded for python 3.10 on Linux
-              echo "ray[default,data,serve,tune] @ https://s3-us-west-2.amazonaws.com/ray-wheels/latest/ray-3.0.0.dev0-cp310-cp310-manylinux2014_x86_64.whl" >> requirements_distributed.txt
-            else
-              echo "ray[default,data,serve,tune]==$RAY_VERSION" >> requirements_distributed.txt
-            fi
-          fi
-
-          if [ "$PYTORCH" == "nightly" ]; then
-            extra_index_url=https://download.pytorch.org/whl/nightly/cpu
-            pip install --pre torch torchtext torchvision torchaudio --extra-index-url $extra_index_url
-
-          else
-            extra_index_url=https://download.pytorch.org/whl/cpu
-            pip install torch==$PYTORCH torchtext torchvision torchaudio --extra-index-url $extra_index_url
-          fi
-
-          pip install '.[test]' --extra-index-url $extra_index_url
           pip list
+        shell: bash
+
+      - name: Install Dependencies for dev
+        run: |
+          echo "starting dev"
+
+          pip install flake8
+          echo "Installed flake8 successfully."
+
+          pip install flake8-pyproject
+          echo "Installed flake8-pyproject successfully."
+
+          pip install pre-commit
+          echo "Installed pre-commit successfully."
+
+          pip install setuptools
+          echo "Installed setuptools successfully."
+
+      - name: Install Dependencies for test
+        run: |
+          echo "starting test"
+          pip install pytest
+          echo "Installed pytest successfully."
+
+          pip install pytest-timeout
+          echo "Installed pytest-timeout successfully."
+
+          pip install pytest-cov
+          echo "Installed pytest-cov successfully."
+
+          pip install tifffile
+          echo "Installed tifffile successfully."
+
+          pip install wget
+          echo "Installed wget successfully."
+
+          pip install six>=1.13.0
+          echo "Installed six successfully."
+
+          pip install aim
+          echo "Installed aim successfully."
+
+          pip install wandb
+          echo "Installed wandb successfully."
+
+          pip install comet_ml
+          echo "Installed comet_ml successfully."
+
+          pip install mlflow
+          echo "Installed mlflow successfully."
+
+          pip install "sqlalchemy<2"
+          echo "Installed sqlalchemy successfully."
+
+          pip install hpbandster
+          echo "Installed hpbandster successfully."
+
+          pip install ConfigSpace==0.7.1
+          echo "Installed ConfigSpace successfully."
+
+          pip install ax-platform
+          echo "Installed ax-platform successfully."
+
+          pip install bayesian-optimization
+          echo "Installed bayesian-optimization successfully."
+
+          pip install flaml[blendsearch]
+          echo "Installed flaml[blendsearch] successfully."
+
+          pip install HEBO
+          echo "Installed HEBO successfully."
+
+          pip install nevergrad
+          echo "Installed nevergrad successfully."
+
+          pip install optuna
+          echo "Installed optuna successfully."
+
+          pip install scikit-optimize
+          echo "Installed scikit-optimize successfully."
+
+          pip install zoopt
+          echo "Installed zoopt successfully."
+
+      - name: Install Dependencies for benchmarking
+        run: |
+          echo "starting benchmarking"
+          pip install s3fs>=2022.8.2
+          echo "Installed s3fs successfully."
+
+      - name: Install Dependencies for distributed
+        run: |
+          echo "starting distributed"
+          pip install awscli
+          echo "Installed awscli successfully."
+
+          pip install "dask[dataframe]<2023.4.0"
+          echo "Installed dask[dataframe] successfully."
+
+          pip install "deepspeed!=0.11.0,<0.13.0"
+          echo "Installed deepspeed successfully."
+
+          pip install getdaft[ray]==0.1.20
+          echo "Installed getdaft[ray] successfully."
+
+          pip install GPUtil
+          echo "Installed GPUtil successfully."
+
+          pip install pyarrow
+          echo "Installed pyarrow successfully."
+
+          pip install ray[default,data,serve,tune]==2.3.1
+          echo "Installed ray[default,data,serve,tune] successfully."
+
+          pip install tblib
+          echo "Installed tblib successfully."
+
+          pip install "tensorboardX<2.3"
+          echo "Installed tensorboardX successfully."
+
+      - name: Install Dependencies for explain
+        run: |
+          echo "starting explain"
+          pip install captum
+          echo "Installed captum successfully."
+
+      - name: Install Dependencies for extra
+        run: |
+          echo "starting extra"
+          pip install horovod[pytorch]>=0.24.0,!=0.26.0
+          echo "Installed horovod[pytorch] successfully."
+
+          pip install modin[ray]
+          echo "Installed modin[ray] successfully."
+
+          pip install predibase>=2023.10.2
+          echo "Installed predibase successfully."
+
+      - name: Install Dependencies for hyperopt
+        run: |
+          echo "starting hyperopt"
+          pip install hyperopt
+          echo "Installed hyperopt successfully."
+
+          pip install ray[default,tune]>=2.0.0
+          echo "Installed ray[default,tune] successfully."
+
+      - name: Install Dependencies for llm
+        run: |
+          echo "starting llm"
+          pip install accelerate
+          echo "Installed accelerate successfully."
 
-          if [ "$PYTORCH" == "nightly" ]; then
-            python -c "from packaging import version; import torch; assert version.parse(torch.__version__).release >= version.parse(\"2.0.0\").release, f\"torch {version.parse(torch.__version__).release} < version.parse(\'2.0.0\').release\""
-          else
-            python -c "from packaging import version; import torch; assert version.parse(torch.__version__).release == version.parse(\"$PYTORCH\").release, f\"torch {version.parse(torch.__version__).release} != version.parse(\'$PYTORCH\').release\""
-          fi
-
-          if [ "$MARKERS" == "distributed" ]; then
-            python -c "from packaging import version; import ray; assert version.parse(ray.__version__).release == version.parse(\"$RAY_VERSION\").release, f\"ray {version.parse(ray.__version__).release} != version.parse(\'$RAY_VERSION\').release\""
-          else
-            python -c "import importlib.util; assert importlib.util.find_spec('ray') is None, \"found ray but expected it to not be installed\""
-          fi
+          pip install faiss-cpu
+          echo "Installed faiss-cpu successfully."
+
+          pip install loralib
+          echo "Installed loralib successfully."
+
+          pip install peft>=0.10.0
+          echo "Installed peft successfully."
+
+          pip install sentence-transformers
+          echo "Installed sentence-transformers successfully."
+
+      - name: Install Dependencies for serve
+        run: |
+          echo "starting serve"
+          pip install cartonml-nightly
+          echo "Installed cartonml-nightly successfully."
+
+          pip install fastapi
+          echo "Installed fastapi successfully."
+
+          pip install httpx
+          echo "Installed httpx successfully."
+
+          pip install "neuropod==0.3.0rc6 ; platform_system != 'Windows' and python_version < '3.9'"
+          echo "Installed neuropod successfully."
+
+          pip install python-multipart
+          echo "Installed python-multipart successfully."
+
+          pip install uvicorn
+          echo "Installed uvicorn successfully."
+
+          pip install starlette
+          echo "Installed starlette successfully."
+
+      - name: Install Dependencies for tree
+        run: |
+          echo "starting tree"
+          pip install hummingbird-ml>=0.4.8
+          echo "Installed hummingbird-ml successfully."
+
+          pip install lightgbm
+          echo "Installed lightgbm successfully."
+
+          pip install lightgbm-ray
+          echo "Installed lightgbm-ray successfully."
+
+      - name: Install Dependencies for viz
+        run: |
+          echo "starting viz"
+          pip install hiplot
+          echo "Installed hiplot successfully."
+
+          pip install matplotlib==3.9.3
+          echo "Installed matplotlib successfully."
+
+          pip install ptitprince
+          echo "Installed ptitprince successfully."
+
+          pip install "seaborn>=0.7,<0.12"
+          echo "Installed seaborn successfully."
+
+      - name: test weird GPy issue
+        run: |
+          # Extract and install each dependency individually
+          for dep in $(python -c "import tomllib; print(' '.join(tomllib.load(open('pyproject.toml', 'rb'))['project']['optional-dependencies']['test']))"); do
+            pip install "$dep" && echo "Installed $dep successfully."
+          done
+
+          # Install the core package
+          pip install . && echo "Core package installed successfully."
         shell: bash
 
-      - name: Install Neuropod backend
+      - name: Debug dependency tree
         run: |
-          sudo mkdir -p "$NEUROPOD_BASE_DIR"
-          curl -L https://github.com/uber/neuropod/releases/download/v${{ env.NEUROPOD_VERISON }}/libneuropod-cpu-linux-v${{ env.NEUROPOD_VERISON }}-torchscript-${{ env.TORCHSCRIPT_VERSION }}-backend.tar.gz | sudo tar -xz -C "$NEUROPOD_BASE_DIR"
+          python -m pip install pipdeptree
+          pipdeptree > dependency-tree.txt
+          cat dependency-tree.txt
         shell: bash
 
+
       - name: Unit Tests
         run: |
           RUN_PRIVATE=$IS_NOT_FORK LUDWIG_TEST_SUITE_TIMEOUT_S=5400 pytest -v --timeout 300 --durations 100 -m "$MARKERS and not slow and not combinatorial and not horovod and not llm" --junitxml pytest.xml tests/ludwig
@@ -192,174 +372,179 @@ jobs:
 
       - name: Upload Unit Test Results
         if: ${{ always() && !env.ACT }}
-        uses: actions/upload-artifact@v2
+        uses: actions/upload-artifact@v4
         with:
           name: Unit Test Results (Python ${{ matrix.python-version }} ${{ matrix.test-markers }})
-          path: pytest.xml
+          path: pytest-${{ matrix.python-version }}-${{ matrix.test-markers }}.xml
 
-  integration-tests:
-    name: ${{ matrix.test-markers }}
-    runs-on: ubuntu-latest
-    strategy:
-      fail-fast: false
-      matrix:
-        test-markers:
-          - "integration_tests_a"
-          - "integration_tests_b"
-          - "integration_tests_c"
-          - "integration_tests_d"
-          - "integration_tests_e"
-          - "integration_tests_f"
+  # integration-tests:
+  #   name: ${{ matrix.test-markers }}
+  #   runs-on: ubuntu-latest
+  #   strategy:
+  #     fail-fast: false
+  #     matrix:
+  #       test-markers:
+  #         - "integration_tests_a"
+  #         - "integration_tests_b"
+  #         - "integration_tests_c"
+  #         - "integration_tests_d"
+  #         - "integration_tests_e"
+  #         - "integration_tests_f"
 
-    env:
-      AWS_ACCESS_KEY_ID: ${{ secrets.LUDWIG_TESTS_AWS_ACCESS_KEY_ID }}
-      AWS_SECRET_ACCESS_KEY: ${{ secrets.LUDWIG_TESTS_AWS_SECRET_ACCESS_KEY }}
-      KAGGLE_USERNAME: ${{ secrets.KAGGLE_USERNAME }}
-      KAGGLE_KEY: ${{ secrets.KAGGLE_KEY }}
-      IS_NOT_FORK: ${{ !(github.event.pull_request.base.repo.full_name == 'ludwig-ai/ludwig' && github.event.pull_request.head.repo.fork) }}
-      MARKERS: ${{ matrix.test-markers }}
+  #   env:
+  #     AWS_ACCESS_KEY_ID: ${{ secrets.LUDWIG_TESTS_AWS_ACCESS_KEY_ID }}
+  #     AWS_SECRET_ACCESS_KEY: ${{ secrets.LUDWIG_TESTS_AWS_SECRET_ACCESS_KEY }}
+  #     KAGGLE_USERNAME: ${{ secrets.KAGGLE_USERNAME }}
+  #     KAGGLE_KEY: ${{ secrets.KAGGLE_KEY }}
+  #     IS_NOT_FORK: ${{ !(github.event.pull_request.base.repo.full_name == 'ludwig-ai/ludwig' && github.event.pull_request.head.repo.fork) }}
+  #     MARKERS: ${{ matrix.test-markers }}
+
+  #   services:
+  #     minio:
+  #       image: fclairamb/minio-github-actions
+  #       env:
+  #         MINIO_ACCESS_KEY: minio
+  #         MINIO_SECRET_KEY: minio123
+  #       ports:
+  #         - 9000:9000
 
-    services:
-      minio:
-        image: fclairamb/minio-github-actions
-        env:
-          MINIO_ACCESS_KEY: minio
-          MINIO_SECRET_KEY: minio123
-        ports:
-          - 9000:9000
+  #   timeout-minutes: 90
+  #   steps:
+  #     - uses: actions/checkout@v2
+  #     - name: Set up Python 3.10
+  #       uses: actions/setup-python@v2
+  #       with:
+  #         python-version: "3.10"
 
-    timeout-minutes: 90
-    steps:
-      - uses: actions/checkout@v2
-      - name: Set up Python 3.10
-        uses: actions/setup-python@v2
-        with:
-          python-version: "3.10"
+  #     - name: Setup Linux
+  #       if: runner.os == 'linux'
+  #       run: |
+  #         sudo apt-get update && sudo apt-get install -y cmake libsndfile1
 
-      - name: Setup Linux
-        if: runner.os == 'linux'
-        run: |
-          sudo apt-get update && sudo apt-get install -y cmake libsndfile1
+  #     - name: Setup macOS
+  #       if: runner.os == 'macOS'
+  #       run: |
+  #         brew install libuv
 
-      - name: Setup macOS
-        if: runner.os == 'macOS'
-        run: |
-          brew install libuv
+  #     - name: Install dependencies
+  #       run: |
+  #         python --version
+  #         pip --version
+  #         python -m pip install -U pip
 
-      - name: Install dependencies
-        run: |
-          python --version
-          pip --version
-          python -m pip install -U pip
+  #         # remove torch and ray from the dependencies so we can add them depending on the matrix args for the job.
+  #         cat requirements.txt | sed '/^torch[>=<\b]/d' | sed '/^torchtext/d' | sed '/^torchvision/d' | sed '/^torchaudio/d' > requirements-temp && mv requirements-temp requirements.txt
+  #         cat requirements_distributed.txt | sed '/^ray[\[]/d'
+  #         pip install torch==2.0.0 torchtext torchvision torchaudio
+  #         pip install ray==2.3.0
+  #         pip install '.[test]'
+  #         pip list
+  #       shell: bash
 
-          # remove torch and ray from the dependencies so we can add them depending on the matrix args for the job.
-          cat requirements.txt | sed '/^torch[>=<\b]/d' | sed '/^torchtext/d' | sed '/^torchvision/d' | sed '/^torchaudio/d' > requirements-temp && mv requirements-temp requirements.txt
-          cat requirements_distributed.txt | sed '/^ray[\[]/d'
-          pip install torch==2.0.0 torchtext torchvision torchaudio
-          pip install ray==2.3.0
-          pip install '.[test]'
-          pip list
-        shell: bash
+  #     - name: Free Disk Space (Ubuntu)
+  #       uses: jlumbroso/free-disk-space@main
+  #       with:
+  #         tool-cache: false
+  #         android: true
+  #         dotnet: true
+  #         haskell: true
+  #         large-packages: false
+  #         docker-images: true
+  #         swap-storage: true
+
+  #     - name: Clean out /tmp directory
+  #       run: |
+  #         sudo rm -rf /tmp/*
 
-      - name: Free Disk Space (Ubuntu)
-        uses: jlumbroso/free-disk-space@main
-        with:
-          tool-cache: false
-          android: true
-          dotnet: true
-          haskell: true
-          large-packages: false
-          docker-images: true
-          swap-storage: true
-
-      - name: Clean out /tmp directory
-        run: |
-          sudo rm -rf /tmp/*
+  #     - name: Integration Tests
+  #       run: |
+  #         RUN_PRIVATE=$IS_NOT_FORK LUDWIG_TEST_SUITE_TIMEOUT_S=7200 pytest -v --timeout 300 --durations 100 -m "not slow and not combinatorial and not horovod and not llm and $MARKERS" --junitxml pytest.xml tests/integration_tests
 
-      - name: Integration Tests
-        run: |
-          RUN_PRIVATE=$IS_NOT_FORK LUDWIG_TEST_SUITE_TIMEOUT_S=7200 pytest -v --timeout 300 --durations 100 -m "not slow and not combinatorial and not horovod and not llm and $MARKERS" --junitxml pytest.xml tests/integration_tests
+  # llm-tests:
+  #   name: LLM Tests
+  #   runs-on: ubuntu-latest
 
-  llm-tests:
-    name: LLM Tests
-    runs-on: ubuntu-latest
+  #   timeout-minutes: 60
+  #   steps:
+  #     - uses: actions/checkout@v2
+  #     - name: Set up Python 3.9
+  #       uses: actions/setup-python@v2
+  #       with:
+  #         python-version: 3.9
 
-    timeout-minutes: 60
-    steps:
-      - uses: actions/checkout@v2
-      - name: Set up Python 3.9
-        uses: actions/setup-python@v2
-        with:
-          python-version: 3.9
+  #     - name: Setup Linux
+  #       if: runner.os == 'linux'
+  #       run: |
+  #         sudo apt-get update && sudo apt-get install -y cmake libsndfile1
 
-      - name: Setup Linux
-        if: runner.os == 'linux'
-        run: |
-          sudo apt-get update && sudo apt-get install -y cmake libsndfile1
+  #     - name: Setup macOS
+  #       if: runner.os == 'macOS'
+  #       run: |
+  #         brew install libuv
 
-      - name: Setup macOS
-        if: runner.os == 'macOS'
-        run: |
-          brew install libuv
+  #     - name: Install dependencies
+  #       run: |
+  #         python --version
+  #         pip --version
+  #         python -m pip install -U pip
 
-      - name: Install dependencies
-        run: |
-          python --version
-          pip --version
-          python -m pip install -U pip
+  #         # remove torch and ray from the dependencies so we can add them depending on the matrix args for the job.
+  #         cat requirements.txt | sed '/^torch[>=<\b]/d' | sed '/^torchtext/d' | sed '/^torchvision/d' | sed '/^torchaudio/d' > requirements-temp && mv requirements-temp requirements.txt
+  #         cat requirements_distributed.txt | sed '/^ray[\[]/d'
+  #         pip install torch==2.0.0 torchtext torchvision torchaudio
+  #         pip install ray==2.3.0
+  #         pip install '.[test]'
+  #         pip list
+  #       shell: bash
 
-          # remove torch and ray from the dependencies so we can add them depending on the matrix args for the job.
-          cat requirements.txt | sed '/^torch[>=<\b]/d' | sed '/^torchtext/d' | sed '/^torchvision/d' | sed '/^torchaudio/d' > requirements-temp && mv requirements-temp requirements.txt
-          cat requirements_distributed.txt | sed '/^ray[\[]/d'
-          pip install torch==2.0.0 torchtext torchvision torchaudio
-          pip install ray==2.3.0
-          pip install '.[test]'
-          pip list
-        shell: bash
+  #     - name: LLM Tests
+  #       run: |
+  #         pytest -vs --durations 100 -m "llm" --junitxml pytest.xml tests
 
-      - name: LLM Tests
-        run: |
-          pytest -vs --durations 100 -m "llm" --junitxml pytest.xml tests
 
-  combinatorial-tests:
-    name: Combinatorial Tests
-    runs-on: ubuntu-latest
+  #COMMENTED OUT COMBINATORIAL TEST ########################################
 
-    timeout-minutes: 60
-    steps:
-      - uses: actions/checkout@v2
-      - name: Set up Python 3.8
-        uses: actions/setup-python@v2
-        with:
-          python-version: 3.8
+  # combinatorial-tests:
+  #   name: Combinatorial Tests
+  #   runs-on: ubuntu-latest
 
-      - name: Setup Linux
-        if: runner.os == 'linux'
-        run: |
-          sudo apt-get update && sudo apt-get install -y cmake libsndfile1
+  #   timeout-minutes: 60
+  #   steps:
+  #     - uses: actions/checkout@v2
+  #     - name: Set up Python 3.11
+  #       uses: actions/setup-python@v5
+  #       with:
+  #         python-version: "3.11"
 
-      - name: Setup macOS
-        if: runner.os == 'macOS'
-        run: |
-          brew install libuv
+  #     - name: Setup Linux
+  #       if: runner.os == 'linux'
+  #       run: |
+  #         sudo apt-get update && sudo apt-get install -y cmake libsndfile1
 
-      - name: Install dependencies
-        run: |
-          python --version
-          pip --version
-          python -m pip install -U pip
-          pip install torch==2.0.0 torchtext torchvision torchaudio
-          pip install '.[test]'
-          pip list
-        shell: bash
+  #     - name: Setup macOS
+  #       if: runner.os == 'macOS'
+  #       run: |
+  #         brew install libuv
 
-      - name: Testing combinatorial config generation code
-        run: |
-          pytest -vs --durations 100 -m "combinatorial" --junitxml pytest.xml tests/ludwig/config_sampling
+  #     - name: Install dependencies
+  #       run: |
+  #         python --version
+  #         pip --version
+  #         python -m pip install -U pip
+  #         pip install '.[test]'
+  #         pip list
+  #       shell: bash
+
+  #     - name: Testing combinatorial config generation code
+  #       run: |
+  #         pytest -vs --durations 100 -m "combinatorial" --junitxml pytest.xml tests/ludwig/config_sampling
+
+  #     - name: Combinatorial Tests
+  #       run: |
+  #         pytest -rx --durations 100 -m "combinatorial" --junitxml pytest.xml tests/training_success
+
+  #COMMENTED OUT COMBINATORIAL TEST ########################################
 
-      - name: Combinatorial Tests
-        run: |
-          pytest -rx --durations 100 -m "combinatorial" --junitxml pytest.xml tests/training_success
 
   test-minimal-install:
     name: Test Minimal Install
@@ -368,10 +553,10 @@ jobs:
     timeout-minutes: 15
     steps:
       - uses: actions/checkout@v2
-      - name: Set up Python 3.8
+      - name: Set up Python 3.11
         uses: actions/setup-python@v2
         with:
-          python-version: 3.8
+          python-version: "3.11"
 
       - name: Setup Linux
         if: runner.os == 'linux'
@@ -388,9 +573,7 @@ jobs:
           python --version
           pip --version
           python -m pip install -U pip
-          pip install ray==2.3.0
-          pip install '.'
-          pip install torch==2.0.0 torchtext torchvision torchaudio
+          pip install -e '.'
           pip list
         shell: bash
       - name: Check Install
@@ -560,7 +743,7 @@ jobs:
     steps:
       - name: Upload
         if: ${{ !env.ACT }}
-        uses: actions/upload-artifact@v2
+        uses: actions/upload-artifact@v4
         with:
           name: Event File
           path: ${{ github.event_path }}
diff --git a/.github/workflows/pytest_slow.yml b/.github/workflows/pytest_slow.yml
index f2f8b493bad..9e93c2903fc 100644
--- a/.github/workflows/pytest_slow.yml
+++ b/.github/workflows/pytest_slow.yml
@@ -50,7 +50,7 @@ jobs:
           python --version
           pip --version
           python -m pip install -U pip
-          pip install torch==2.1.0 torchtext torchvision torchaudio
+          pip install torch==2.1.0 torchvision torchaudio
           pip install ray==2.3.1
           pip install '.[test]'
 
diff --git a/.gitignore b/.gitignore
index 1c234c6996e..a4166a54fab 100644
--- a/.gitignore
+++ b/.gitignore
@@ -140,3 +140,5 @@ examples/*/visualizations/
 
 # benchmarking configs
 ludwig/benchmarking/configs/
+pytest.xml
+ludwig.code-workspace
diff --git a/.pre-commit-config.yaml b/.pre-commit-config.yaml
index 1c6390db514..d25aac6c912 100644
--- a/.pre-commit-config.yaml
+++ b/.pre-commit-config.yaml
@@ -34,7 +34,7 @@ repos:
       - id: pyupgrade
         args: [--py36-plus]
   - repo: https://github.com/PyCQA/docformatter
-    rev: v1.5.1
+    rev: 06907d0
     hooks:
       - id: docformatter
         args: [--in-place, --wrap-summaries=115, --wrap-descriptions=120]
diff --git a/docker/ludwig-ray-gpu/Dockerfile b/docker/ludwig-ray-gpu/Dockerfile
index 7721126f931..af434aac6a5 100644
--- a/docker/ludwig-ray-gpu/Dockerfile
+++ b/docker/ludwig-ray-gpu/Dockerfile
@@ -50,7 +50,7 @@ RUN pip install -U pip
 
 WORKDIR /ludwig
 
-RUN pip install --no-cache-dir torch==2.1.0 torchtext torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu118
+RUN pip install --no-cache-dir torch==2.1.0  torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu118
 
 COPY . .
 RUN pip install --no-cache-dir '.[full]' --extra-index-url https://download.pytorch.org/whl/cu118
diff --git a/docker/ludwig-ray/Dockerfile b/docker/ludwig-ray/Dockerfile
index 6075cae2e89..2c460e4a5ff 100644
--- a/docker/ludwig-ray/Dockerfile
+++ b/docker/ludwig-ray/Dockerfile
@@ -36,7 +36,7 @@ RUN pip install -U pip
 
 WORKDIR /ludwig
 
-RUN pip install --no-cache-dir torch==2.1.0 torchtext torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cpu
+RUN pip install --no-cache-dir torch==2.1.0 torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cpu
 
 COPY . .
 RUN pip install --no-cache-dir '.[full]' --extra-index-url https://download.pytorch.org/whl/cpu
diff --git a/docker/ludwig/Dockerfile b/docker/ludwig/Dockerfile
index 73a5285380f..a94e014cd0e 100644
--- a/docker/ludwig/Dockerfile
+++ b/docker/ludwig/Dockerfile
@@ -24,7 +24,7 @@ RUN pip install -U pip
 
 WORKDIR /ludwig
 
-RUN pip install --no-cache-dir torch==2.0.0 torchtext torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cpu
+RUN pip install --no-cache-dir torch==2.0.0 torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cpu
 
 COPY . .
 RUN pip install --no-cache-dir '.[full]'
diff --git a/docker/ludwig_hatch/Dockerfile b/docker/ludwig_hatch/Dockerfile
new file mode 100644
index 00000000000..0bb3a038b61
--- /dev/null
+++ b/docker/ludwig_hatch/Dockerfile
@@ -0,0 +1,17 @@
+FROM python:3.12
+
+ENV PATH="/root/.local/bin:$PATH"
+RUN apt-get -y update
+RUN apt-get -y install pipx
+RUN apt-get -y install git libsndfile1 build-essential g++ cmake ffmpeg sox libsox-dev
+RUN pipx ensurepath --force
+RUN pipx install hatch
+RUN python3 -m pip install --upgrade pipx
+WORKDIR /ludwig
+#COPY /ludwig/ .
+COPY . .
+
+RUN hatch env create
+RUN hatch build
+
+ENTRYPOINT ["ludwig"]
diff --git a/ludwig/__about__.py b/ludwig/__about__.py
new file mode 100644
index 00000000000..9a34ccc9fa7
--- /dev/null
+++ b/ludwig/__about__.py
@@ -0,0 +1 @@
+__version__ = "1.13.0"
diff --git a/ludwig/api.py b/ludwig/api.py
index 691219e201a..063853a7104 100644
--- a/ludwig/api.py
+++ b/ludwig/api.py
@@ -2015,9 +2015,9 @@ def to_torchscript(
         # Inputs
 
         :param  model_only (bool, optional): If True, only the ECD model will be converted to Torchscript. Else,
-            preprocessing and postprocessing steps will also be converted to Torchscript.
-        :param device (TorchDevice, optional): If None, the model will be converted to Torchscript on the same device to
-            ensure maximum model parity.
+        preprocessing and postprocessing steps will also be converted to Torchscript. :param device (TorchDevice,
+        optional): If None, the model will be converted to Torchscript on the same device to     ensure maximum model
+        parity.
 
         # Returns
 
@@ -2086,11 +2086,8 @@ def create_model(config_obj: Union[ModelConfig, dict], random_seed: int = defaul
 
         # Inputs
         :param config_obj: (Union[Config, dict]) Ludwig config object
-        :param random_seed: (int, default: ludwig default random seed) Random
-            seed used for weights initialization,
-            splits and any other random function.
-
-        # Return
+        :param random_seed: (int, default: ludwig default random seed) Random seed used for weights initialization,
+            splits and any other random function. # Return
         :return: (ludwig.models.BaseModel) Instance of the Ludwig model object.
         """
         if isinstance(config_obj, dict):
@@ -2136,7 +2133,7 @@ def is_merge_and_unload_set(self) -> bool:
 
         # Return
 
-            :return (bool): whether merge_and_unload should be done.
+        :return (bool): whether merge_and_unload should be done.
         """
         # TODO: In the future, it may be possible to move up the model type check into the BaseModel class.
         return self.config_obj.model_type == MODEL_LLM and self.model.is_merge_and_unload_set()
diff --git a/ludwig/automl/base_config.py b/ludwig/automl/base_config.py
index 5384c643a50..2cf4265492e 100644
--- a/ludwig/automl/base_config.py
+++ b/ludwig/automl/base_config.py
@@ -79,9 +79,8 @@ class DatasetInfo:
 def allocate_experiment_resources(resources: Resources) -> dict:
     """Allocates ray trial resources based on available resources.
 
-    # Inputs
-    :param resources (dict) specifies all available GPUs, CPUs and associated
-        metadata of the machines (i.e. memory)
+    # Inputs :param resources (dict) specifies all available GPUs, CPUs and associated     metadata of the machines
+    (i.e. memory)
 
     # Return
     :return: (dict) gpu and cpu resources per trial
@@ -260,9 +259,7 @@ def get_dataset_info(df: Union[pd.DataFrame, dd.core.DataFrame]) -> DatasetInfo:
     inference.
 
     # Inputs
-    :param df: (Union[pd.DataFrame, dd.core.DataFrame]) Pandas or Dask dataframe.
-
-    # Return
+    :param df: (Union[pd.DataFrame, dd.core.DataFrame]) Pandas or Dask dataframe.  # Return
     :return: (DatasetInfo) Structure containing list of FieldInfo objects.
     """
     source = wrap_data_source(df)
@@ -297,9 +294,7 @@ def get_dataset_info_from_source(source: DataSource) -> DatasetInfo:
     inference.
 
     # Inputs
-    :param source: (DataSource) A wrapper around a data source, which may represent a pandas or Dask dataframe.
-
-    # Return
+    :param source: (DataSource) A wrapper around a data source, which may represent a pandas or Dask dataframe. # Return
     :return: (DatasetInfo) Structure containing list of FieldInfo objects.
     """
     row_count = len(source)
@@ -355,10 +350,8 @@ def get_features_config(
 
     # Inputs
     :param fields: (List[FieldInfo]) FieldInfo objects for all fields in dataset
-    :param row_count: (int) total number of entries in original dataset
-    :param target_name (str, List[str]) name of target feature
-
-    # Return
+    :param row_count: (int) total number of entries in original dataset :param target_name (str, List[str]) name of
+        target feature # Return
     :return: (dict) section of auto_train config for input_features and output_features
     """
     targets = convert_targets(target_name)
@@ -379,10 +372,8 @@ def get_config_from_metadata(metadata: List[FieldMetadata], targets: Set[str] =
     """Builds input/output feature sections of auto-train config using field metadata.
 
     # Inputs
-    :param metadata: (List[FieldMetadata]) field descriptions
-    :param targets (Set[str]) names of target features
-
-    # Return
+    :param metadata: (List[FieldMetadata]) field descriptions :param targets (Set[str]) names of target features #
+        Return
     :return: (dict) section of auto_train config for input_features and output_features
     """
     config = {
@@ -405,10 +396,8 @@ def get_field_metadata(fields: List[FieldInfo], row_count: int, targets: Set[str
 
     # Inputs
     :param fields: (List[FieldInfo]) FieldInfo objects for all fields in dataset
-    :param row_count: (int) total number of entries in original dataset
-    :param targets (Set[str]) names of target features
-
-    # Return
+    :param row_count: (int) total number of entries in original dataset :param targets (Set[str]) names of target
+        features # Return
     :return: (List[FieldMetadata]) list of objects containing metadata for each field
     """
 
diff --git a/ludwig/backend/_ray210_compat.py b/ludwig/backend/_ray210_compat.py
index a05c64f3e20..afe1b705940 100644
--- a/ludwig/backend/_ray210_compat.py
+++ b/ludwig/backend/_ray210_compat.py
@@ -19,8 +19,8 @@
 class TunerRay210(Tuner):
     """HACK(geoffrey): This is a temporary fix to support Ray 2.1.0.
 
-    Specifically, this Tuner ensures that TunerInternalRay210 is called by the class.
-    For more details, see TunerInternalRay210.
+    Specifically, this Tuner ensures that TunerInternalRay210 is called by the class. For more details, see
+    TunerInternalRay210.
     """
 
     def __init__(
@@ -120,8 +120,9 @@ def restore(
 class TunerInternalRay210(TunerInternal):
     """HACK(geoffrey): This is a temporary fix to support Ray 2.1.0.
 
-    This TunerInternal ensures that a division by zero is avoided when running zero-CPU hyperopt trials.
-    This is fixed in ray>=2.2 (but not ray<=2.1) here: https://github.com/ray-project/ray/pull/30598
+    This TunerInternal ensures that a division by zero is avoided when running zero-CPU hyperopt trials. This is fixed
+    in ray>=2.2 (but not ray<=2.1) here:
+    https://github.com/ray-project/ray/pull/30598
     """
 
     def _expected_utilization(self, cpus_per_trial, cpus_total):
diff --git a/ludwig/backend/datasource.py b/ludwig/backend/datasource.py
index 8b67032c321..aa965da8463 100644
--- a/ludwig/backend/datasource.py
+++ b/ludwig/backend/datasource.py
@@ -88,8 +88,8 @@ def _open_input_source(
 
         The default implementation opens the source path as a sequential input stream.
 
-        Implementations that do not support streaming reads (e.g. that require random
-        access) should override this method.
+        Implementations that do not support streaming reads (e.g. that require random access) should override this
+        method.
         """
         if path is None or is_http(path):
             return contextlib.nullcontext()
diff --git a/ludwig/backend/deepspeed.py b/ludwig/backend/deepspeed.py
index 41ed3718863..b4661334ea6 100644
--- a/ludwig/backend/deepspeed.py
+++ b/ludwig/backend/deepspeed.py
@@ -17,7 +17,7 @@ def __init__(
         fp16: Optional[Dict[str, Any]] = None,
         bf16: Optional[Dict[str, Any]] = None,
         compression_training: Optional[Dict[str, Any]] = None,
-        **kwargs
+        **kwargs,
     ):
         super().__init__(**kwargs)
         self.zero_optimization = zero_optimization
diff --git a/ludwig/benchmarking/summary_dataclasses.py b/ludwig/benchmarking/summary_dataclasses.py
index af18e5bc80d..ba391b2c87b 100644
--- a/ludwig/benchmarking/summary_dataclasses.py
+++ b/ludwig/benchmarking/summary_dataclasses.py
@@ -8,7 +8,8 @@
 import ludwig.modules.metric_modules  # noqa: F401
 from ludwig.benchmarking.utils import format_memory, format_time
 from ludwig.globals import MODEL_FILE_NAME, MODEL_HYPERPARAMETERS_FILE_NAME
-from ludwig.modules.metric_registry import get_metric_classes, metric_feature_type_registry  # noqa: F401
+from ludwig.modules.metric_registry import get_metric_classes  # noqa: F401
+from ludwig.modules.metric_registry import metric_feature_type_registry
 from ludwig.types import ModelConfigDict
 from ludwig.utils.data_utils import load_json
 
diff --git a/ludwig/callbacks.py b/ludwig/callbacks.py
index b7c4673789d..3e08962e855 100644
--- a/ludwig/callbacks.py
+++ b/ludwig/callbacks.py
@@ -48,7 +48,7 @@ def on_preprocess_end(self, training_set, validation_set, test_set, training_set
         :param test_set: The test set.
         :type test_set: ludwig.dataset.base.Dataset
         :param training_set_metadata: Values inferred from the training set, including preprocessing settings,
-                                      vocabularies, feature statistics, etc. Same as training_set_metadata.json.
+            vocabularies, feature statistics, etc. Same as training_set_metadata.json.
         """
 
         pass
@@ -374,7 +374,6 @@ def prepare_ray_tune(self, train_fn: Callable, tune_config: Dict[str, Any], tune
         :param train_fn: The function which runs the experiment trial.
         :param tune_config: The ray tune configuration dictionary.
         :param tune_callbacks: List of callbacks (not used yet).
-
         :returns: Tuple[Callable, Dict] The train_fn and tune_config, which will be passed to ray tune.
         """
         return train_fn, tune_config
diff --git a/ludwig/config_validation/checks.py b/ludwig/config_validation/checks.py
index a15b44c4a45..9f77ad61ef9 100644
--- a/ludwig/config_validation/checks.py
+++ b/ludwig/config_validation/checks.py
@@ -358,7 +358,8 @@ def check_hyperopt_parameter_dicts(config: "ModelConfig") -> None:  # noqa: F821
     if config.hyperopt is None:
         return
 
-    from ludwig.schema.hyperopt.utils import get_parameter_cls, parameter_config_registry  # noqa: F401
+    from ludwig.schema.hyperopt.utils import get_parameter_cls  # noqa: F401
+    from ludwig.schema.hyperopt.utils import parameter_config_registry
 
     for parameter, space in config.hyperopt.parameters.items():
         # skip nested hyperopt parameters
diff --git a/ludwig/config_validation/validation.py b/ludwig/config_validation/validation.py
index 30725d7e49b..cd4e0dfadea 100644
--- a/ludwig/config_validation/validation.py
+++ b/ludwig/config_validation/validation.py
@@ -11,9 +11,11 @@
 
 # TODO(travis): figure out why we need these imports to avoid circular import error
 from ludwig.schema.combiners.utils import get_combiner_jsonschema  # noqa
-from ludwig.schema.features.utils import get_input_feature_jsonschema, get_output_feature_jsonschema  # noqa
+from ludwig.schema.features.utils import get_input_feature_jsonschema  # noqa
+from ludwig.schema.features.utils import get_output_feature_jsonschema
 from ludwig.schema.hyperopt import get_hyperopt_jsonschema  # noqa
-from ludwig.schema.trainer import get_model_type_jsonschema, get_trainer_jsonschema  # noqa
+from ludwig.schema.trainer import get_model_type_jsonschema  # noqa
+from ludwig.schema.trainer import get_trainer_jsonschema
 from ludwig.schema.utils import unload_jsonschema_from_marshmallow_class
 
 VALIDATION_LOCK = Lock()
diff --git a/ludwig/contrib.py b/ludwig/contrib.py
index d69085c5356..3c30bf6116f 100644
--- a/ludwig/contrib.py
+++ b/ludwig/contrib.py
@@ -12,7 +12,6 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 # ==============================================================================
-
 """Module for handling contributed support."""
 
 import argparse
diff --git a/ludwig/contribs/__init__.py b/ludwig/contribs/__init__.py
index dd823ed44e6..c8f7c2ebd68 100644
--- a/ludwig/contribs/__init__.py
+++ b/ludwig/contribs/__init__.py
@@ -12,7 +12,6 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 # ==============================================================================
-
 """All contrib classes must implement the `ludwig.callbacks.Callback` interface.
 
 If you don't want to handle the call, either provide an empty method with `pass`, or just don't implement the method.
diff --git a/ludwig/data/preprocessing.py b/ludwig/data/preprocessing.py
index 51887a7d6bf..3754f959bc5 100644
--- a/ludwig/data/preprocessing.py
+++ b/ludwig/data/preprocessing.py
@@ -2086,12 +2086,12 @@ def _preprocess_file_for_training(
 
     :param features: list of all features (input + output)
     :param dataset: path to the data
-    :param training_set:  training data
+    :param training_set: training data
     :param validation_set: validation data
     :param test_set: test data
     :param training_set_metadata: train set metadata
-    :param skip_save_processed_input: if False, the pre-processed data is saved
-    as .hdf5 files in the same location as the csv files with the same names.
+    :param skip_save_processed_input: if False, the pre-processed data is saved as .hdf5 files in the same location as
+        the csv files with the same names.
     :param preprocessing_params: preprocessing parameters
     :param random_seed: random seed
     :return: training, test, validation datasets, training metadata
diff --git a/ludwig/data/sampler.py b/ludwig/data/sampler.py
index 08487f7fc13..e062e9b0f40 100644
--- a/ludwig/data/sampler.py
+++ b/ludwig/data/sampler.py
@@ -64,8 +64,8 @@ def __len__(self):
     def set_epoch(self, epoch):
         """Sets the epoch for this sampler.
 
-        When `shuffle=True`, this ensures all replicas use a different random ordering
-        for each epoch. Otherwise, the next iteration of this sampler will yield the same ordering.
+        When `shuffle=True`, this ensures all replicas use a different random ordering for each epoch. Otherwise, the
+        next iteration of this sampler will yield the same ordering.
 
         :param epoch: (int) epoch number
         """
diff --git a/ludwig/datasets/__init__.py b/ludwig/datasets/__init__.py
index 16795366f41..dbe45169376 100644
--- a/ludwig/datasets/__init__.py
+++ b/ludwig/datasets/__init__.py
@@ -242,7 +242,7 @@ def get_datasets_output_features(
     :param include_competitions: (bool) whether to include the output features from kaggle competition datasets
     :param include_data_modalities: (bool) whether to include the data modalities associated with the prediction task
     :return: (dict) dictionary with the output features for each dataset or a dictionary with the output features for
-                    the specified dataset
+        the specified dataset
     """
     ordered_configs = OrderedDict(sorted(_get_dataset_configs().items()))
     competition_datasets = []
@@ -321,10 +321,8 @@ def _get_hf_dataset_and_subsample(dataset_name: str) -> Tuple[str, Optional[str]
 
     The dataset name should follow the format "{HF_PREFIX}{hf_id}--{hf_subsample}"
 
-    Examples (Dataset Name --> HF ID; HF subsample):
-    "hf://wikisql" --> "wikisql"; None
-    "hf://ColumbiaNLP/FLUTE" --> "ColumbiaNLP/FLUTE"; None
-    "hf://mstz/adult--income" --> "mstz/adult"; "income"
+    Examples (Dataset Name --> HF ID; HF subsample): "hf://wikisql" --> "wikisql"; None "hf://ColumbiaNLP/FLUTE" -->
+    "ColumbiaNLP/FLUTE"; None "hf://mstz/adult--income" --> "mstz/adult"; "income"
     """
     dataset_name = dataset_name[len(HF_PREFIX) :]
     dataset_name = dataset_name.split("--")
diff --git a/ludwig/datasets/loaders/mnist.py b/ludwig/datasets/loaders/mnist.py
index 28bc1efae33..5da5cc1c245 100644
--- a/ludwig/datasets/loaders/mnist.py
+++ b/ludwig/datasets/loaders/mnist.py
@@ -58,11 +58,8 @@ def load_unprocessed_dataframe(self, file_paths: List[str]) -> pd.DataFrame:
     def read_source_dataset(self, dataset="training", path="."):
         """Create a directory for training and test and extract all the images and labels to this destination.
 
-        :args:
-            dataset (str) : the label for the dataset
-            path (str): the raw dataset path
-        :returns:
-            A tuple of the label for the image, the file array, the size and rows and columns for the image
+        :args:     dataset (str) : the label for the dataset     path (str): the raw dataset path
+        :returns: A tuple of the label for the image, the file array, the size and rows and columns for the image
         """
         if dataset == "training":
             fname_img = os.path.join(path, "train-images-idx3-ubyte")
@@ -87,13 +84,9 @@ def read_source_dataset(self, dataset="training", path="."):
     def write_output_dataset(self, labels, images, output_dir):
         """Create output directories where we write out the images.
 
-        :args:
-            labels (str) : the labels for the image
-            data (np.array) : the binary array corresponding to the image
-            output_dir (str) : the output directory that we need to write to
-            path (str): the raw dataset path
-        :returns:
-            A tuple of the label for the image, the file array, the size and rows and columns for the image
+        :args:     labels (str) : the labels for the image     data (np.array) : the binary array corresponding to the
+        image     output_dir (str) : the output directory that we need to write to     path (str): the raw dataset path
+        :returns: A tuple of the label for the image, the file array, the size and rows and columns for the image
         """
         # create child image output directories
         output_dirs = [os.path.join(output_dir, str(i)) for i in range(NUM_LABELS)]
diff --git a/ludwig/datasets/loaders/split_loaders.py b/ludwig/datasets/loaders/split_loaders.py
index 19963c83cd3..f83f605ce15 100644
--- a/ludwig/datasets/loaders/split_loaders.py
+++ b/ludwig/datasets/loaders/split_loaders.py
@@ -21,10 +21,12 @@
 
 class RandomSplitLoader(DatasetLoader):
     """Adds a random split column to the dataset, with fixed proportions of:
-     train: 70%
+
+    train: 70%
      validation: 10%
      test: 20%
-    ."""
+    .
+    """
 
     def transform_dataframe(self, dataframe: pd.DataFrame) -> pd.DataFrame:
         df = super().transform_dataframe(dataframe)
diff --git a/ludwig/decoders/llm_decoders.py b/ludwig/decoders/llm_decoders.py
index eafc84bacc1..5763f5a5868 100644
--- a/ludwig/decoders/llm_decoders.py
+++ b/ludwig/decoders/llm_decoders.py
@@ -1,3 +1,4 @@
+# flake8: noqa: E501
 import logging
 import re
 from typing import Any, Dict, List, Union
@@ -91,7 +92,6 @@ def __init__(
             # Transformer Tokenizers
             self.tokenizer_vocab_size = self.tokenizer.tokenizer.vocab_size
         else:
-            # TorchText Tokenizers
             self.tokenizer_vocab_size = len(self.tokenizer.vocab)
 
         # Maximum number of new tokens that will be generated
diff --git a/ludwig/distributed/_ray_210_compat.py b/ludwig/distributed/_ray_210_compat.py
index 59dd2962b5f..e0adfb5512f 100644
--- a/ludwig/distributed/_ray_210_compat.py
+++ b/ludwig/distributed/_ray_210_compat.py
@@ -8,8 +8,7 @@
 class HorovodTrainerRay210(HorovodTrainer):
     """HACK(geoffrey): This is a temporary fix to support Ray 2.1.0.
 
-    Specifically, this Trainer ensures that TunerRay210 is called by the class.
-    For more details, see TunerRay210.
+    Specifically, this Trainer ensures that TunerRay210 is called by the class. For more details, see TunerRay210.
     """
 
     def fit(self) -> Result:
diff --git a/ludwig/distributed/deepspeed.py b/ludwig/distributed/deepspeed.py
index 9b3a04b5135..a5677f66538 100644
--- a/ludwig/distributed/deepspeed.py
+++ b/ludwig/distributed/deepspeed.py
@@ -47,7 +47,7 @@ def __init__(
         fp16: Optional[Dict[str, Any]] = None,
         bf16: Optional[Dict[str, Any]] = None,
         compression_training: Optional[Dict[str, Any]] = None,
-        **kwargs
+        **kwargs,
     ):
         # If we're initializing from a `deepspeed` CLI command, deepspeed will have already been initialized, as
         # indicated by the presence of the LOCAL_RANK var. Otherwise, we're initializing from Ray / torchrun, and will
diff --git a/ludwig/experiment.py b/ludwig/experiment.py
index 73e5a3fafa0..7615a4f9783 100644
--- a/ludwig/experiment.py
+++ b/ludwig/experiment.py
@@ -257,15 +257,12 @@ def kfold_cross_validate_cli(
 
     # Inputs
     :param k_fold: (int) number of folds to create for the cross-validation
-    :param config: (Union[str, dict], default: None) a dictionary or file path
-            containing model configuration. Refer to the [User Guide]
-           (http://ludwig.ai/user_guide/#model-config) for details.
+    :param config: (Union[str, dict], default: None) a dictionary or file path containing model configuration. Refer to
+        the [User Guide] (http://ludwig.ai/user_guide/#model-config) for details.
     :param dataset: (string, default: None)
     :param output_directory: (string, default: 'results')
     :param random_seed: (int) Random seed used k-fold splits.
-    :param skip_save_k_fold_split_indices: (boolean, default: False) Disables
-            saving k-fold split indices
-
+    :param skip_save_k_fold_split_indices: (boolean, default: False) Disables saving k-fold split indices
     :return: None
     """
 
diff --git a/ludwig/explain/captum.py b/ludwig/explain/captum.py
index 081568e18f7..643d5d2ddc3 100644
--- a/ludwig/explain/captum.py
+++ b/ludwig/explain/captum.py
@@ -273,10 +273,7 @@ def get_input_tensors(
     # Inputs
 
     :param model: The LudwigModel to use for encoding.
-    :param input_set: The input data to encode of shape [batch size, num input features].
-
-    # Return
-
+    :param input_set: The input data to encode of shape [batch size, num input features].  # Return
     :return: A list of variables, one for each input feature. Shape of each variable is [batch size, embedding size].
     """
     # Ignore sample_ratio and sample_size from the model config, since we want to explain all the data.
diff --git a/ludwig/features/image_feature.py b/ludwig/features/image_feature.py
index c9c8416b8b8..a1488440f41 100644
--- a/ludwig/features/image_feature.py
+++ b/ludwig/features/image_feature.py
@@ -465,27 +465,20 @@ def _read_image_if_bytes_obj_and_resize(
         standardize_image: str,
         channel_class_map: torch.Tensor,
     ) -> Optional[np.ndarray]:
-        """
-        :param img_entry Union[bytes, torch.Tensor, np.ndarray, str]: if str file path to the
-            image else torch.Tensor of the image itself
-        :param img_width: expected width of the image
-        :param img_height: expected height of the image
-        :param should_resize: Should the image be resized?
-        :param resize_method: type of resizing method
-        :param num_channels: expected number of channels in the first image
-        :param user_specified_num_channels: did the user specify num channels?
-        :param standardize_image: specifies whether to standarize image with imagenet1k specifications
-        :param channel_class_map: A tensor mapping channel values to classes, where dim=0 is the class
-        :return: image object as a numpy array
-
-        Helper method to read and resize an image according to model definition.
-        If the user doesn't specify a number of channels, we use the first image
-        in the dataset as the source of truth. If any image in the dataset
-        doesn't have the same number of channels as the first image,
-        raise an exception.
-
-        If the user specifies a number of channels, we try to convert all the
-        images to the specifications by dropping channels/padding 0 channels
+        """:param img_entry Union[bytes, torch.Tensor, np.ndarray, str]: if str file path to the image else
+        torch.Tensor of the image itself :param img_width: expected width of the image :param img_height: expected
+        height of the image :param should_resize: Should the image be resized? :param resize_method: type of
+        resizing method :param num_channels: expected number of channels in the first image :param
+        user_specified_num_channels: did the user specify num channels? :param standardize_image: specifies whether
+        to standarize image with imagenet1k specifications :param channel_class_map: A tensor mapping channel
+        values to classes, where dim=0 is the class :return: image object as a numpy array.
+
+        Helper method to read and resize an image according to model definition. If the user doesn't specify a number of
+        channels, we use the first image in the dataset as the source of truth. If any image in the dataset doesn't have
+        the same number of channels as the first image, raise an exception.
+
+        If the user specifies a number of channels, we try to convert all the images to the specifications by dropping
+        channels/padding 0 channels
         """
 
         if isinstance(img_entry, bytes):
diff --git a/ludwig/models/llm.py b/ludwig/models/llm.py
index 862ee55ff4a..3a970cd2013 100644
--- a/ludwig/models/llm.py
+++ b/ludwig/models/llm.py
@@ -364,7 +364,7 @@ def is_merge_and_unload_set(self) -> bool:
 
         # Return
 
-            :return (bool): whether merge_and_unload should be done.
+        :return (bool): whether merge_and_unload should be done.
         """
         return (
             self.config_obj.adapter is not None
diff --git a/ludwig/preprocess.py b/ludwig/preprocess.py
index f4b427bd076..4b268d6482e 100644
--- a/ludwig/preprocess.py
+++ b/ludwig/preprocess.py
@@ -45,11 +45,10 @@ def preprocess_cli(
     logging_level: int = logging.INFO,
     callbacks: List[Callback] = None,
     backend: Union[Backend, str] = None,
-    **kwargs
+    **kwargs,
 ) -> None:
-    """*train* defines the entire training procedure used by Ludwig's
-    internals. Requires most of the parameters that are taken into the model.
-    Builds a full ludwig model and performs the training.
+    """*train* defines the entire training procedure used by Ludwig's internals. Requires most of the parameters
+    that are taken into the model. Builds a full ludwig model and performs the training.
 
     :param preprocessing_config: (Union[str, dict]) in-memory representation of
             config or string path to a YAML config file.
diff --git a/ludwig/schema/__init__.py b/ludwig/schema/__init__.py
index 77ecd60c32c..b217aab6a4b 100644
--- a/ludwig/schema/__init__.py
+++ b/ludwig/schema/__init__.py
@@ -1,5 +1,7 @@
 # TODO(travis): figure out why we need these imports to avoid circular import error
 from ludwig.schema.combiners.utils import get_combiner_jsonschema  # noqa
-from ludwig.schema.features.utils import get_input_feature_jsonschema, get_output_feature_jsonschema  # noqa
+from ludwig.schema.features.utils import get_input_feature_jsonschema  # noqa
+from ludwig.schema.features.utils import get_output_feature_jsonschema
 from ludwig.schema.hyperopt import get_hyperopt_jsonschema  # noqa
-from ludwig.schema.trainer import get_model_type_jsonschema, get_trainer_jsonschema  # noqa
+from ludwig.schema.trainer import get_model_type_jsonschema  # noqa
+from ludwig.schema.trainer import get_trainer_jsonschema
diff --git a/ludwig/schema/features/loss/__init__.py b/ludwig/schema/features/loss/__init__.py
index e7ee0534df8..249d91e87bc 100644
--- a/ludwig/schema/features/loss/__init__.py
+++ b/ludwig/schema/features/loss/__init__.py
@@ -1 +1,2 @@
-from ludwig.schema.features.loss.loss import get_loss_classes, get_loss_cls, get_loss_schema_registry  # noqa
+from ludwig.schema.features.loss.loss import get_loss_classes  # noqa
+from ludwig.schema.features.loss.loss import get_loss_cls, get_loss_schema_registry
diff --git a/ludwig/schema/features/utils.py b/ludwig/schema/features/utils.py
index 34abd2eee15..0ce5ce20dbb 100644
--- a/ludwig/schema/features/utils.py
+++ b/ludwig/schema/features/utils.py
@@ -18,16 +18,16 @@
 
 input_mixin_registry = Registry()
 output_mixin_registry = Registry()
+"""As of Ludwig v0.7, ECD models support the full range of feature parameters available in Ludwig, so any feature
+schema can be registered into it.
 
-"""
-As of Ludwig v0.7, ECD models support the full range of feature parameters available in Ludwig, so any feature schema
-can be registered into it. See `BinaryDefaultsConfig` for an example.
+See `BinaryDefaultsConfig` for an example.
 """
 ecd_defaults_config_registry = Registry()
+"""As of Ludwig v0.7, GBM models only support certain feature types and those features may only contain
+preprocessing parameters (in comparison, ECD features can specify encoders and other parameters).
 
-"""
-As of Ludwig v0.7, GBM models only support certain feature types and those features may only contain preprocessing
-parameters (in comparison, ECD features can specify encoders and other parameters). This is why the two model types have
+This is why the two model types have
 separate defaults registries. See `BinaryInputFeatureConfigMixin` for an example of a schema pattern that is designed to
 be registered by this registry (whereas, conversely, `BinaryDefaultsConfig` is an example of one to be registered with
 the ECD defaults registry).
diff --git a/ludwig/schema/llms/model_parameters.py b/ludwig/schema/llms/model_parameters.py
index 9c8ef70b02c..aad909eec56 100644
--- a/ludwig/schema/llms/model_parameters.py
+++ b/ludwig/schema/llms/model_parameters.py
@@ -12,10 +12,9 @@ class RoPEScalingConfig(schema_utils.BaseMarshmallowConfig):
     """Dynamic RoPE-scaling (rotary position embeddings) to extend the context length of LLM like LLaMA, GPT-NeoX,
     or Falcon.
 
-    This parameter is a dictionary containing the scaling configuration
-    for the RoPE embeddings. Currently supports three scaling strategies: linear and dynamic. Their
-    scaling factor must be an float greater than 1. The expected format is {'type': strategy name,
-    'factor': scaling factor}
+    This parameter is a dictionary containing the scaling configuration for the RoPE embeddings. Currently supports
+    three scaling strategies: linear and dynamic. Their scaling factor must be an float greater than 1. The expected
+    format is {'type': strategy name, 'factor': scaling factor}
     """
 
     def __post_init__(self):
diff --git a/ludwig/schema/split.py b/ludwig/schema/split.py
index 36410ad467e..25779be054d 100644
--- a/ludwig/schema/split.py
+++ b/ludwig/schema/split.py
@@ -122,8 +122,8 @@ class DateTimeSplitConfig(BaseSplitConfig):
 class HashSplitConfig(BaseSplitConfig):
     """This Dataclass generates a schema for the hash splitting config.
 
-    This is useful for deterministically splitting on a unique ID. Even when additional rows are added to the dataset
-    in the future, each ID will retain its original split assignment.
+    This is useful for deterministically splitting on a unique ID. Even when additional rows are added to the dataset in
+    the future, each ID will retain its original split assignment.
 
     This approach does not guarantee that the split proportions will be assigned exactly, but the larger the dataset,
     the more closely the assignment should match the given proportions.
diff --git a/ludwig/schema/trainer.py b/ludwig/schema/trainer.py
index 322e4df2b06..a164e86fed5 100644
--- a/ludwig/schema/trainer.py
+++ b/ludwig/schema/trainer.py
@@ -1,8 +1,10 @@
+# flake8: noqa: E501
 import re
 from abc import ABC
 from typing import Optional, Type, Union
 
 import torch
+import torch._dynamo
 from packaging.version import parse as parse_version
 
 from ludwig.api_annotations import DeveloperAPI
@@ -33,6 +35,9 @@
 
 _torch_200 = parse_version(torch.__version__) >= parse_version("2.0")
 
+# this is a workarrpund to avoid an error regarding torch.compile.
+# TODO Fix torch.compile and dynamo problems
+torch._dynamo.config.suppress_errors = True
 
 trainer_schema_registry = Registry()
 _llm_trainer_schema_registry = Registry()
diff --git a/ludwig/schema/utils.py b/ludwig/schema/utils.py
index dc876c98490..cd049ea44ba 100644
--- a/ludwig/schema/utils.py
+++ b/ludwig/schema/utils.py
@@ -2,7 +2,7 @@
 import os
 import warnings
 from abc import ABC, abstractmethod
-from dataclasses import field, Field
+from dataclasses import Field, field
 from functools import lru_cache
 from typing import Any
 from typing import Dict as TDict
diff --git a/ludwig/train.py b/ludwig/train.py
index fb0f28ff003..37e6bb6f618 100644
--- a/ludwig/train.py
+++ b/ludwig/train.py
@@ -59,11 +59,10 @@ def train_cli(
     backend: Union[Backend, str] = None,
     random_seed: int = default_random_seed,
     logging_level: int = logging.INFO,
-    **kwargs
+    **kwargs,
 ) -> None:
-    """*train* defines the entire training procedure used by Ludwig's
-    internals. Requires most of the parameters that are taken into the model.
-    Builds a full ludwig model and performs the training.
+    """*train* defines the entire training procedure used by Ludwig's internals. Requires most of the parameters
+    that are taken into the model. Builds a full ludwig model and performs the training.
 
     :param config: (Union[str, dict]) in-memory representation of
             config or string path to a YAML config file.
diff --git a/ludwig/trainers/trainer_lightgbm.py b/ludwig/trainers/trainer_lightgbm.py
index 74f0df1d98d..d15982d71a9 100644
--- a/ludwig/trainers/trainer_lightgbm.py
+++ b/ludwig/trainers/trainer_lightgbm.py
@@ -419,8 +419,7 @@ def check_progress_on_validation(
     ) -> bool:
         """Checks the history of validation scores.
 
-        Uses history of validation scores to decide whether training
-        should stop.
+        Uses history of validation scores to decide whether training should stop.
 
         Saves the model if scores have improved.
         """
diff --git a/ludwig/utils/automl/type_inference.py b/ludwig/utils/automl/type_inference.py
index d28cfbd56e2..f55f96e782f 100644
--- a/ludwig/utils/automl/type_inference.py
+++ b/ludwig/utils/automl/type_inference.py
@@ -22,9 +22,7 @@ def infer_type(field: FieldInfo, missing_value_percent: float, row_count: int) -
     # Inputs
     :param field: (FieldInfo) object describing field
     :param missing_value_percent: (float) percent of missing values in the column
-    :param row_count: (int) total number of entries in original dataset
-
-    # Return
+    :param row_count: (int) total number of entries in original dataset  # Return
     :return: (str) feature type
     """
     if field.dtype == DATE or field.dtype.startswith("datetime"):
diff --git a/ludwig/utils/calibration.py b/ludwig/utils/calibration.py
index 3ecdc099708..6b5458df5f3 100644
--- a/ludwig/utils/calibration.py
+++ b/ludwig/utils/calibration.py
@@ -304,7 +304,8 @@ def regularization_terms(self) -> torch.Tensor:
         """Off-Diagonal and Intercept Regularisation (ODIR).
 
         Described in "Beyond temperature scaling: Obtaining well-calibrated multiclass probabilities with Dirichlet
-        calibration" https://proceedings.neurips.cc/paper/2019/file/8ca01ea920679a0fe3728441494041b9-Paper.pdf
+        calibration"
+        https://proceedings.neurips.cc/paper/2019/file/8ca01ea920679a0fe3728441494041b9-Paper.pdf
         """
         off_diagonal_entries = torch.masked_select(self.w, ~torch.eye(self.num_classes, dtype=bool))
         weight_matrix_loss = self.off_diagonal_l2 * torch.linalg.vector_norm(off_diagonal_entries)
diff --git a/ludwig/utils/data_utils.py b/ludwig/utils/data_utils.py
index 471605915a8..b06f753a174 100644
--- a/ludwig/utils/data_utils.py
+++ b/ludwig/utils/data_utils.py
@@ -700,14 +700,12 @@ def class_counts(dataset, labels_field):
 def load_from_file(file_name, field=None, dtype=int, ground_truth_split=2):
     """Load experiment data from supported file formats.
 
-    Experiment data can be test/train statistics, model predictions,
-    probability, ground truth,  ground truth metadata.
+    Experiment data can be test/train statistics, model predictions, probability, ground truth,  ground truth metadata.
     :param file_name: Path to file to be loaded
     :param field: Target Prediction field.
     :param dtype:
-    :param ground_truth_split: Ground truth split filter where 0 is train 1 is
-    validation and 2 is test split. By default test split is used when loading
-    ground truth from hdf5.
+    :param ground_truth_split: Ground truth split filter where 0 is train 1 is validation and 2 is test split. By
+        default test split is used when loading ground truth from hdf5.
     :return: Experiment data as array
     """
     if file_name.endswith(".hdf5") and field is not None:
@@ -753,7 +751,7 @@ def add_sequence_feature_column(df, col_name, seq_length):
     delimited strings composed of preceding values of the same column up to seq_length. For example values of the
     i-th row of the new column will be a space-delimited string of df[col_name][i-seq_length].
 
-     :param df: input dataframe
+    :param df: input dataframe
     :param col_name: column name containing sequential data
     :param seq_length: length of an array of preceeding column values to use
     """
diff --git a/ludwig/utils/entmax/__init__.py b/ludwig/utils/entmax/__init__.py
index 7eb4162ff95..882b5938146 100644
--- a/ludwig/utils/entmax/__init__.py
+++ b/ludwig/utils/entmax/__init__.py
@@ -1,6 +1,6 @@
 __version__ = "1.1.dev0"
 
-from ludwig.utils.entmax.activations import entmax15, Entmax15, sparsemax, Sparsemax
+from ludwig.utils.entmax.activations import Entmax15, entmax15, Sparsemax, sparsemax
 from ludwig.utils.entmax.losses import (
     entmax15_loss,
     Entmax15Loss,
diff --git a/ludwig/utils/horovod_utils.py b/ludwig/utils/horovod_utils.py
index f4482596a00..a8f6106e8ef 100644
--- a/ludwig/utils/horovod_utils.py
+++ b/ludwig/utils/horovod_utils.py
@@ -45,14 +45,13 @@ def has_horovodrun():
 def gather_all_tensors(result: torch.Tensor, group: Optional[Any] = None) -> List[torch.Tensor]:
     """Function to gather all tensors from several processes onto a list that is broadcast to all processes.
 
-    Works on tensors that have the same number of dimensions, but where each dimension may differ. In this case
-    tensors are padded, gathered and then trimmed to secure equal workload for all processes.
+    Works on tensors that have the same number of dimensions, but where each dimension may differ. In this case tensors
+    are padded, gathered and then trimmed to secure equal workload for all processes.
 
     :param result: the value to sync
     :param group: the process group to gather results from (not supported: always uses world)
-
-    :return: list with size equal to the process group where gathered_result[i]
-             corresponds to result tensor from process i
+    :return: list with size equal to the process group where gathered_result[i] corresponds to result tensor from
+        process i
     """
     if group is not None:
         raise ValueError("Horovod does not support allgather using a subcommunicator at this time. " "Unset `group`.")
diff --git a/ludwig/utils/image_utils.py b/ludwig/utils/image_utils.py
index 6395a2cbe17..a2fae951777 100644
--- a/ludwig/utils/image_utils.py
+++ b/ludwig/utils/image_utils.py
@@ -208,7 +208,7 @@ def pad(
     img: torch.Tensor,
     new_size: Union[int, Tuple[int, int]],
 ) -> torch.Tensor:
-    """torchscript-compatible implementation of pad.
+    """Torchscript-compatible implementation of pad.
 
     Args:
         img (torch.Tensor): image with shape [..., height, width] to pad
@@ -231,7 +231,7 @@ def crop(
     img: torch.Tensor,
     new_size: Union[int, Tuple[int, int]],
 ) -> torch.Tensor:
-    """torchscript-compatible implementation of crop.
+    """Torchscript-compatible implementation of crop.
 
     Args:
         img (torch.Tensor): image with shape [..., height, width] to crop
@@ -246,7 +246,7 @@ def crop(
 
 @DeveloperAPI
 def crop_or_pad(img: torch.Tensor, new_size: Union[int, Tuple[int, int]]):
-    """torchscript-compatible implementation of resize using constants.CROP_OR_PAD.
+    """Torchscript-compatible implementation of resize using constants.CROP_OR_PAD.
 
     Args:
         img (torch.Tensor): image with shape [..., height, width] to resize
@@ -271,7 +271,7 @@ def resize_image(
     crop_or_pad_constant: str = CROP_OR_PAD,
     interpolate_constant: str = INTERPOLATE,
 ) -> torch.Tensor:
-    """torchscript-compatible implementation of resize.
+    """Torchscript-compatible implementation of resize.
 
     Args:
         img (torch.Tensor): image with shape [..., height, width] to resize
@@ -442,9 +442,8 @@ def to_tuple(v: Union[int, Tuple[int, int]]) -> Tuple[int, int]:
 def to_np_tuple(prop: Union[int, Iterable]) -> np.ndarray:
     """Creates a np array of length 2 from a Conv2D property.
 
-    E.g., stride=(2, 3) gets converted into np.array([2, 3]), where the
-    height_stride = 2 and width_stride = 3. stride=2 gets converted into
-    np.array([2, 2]).
+    E.g., stride=(2, 3) gets converted into np.array([2, 3]), where the height_stride = 2 and width_stride = 3. stride=2
+    gets converted into np.array([2, 2]).
     """
     if type(prop) is int:
         return np.ones(2).astype(int) * prop
diff --git a/ludwig/utils/llm_utils.py b/ludwig/utils/llm_utils.py
index 29452237e33..a29ef380868 100644
--- a/ludwig/utils/llm_utils.py
+++ b/ludwig/utils/llm_utils.py
@@ -158,7 +158,8 @@ def initialize_adapter(
             logger.info(f"Using pretrained adapter weights: {config_obj.adapter.pretrained_adapter_weights}")
 
             # Leave this import inline to support a minimal install of Ludwig
-            from peft import MODEL_TYPE_TO_PEFT_MODEL_MAPPING, PeftConfig  # noqa
+            from peft import MODEL_TYPE_TO_PEFT_MODEL_MAPPING  # noqa
+            from peft import PeftConfig
 
             peft_config = PeftConfig.from_pretrained(config_obj.adapter.pretrained_adapter_weights)
 
diff --git a/ludwig/utils/tokenizers.py b/ludwig/utils/tokenizers.py
index 6fb2d2117a4..99cde68d51a 100644
--- a/ludwig/utils/tokenizers.py
+++ b/ludwig/utils/tokenizers.py
@@ -15,24 +15,16 @@
 
 import logging
 from abc import abstractmethod
-from typing import Any, Dict, List, Optional, Union
+from typing import Any, List, Union
 
 import torch
-import torchtext
 
-from ludwig.constants import PADDING_SYMBOL, UNKNOWN_SYMBOL
-from ludwig.utils.data_utils import load_json
 from ludwig.utils.hf_utils import load_pretrained_hf_tokenizer
 from ludwig.utils.nlp_utils import load_nlp_pipeline, process_text
 
 logger = logging.getLogger(__name__)
-torchtext_version = torch.torch_version.TorchVersion(torchtext.__version__)
 
 TORCHSCRIPT_COMPATIBLE_TOKENIZERS = {"space", "space_punct", "comma", "underscore", "characters"}
-TORCHTEXT_0_12_0_TOKENIZERS = {"sentencepiece", "clip", "gpt2bpe"}
-TORCHTEXT_0_13_0_TOKENIZERS = {"bert"}
-
-HF_TOKENIZER_SAMPLE_INPUTS = ["UNwant\u00E9d,running", "ah\u535A\u63A8zz", " \tHeLLo!how  \n Are yoU? [UNK]"]
 
 
 class BaseTokenizer:
@@ -140,9 +132,25 @@ def __init__(self, ngram_size: int = 2, **kwargs):
         self.n = ngram_size or 2
 
     def get_tokens(self, tokens: List[str]) -> List[str]:
-        from torchtext.data.utils import ngrams_iterator
+        return list(self._ngrams_iterator(tokens, ngrams=self.n))
+
+    def _ngrams_iterator(self, token_list, ngrams):
+        """Return an iterator that yields the given tokens and their ngrams. This code is taken from
+        https://pytorch.org/text/stable/_modules/torchtext/data/utils.html#ngrams_iterator.
+
+        Args:
+            token_list: A list of tokens
+            ngrams: the number of ngrams.
+        """
+
+        def _get_ngrams(n):
+            return zip(*[token_list[i:] for i in range(n)])
 
-        return list(ngrams_iterator(tokens, ngrams=self.n))
+        for x in token_list:
+            yield x
+        for n in range(2, ngrams + 1):
+            for x in _get_ngrams(n):
+                yield " ".join(x)
 
 
 class SpacePunctuationStringToListTokenizer(torch.nn.Module):
@@ -897,7 +905,7 @@ def convert_token_to_id(self, token: str) -> int:
 
 
 tokenizer_registry = {
-    # Torchscript-compatible tokenizers. Torchtext tokenizers are also available below (requires torchtext>=0.12.0).
+    # Torchscript-compatible tokenizers.
     "space": SpaceStringToListTokenizer,
     "space_punct": SpacePunctuationStringToListTokenizer,
     "ngram": NgramTokenizer,
@@ -1005,233 +1013,40 @@ def convert_token_to_id(self, token: str) -> int:
     "multi_lemmatize_remove_stopwords": MultiLemmatizeRemoveStopwordsTokenizer,
 }
 
-"""torchtext 0.12.0 tokenizers.
-
-Only available with torchtext>=0.12.0.
-"""
-
-
-class SentencePieceTokenizer(torch.nn.Module):
-    def __init__(self, pretrained_model_name_or_path: Optional[str] = None, **kwargs):
-        super().__init__()
-        if pretrained_model_name_or_path is None:
-            pretrained_model_name_or_path = "https://download.pytorch.org/models/text/xlmr.sentencepiece.bpe.model"
-        self.tokenizer = torchtext.transforms.SentencePieceTokenizer(sp_model_path=pretrained_model_name_or_path)
-
-    def forward(self, v: Union[str, List[str], torch.Tensor]):
-        if isinstance(v, torch.Tensor):
-            raise ValueError(f"Unsupported input: {v}")
-        return self.tokenizer(v)
-
-
-class _BPETokenizer(torch.nn.Module):
-    """Superclass for tokenizers that use BPE, such as CLIPTokenizer and GPT2BPETokenizer."""
-
-    def __init__(self, pretrained_model_name_or_path: str, vocab_file: str):
-        super().__init__()
-        self.str2idx, self.idx2str = self._init_vocab(vocab_file)
-        self.tokenizer = self._init_tokenizer(pretrained_model_name_or_path, vocab_file)
-
-    def _init_vocab(self, vocab_file: str) -> Dict[str, str]:
-        """Loads the vocab from the vocab file."""
-        str2idx = load_json(torchtext.utils.get_asset_local_path(vocab_file))
-        _, idx2str = zip(*sorted((v, k) for k, v in str2idx.items()))
-        return str2idx, idx2str
-
-    def _init_tokenizer(self, pretrained_model_name_or_path: str, vocab_file: str) -> Any:
-        """Initializes and returns the tokenizer."""
-        raise NotImplementedError
-
-    def forward(self, v: Union[str, List[str], torch.Tensor]) -> Any:
-        """Implements forward pass for tokenizer.
-
-        BPE tokenizers from torchtext return ids directly, which is inconsistent with the Ludwig tokenizer API. The
-        below implementation works around this by converting the ids back to their original string tokens.
-        """
-        if isinstance(v, torch.Tensor):
-            raise ValueError(f"Unsupported input: {v}")
-
-        inputs: List[str] = []
-        # Ludwig calls map on List[str] objects, so we need to handle individual strings as well.
-        if isinstance(v, str):
-            inputs.append(v)
-        else:
-            inputs.extend(v)
-
-        token_ids = self.tokenizer(inputs)
-        assert torch.jit.isinstance(token_ids, List[List[str]])
-
-        tokens = [[self.idx2str[int(unit_idx)] for unit_idx in sequence] for sequence in token_ids]
-        return tokens[0] if isinstance(v, str) else tokens
 
-    def get_vocab(self) -> Dict[str, str]:
-        return self.str2idx
-
-
-class CLIPTokenizer(_BPETokenizer):
-    def __init__(self, pretrained_model_name_or_path: Optional[str] = None, vocab_file: Optional[str] = None, **kwargs):
-        if pretrained_model_name_or_path is None:
-            pretrained_model_name_or_path = "http://download.pytorch.org/models/text/clip_merges.bpe"
-        if vocab_file is None:
-            vocab_file = "http://download.pytorch.org/models/text/clip_encoder.json"
-        super().__init__(pretrained_model_name_or_path, vocab_file)
-
-    def _init_tokenizer(self, pretrained_model_name_or_path: str, vocab_file: str):
-        return torchtext.transforms.CLIPTokenizer(
-            encoder_json_path=vocab_file, merges_path=pretrained_model_name_or_path
-        )
-
-
-class GPT2BPETokenizer(_BPETokenizer):
-    def __init__(self, pretrained_model_name_or_path: Optional[str] = None, vocab_file: Optional[str] = None, **kwargs):
-        if pretrained_model_name_or_path is None:
-            pretrained_model_name_or_path = "https://download.pytorch.org/models/text/gpt2_bpe_vocab.bpe"
-        if vocab_file is None:
-            vocab_file = "https://download.pytorch.org/models/text/gpt2_bpe_encoder.json"
-        super().__init__(pretrained_model_name_or_path, vocab_file)
-
-    def _init_tokenizer(self, pretrained_model_name_or_path: str, vocab_file: str):
-        return torchtext.transforms.GPT2BPETokenizer(
-            encoder_json_path=vocab_file, vocab_bpe_path=pretrained_model_name_or_path
-        )
+class HFTokenizerShortcutFactory:
+    """This factory can be used to build HuggingFace tokenizers form a shortcut string.
 
+    Those shortcuts were originally used for torchtext tokenizers. They also guarantee backward compatibility.
+    """
 
-tokenizer_registry.update(
-    {
-        "sentencepiece": SentencePieceTokenizer,
-        "clip": CLIPTokenizer,
-        "gpt2bpe": GPT2BPETokenizer,
+    MODELS = {
+        "sentencepiece": "FacebookAI/xlm-roberta-base",
+        "clip": "openai/clip-vit-base-patch32",
+        "gpt2bpe": "openai-community/gpt2",
+        "bert": "bert-base-uncased",
     }
-)
-TORCHSCRIPT_COMPATIBLE_TOKENIZERS.update(TORCHTEXT_0_12_0_TOKENIZERS)
-
-
-class BERTTokenizer(torch.nn.Module):
-    def __init__(
-        self,
-        vocab_file: Optional[str] = None,
-        is_hf_tokenizer: Optional[bool] = False,
-        hf_tokenizer_attrs: Optional[Dict[str, Any]] = None,
-        **kwargs,
-    ):
-        super().__init__()
-
-        if vocab_file is None:
-            # If vocab_file not passed in, use default "bert-base-uncased" vocab and kwargs.
-            kwargs = _get_bert_config("bert-base-uncased")
-            vocab_file = kwargs["vocab_file"]
-            vocab = self._init_vocab(vocab_file)
-            hf_tokenizer_attrs = {
-                "pad_token": "[PAD]",
-                "unk_token": "[UNK]",
-                "sep_token_id": vocab["[SEP]"],
-                "cls_token_id": vocab["[CLS]"],
-            }
-        else:
-            vocab = self._init_vocab(vocab_file)
-
-        self.vocab = vocab
-
-        self.is_hf_tokenizer = is_hf_tokenizer
-        if self.is_hf_tokenizer:
-            # Values used by Ludwig extracted from the corresponding HF model.
-            self.pad_token = hf_tokenizer_attrs["pad_token"]  # Used as padding symbol
-            self.unk_token = hf_tokenizer_attrs["unk_token"]  # Used as unknown symbol
-            self.cls_token_id = hf_tokenizer_attrs["cls_token_id"]  # Used as start symbol. Only used if HF.
-            self.sep_token_id = hf_tokenizer_attrs["sep_token_id"]  # Used as stop symbol. Only used if HF.
-            self.never_split = hf_tokenizer_attrs["all_special_tokens"]
-        else:
-            self.pad_token = PADDING_SYMBOL
-            self.unk_token = UNKNOWN_SYMBOL
-            self.cls_token_id = None
-            self.sep_token_id = None
-            self.never_split = [UNKNOWN_SYMBOL]
-
-        tokenizer_kwargs = {}
-        if "do_lower_case" in kwargs:
-            tokenizer_kwargs["do_lower_case"] = kwargs["do_lower_case"]
-        if "strip_accents" in kwargs:
-            tokenizer_kwargs["strip_accents"] = kwargs["strip_accents"]
-
-        # Return tokens as raw tokens only if not being used as a HF tokenizer.
-        self.return_tokens = not self.is_hf_tokenizer
 
-        tokenizer_init_kwargs = {
-            **tokenizer_kwargs,
-            "vocab_path": vocab_file,
-            "return_tokens": self.return_tokens,
-        }
-        if torchtext_version >= (0, 14, 0):
-            # never_split kwarg added in torchtext 0.14.0
-            tokenizer_init_kwargs["never_split"] = self.never_split
+    @classmethod
+    def create_class(cls, model_name: str):
+        """Creating a tokenizer class from a model name."""
 
-        self.tokenizer = torchtext.transforms.BERTTokenizer(**tokenizer_init_kwargs)
+        class DynamicHFTokenizer(torch.nn.Module):
+            def __init__(self, **kwargs):
+                super().__init__()
+                self.tokenizer = load_pretrained_hf_tokenizer(model_name, use_fast=False)
 
-    def _init_vocab(self, vocab_file: str) -> Dict[str, int]:
-        from transformers.models.bert.tokenization_bert import load_vocab
+            def forward(self, v: Union[str, List[str], torch.Tensor]):
+                if isinstance(v, torch.Tensor):
+                    raise ValueError(f"Unsupported input: {v}")
+                return self.tokenizer.tokenize(v)
 
-        return load_vocab(vocab_file)
-
-    def forward(self, v: Union[str, List[str], torch.Tensor]) -> Any:
-        """Implements forward pass for tokenizer.
-
-        If the is_hf_tokenizer flag is set to True, then the output follows the HF convention, i.e. the output is an
-        List[List[int]] of tokens and the cls and sep tokens are automatically added as the start and stop symbols.
-
-        If the is_hf_tokenizer flag is set to False, then the output follows the Ludwig convention, i.e. the output
-        is a List[List[str]] of tokens.
-        """
-        if isinstance(v, torch.Tensor):
-            raise ValueError(f"Unsupported input: {v}")
-
-        inputs: List[str] = []
-        # Ludwig calls map on List[str] objects, so we need to handle individual strings as well.
-        if isinstance(v, str):
-            inputs.append(v)
-        else:
-            inputs.extend(v)
-
-        if self.is_hf_tokenizer:
-            token_ids_str = self.tokenizer(inputs)
-            assert torch.jit.isinstance(token_ids_str, List[List[str]])
-            # Must cast token_ids to ints because they are used directly as indices.
-            token_ids: List[List[int]] = []
-            for token_ids_str_i in token_ids_str:
-                token_ids_i = [int(token_id_str) for token_id_str in token_ids_str_i]
-                token_ids_i = self._add_special_token_ids(token_ids_i)
-                token_ids.append(token_ids_i)
-            return token_ids[0] if isinstance(v, str) else token_ids
-
-        tokens = self.tokenizer(inputs)
-        assert torch.jit.isinstance(tokens, List[List[str]])
-        return tokens[0] if isinstance(v, str) else tokens
-
-    def get_vocab(self) -> Dict[str, int]:
-        return self.vocab
-
-    def get_pad_token(self) -> str:
-        return self.pad_token
-
-    def get_unk_token(self) -> str:
-        return self.unk_token
-
-    def _add_special_token_ids(self, token_ids: List[int]) -> List[int]:
-        """Adds special token ids to the token_ids list."""
-        if torch.jit.isinstance(self.cls_token_id, int) and torch.jit.isinstance(self.sep_token_id, int):
-            token_ids.insert(0, self.cls_token_id)
-            token_ids.append(self.sep_token_id)
-        return token_ids
-
-    def convert_token_to_id(self, token: str) -> int:
-        return self.vocab[token]
+        return DynamicHFTokenizer
 
 
 tokenizer_registry.update(
-    {
-        "bert": BERTTokenizer,
-    }
+    {name: HFTokenizerShortcutFactory.create_class(model) for name, model in HFTokenizerShortcutFactory.MODELS.items()}
 )
-TORCHSCRIPT_COMPATIBLE_TOKENIZERS.update(TORCHTEXT_0_13_0_TOKENIZERS)
 
 
 def get_hf_tokenizer(pretrained_model_name_or_path, **kwargs):
@@ -1242,82 +1057,8 @@ def get_hf_tokenizer(pretrained_model_name_or_path, **kwargs):
     Returns:
         A torchscript-able HF tokenizer if it is available. Else, returns vanilla HF tokenizer.
     """
-    from transformers import BertTokenizer, DistilBertTokenizer, ElectraTokenizer
-
-    # HuggingFace has implemented a DO Repeat Yourself policy for models
-    # https://github.com/huggingface/transformers/issues/19303
-    # We now need to manually track BERT-like tokenizers to map onto the TorchText implementation
-    # until PyTorch improves TorchScript to be able to compile HF tokenizers. This would require
-    #  1. Support for string inputs for torch.jit.trace, or
-    #  2. Support for `kwargs` in torch.jit.script
-    # This is populated in the `get_hf_tokenizer` since the set requires `transformers` to be installed
-    HF_BERTLIKE_TOKENIZER_CLS_SET = {BertTokenizer, DistilBertTokenizer, ElectraTokenizer}
-
-    hf_name = pretrained_model_name_or_path
-    # use_fast=False to leverage python class inheritance
-    # cannot tokenize HF tokenizers directly because HF lacks strict typing and List[str] cannot be traced
-    hf_tokenizer = load_pretrained_hf_tokenizer(hf_name, use_fast=False)
-
-    torchtext_tokenizer = None
-    if "bert" in TORCHSCRIPT_COMPATIBLE_TOKENIZERS and any(
-        isinstance(hf_tokenizer, cls) for cls in HF_BERTLIKE_TOKENIZER_CLS_SET
-    ):
-        tokenizer_kwargs = _get_bert_config(hf_name)
-        torchtext_tokenizer = BERTTokenizer(
-            **tokenizer_kwargs,
-            is_hf_tokenizer=True,
-            hf_tokenizer_attrs={
-                "pad_token": hf_tokenizer.pad_token,
-                "unk_token": hf_tokenizer.unk_token,
-                "cls_token_id": hf_tokenizer.cls_token_id,
-                "sep_token_id": hf_tokenizer.sep_token_id,
-                "all_special_tokens": hf_tokenizer.all_special_tokens,
-            },
-        )
-
-    use_torchtext = torchtext_tokenizer is not None
-    if use_torchtext:
-        # If a torchtext tokenizer is instantiable, tenatively we will use it. However,
-        # if the tokenizer does not pass (lightweight) validation, then we will fall back to the vanilla HF tokenizer.
-        # TODO(geoffrey): can we better validate tokenizer parity before swapping in the TorchText tokenizer?
-        # Samples from https://github.com/huggingface/transformers/blob/main/tests/models/bert/test_tokenization_bert.py
-        for sample_input in HF_TOKENIZER_SAMPLE_INPUTS:
-            hf_output = hf_tokenizer.encode(sample_input)
-            tt_output = torchtext_tokenizer(sample_input)
-            if hf_output != tt_output:
-                use_torchtext = False
-                logger.warning("Falling back to HuggingFace tokenizer because TorchText tokenizer failed validation.")
-                logger.warning(f"Sample input: {sample_input}\nHF output: {hf_output}\nTT output: {tt_output}")
-                break
-
-    if use_torchtext:
-        logger.info(f"Loaded TorchText implementation of {hf_name} tokenizer")
-        return torchtext_tokenizer
-    else:
-        # If hf_name does not have a torchtext equivalent implementation, load the
-        # HuggingFace implementation.
-        logger.info(f"Loaded HuggingFace implementation of {hf_name} tokenizer")
-        return HFTokenizer(hf_name)
-
-
-def _get_bert_config(hf_name):
-    """Gets configs from BERT tokenizers in HuggingFace.
-
-    `vocab_file` is required for BERT tokenizers. `tokenizer_config.json` are optional keyword arguments used to
-    initialize the tokenizer object. If no `tokenizer_config.json` is found, then we instantiate the tokenizer with
-    default arguments.
-    """
-    from huggingface_hub import hf_hub_download
-    from huggingface_hub.utils import EntryNotFoundError
-
-    vocab_file = hf_hub_download(repo_id=hf_name, filename="vocab.txt")
-
-    try:
-        tokenizer_config = load_json(hf_hub_download(repo_id=hf_name, filename="tokenizer_config.json"))
-    except EntryNotFoundError:
-        tokenizer_config = {}
 
-    return {"vocab_file": vocab_file, **tokenizer_config}
+    return HFTokenizer(pretrained_model_name_or_path)
 
 
 tokenizer_registry.update(
@@ -1335,24 +1076,5 @@ def get_tokenizer_from_registry(tokenizer_name: str) -> torch.nn.Module:
     """
     if tokenizer_name in tokenizer_registry:
         return tokenizer_registry[tokenizer_name]
-
-    if (
-        torch.torch_version.TorchVersion(torchtext.__version__) < (0, 12, 0)
-        and tokenizer_name in TORCHTEXT_0_12_0_TOKENIZERS
-    ):
-        raise KeyError(
-            f"torchtext>=0.12.0 is not installed, so '{tokenizer_name}' and the following tokenizers are not "
-            f"available: {TORCHTEXT_0_12_0_TOKENIZERS}"
-        )
-
-    if (
-        torch.torch_version.TorchVersion(torchtext.__version__) < (0, 13, 0)
-        and tokenizer_name in TORCHTEXT_0_13_0_TOKENIZERS
-    ):
-        raise KeyError(
-            f"torchtext>=0.13.0 is not installed, so '{tokenizer_name}' and the following tokenizers are not "
-            f"available: {TORCHTEXT_0_13_0_TOKENIZERS}"
-        )
-
     # Tokenizer does not exist or is unavailable.
     raise KeyError(f"Invalid tokenizer name: '{tokenizer_name}'. Available tokenizers: {tokenizer_registry.keys()}")
diff --git a/ludwig/utils/torch_utils.py b/ludwig/utils/torch_utils.py
index 10be7c762c8..abdcce0d0c5 100644
--- a/ludwig/utils/torch_utils.py
+++ b/ludwig/utils/torch_utils.py
@@ -60,9 +60,7 @@ def place_on_device(x, device):
 def sequence_length_2D(sequence: torch.Tensor) -> torch.Tensor:
     """Returns the number of non-padding elements per sequence in batch.
 
-    :param sequence: (torch.Tensor) A 2D tensor of shape [batch size x max sequence length].
-
-    # Return
+    :param sequence: (torch.Tensor) A 2D tensor of shape [batch size x max sequence length].  # Return
     :returns: (torch.Tensor) The count on non-zero elements per sequence.
     """
     used = (sequence != SpecialSymbol.PADDING.value).type(torch.int32)
@@ -74,9 +72,7 @@ def sequence_length_2D(sequence: torch.Tensor) -> torch.Tensor:
 def sequence_length_3D(sequence: torch.Tensor) -> torch.Tensor:
     """Returns the number of non-zero elements per sequence in batch.
 
-    :param sequence: (torch.Tensor) A 3D tensor of shape [batch size x max sequence length x hidden size].
-
-    # Return
+    :param sequence: (torch.Tensor) A 3D tensor of shape [batch size x max sequence length x hidden size].  # Return
     :returns: (torch.Tensor) The count on non-zero elements per sequence.
     """
     used = torch.sign(torch.amax(torch.abs(sequence), dim=2))
@@ -92,9 +88,7 @@ def sequence_mask(lengths: torch.Tensor, maxlen: Optional[int] = None, dtype: to
 
     :param lengths: (torch.Tensor) A 1d integer tensor of shape [batch size].
     :param maxlen: (Optional[int]) The maximum sequence length.  If not specified, the max(lengths) is used.
-    :param dtype: (type) The type to output.
-
-    # Return
+    :param dtype: (type) The type to output.  # Return
     :returns: (torch.Tensor) A sequence mask tensor of shape (batch_size x maxlen).
     """
     if maxlen is None:
diff --git a/ludwig/utils/triton_utils.py b/ludwig/utils/triton_utils.py
index b0082d317f1..3d81cdb1069 100644
--- a/ludwig/utils/triton_utils.py
+++ b/ludwig/utils/triton_utils.py
@@ -739,8 +739,7 @@ def export_triton(
 
     # Inputs
     :param model: (LudwigModel) A ludwig model.
-    :param data_example: (pd.DataFrame) an example from the dataset.
-        Used to get dimensions throughout the pipeline.
+    :param data_example: (pd.DataFrame) an example from the dataset. Used to get dimensions throughout the pipeline.
     :param output_path: (str) The output path for the model repository.
     :param model_name: (str) The optional model name.
     :param model_version: (Union[int,str]) The optional model verison.
@@ -749,9 +748,7 @@ def export_triton(
     :param predictor_num_instances: (int) number of instances for the predictor.
     :param postprocessor_num_instances: (int) number of instances for the postprocessor (on CPU).
     :param predictor_max_batch_size: (int) max_batch_size parameter for the predictor Triton config.
-    :param max_queue_delay_microseconds: (int) max_queue_delay_microseconds for all Triton configs.
-
-    # Return
+    :param max_queue_delay_microseconds: (int) max_queue_delay_microseconds for all Triton configs.  # Return
     :return: (List[TritonArtifact]) list of TritonArtifacts that contains information about exported artifacts.
     """
 
diff --git a/ludwig/utils/upload_utils.py b/ludwig/utils/upload_utils.py
index f3aed5f8bea..51a0fb87efd 100644
--- a/ludwig/utils/upload_utils.py
+++ b/ludwig/utils/upload_utils.py
@@ -188,8 +188,8 @@ def _validate_upload_parameters(
         )
 
         trained_model_artifacts_path = os.path.join(model_path, MODEL_FILE_NAME, MODEL_WEIGHTS_FILE_NAME)
-        """
-        Make sure the model's saved artifacts either contain:
+        """Make sure the model's saved artifacts either contain:
+
         1. pytorch_model.bin -> regular model training, such as ECD or for LLMs
         2. adapter_model.bin or adapter_model.safetensors -> LLM fine-tuning using PEFT
            <Alex(12/10/2023): TODO>
diff --git a/ludwig/visualize.py b/ludwig/visualize.py
index 61c41e0fbb4..feb97897072 100644
--- a/ludwig/visualize.py
+++ b/ludwig/visualize.py
@@ -60,7 +60,7 @@
 
 
 def _convert_ground_truth(ground_truth, feature_metadata, ground_truth_apply_idx, positive_label):
-    """converts non-np.array representation to be np.array."""
+    """Converts non-np.array representation to be np.array."""
     if "str2idx" in feature_metadata:
         # categorical output feature as binary
         ground_truth = _vectorize_ground_truth(ground_truth, feature_metadata["str2idx"], ground_truth_apply_idx)
@@ -102,8 +102,7 @@ def validate_conf_thresholds_and_probabilities_2d_3d(probabilities, threshold_ou
     """Ensure probabilities and threshold output_feature_names arrays have two members each.
 
     :param probabilities: List of probabilities per model
-    :param threshhold_output_feature_names: List of threshhold output_feature_names per model
-    :raise: RuntimeError
+    :param threshhold_output_feature_names: List of threshhold output_feature_names per model :raise: RuntimeError
     """
     validation_mapping = {
         "probabilities": probabilities,
@@ -122,9 +121,8 @@ def load_data_for_viz(load_type, model_file_statistics, dtype=int, ground_truth_
     """Load JSON files (training stats, evaluation stats...) for a list of models.
 
     :param load_type: type of the data loader to be used.
-    :param model_file_statistics: JSON file or list of json files containing any
-           model experiment stats.
-    :return List of training statistics loaded as json objects.
+    :param model_file_statistics: JSON file or list of json files containing any model experiment stats. :return List of
+        training statistics loaded as json objects.
     """
     supported_load_types = dict(
         load_json=load_json,
@@ -145,9 +143,8 @@ def load_training_stats_for_viz(load_type, model_file_statistics, dtype=int, gro
     """Load model file data (specifically training stats) for a list of models.
 
     :param load_type: type of the data loader to be used.
-    :param model_file_statistics: JSON file or list of json files containing any
-           model experiment stats.
-    :return List of model statistics loaded as TrainingStats objects.
+    :param model_file_statistics: JSON file or list of json files containing any model experiment stats. :return List of
+        model statistics loaded as TrainingStats objects.
     """
     stats_per_model = load_data_for_viz(
         load_type, model_file_statistics, dtype=dtype, ground_truth_split=ground_truth_split
@@ -213,7 +210,7 @@ def _validate_output_feature_name_from_test_stats(output_feature_name, test_stat
 
 
 def _encode_categorical_feature(raw: np.array, str2idx: dict) -> np.array:
-    """encodes raw categorical string value to encoded numeric value.
+    """Encodes raw categorical string value to encoded numeric value.
 
     Args:
     :param raw: (np.array) string categorical representation
@@ -326,10 +323,8 @@ def generate_filename_template_path(output_dir, filename_template):
 
     Create output directory if yet does exist.
     :param output_dir: Directory that will contain the filename_template file
-    :param filename_template: name of the file template to be appended to the
-            filename template path
-    :return: path to filename template inside the output dir or None if the
-             output dir is None
+    :param filename_template: name of the file template to be appended to the filename template path
+    :return: path to filename template inside the output dir or None if the output dir is None
     """
     if output_dir:
         os.makedirs(output_dir, exist_ok=True)
@@ -343,12 +338,8 @@ def compare_performance_cli(test_statistics: Union[str, List[str]], **kwargs: di
 
     # Inputs
 
-    :param test_statistics: (Union[str, List[str]]) path to experiment test
-        statistics file.
-    :param kwargs: (dict) parameters for the requested visualizations.
-
-    # Return
-
+    :param test_statistics: (Union[str, List[str]]) path to experiment test statistics file.
+    :param kwargs: (dict) parameters for the requested visualizations.  # Return
     :return None:
     """
     test_stats_per_model = load_data_for_viz("load_json", test_statistics)
@@ -361,12 +352,8 @@ def learning_curves_cli(training_statistics: Union[str, List[str]], **kwargs: di
 
     # Inputs
 
-    :param training_statistics: (Union[str, List[str]]) path to experiment
-        training statistics file
-    :param kwargs: (dict) parameters for the requested visualizations.
-
-    # Return
-
+    :param training_statistics: (Union[str, List[str]]) path to experiment training statistics file
+    :param kwargs: (dict) parameters for the requested visualizations.  # Return
     :return None:
     """
     train_stats_per_model = load_training_stats_for_viz("load_json", training_statistics)
@@ -588,13 +575,9 @@ def compare_classifiers_multiclass_multimetric_cli(
 
     # Inputs
 
-    :param test_statistics: (Union[str, List[str]]) path to experiment test
-        statistics file.
+    :param test_statistics: (Union[str, List[str]]) path to experiment test statistics file.
     :param ground_truth_metadata: (str) path to ground truth metadata file.
-    :param kwargs: (dict) parameters for the requested visualizations.
-
-    # Return
-
+    :param kwargs: (dict) parameters for the requested visualizations.  # Return
     :return None:
     """
     test_stats_per_model = load_data_for_viz("load_json", test_statistics)
@@ -1175,12 +1158,8 @@ def roc_curves_from_test_statistics_cli(test_statistics: Union[str, List[str]],
     """Load model data from files to be shown by roc_curves_from_test_statistics_cli.
 
     # Inputs
-    :param test_statistics: (Union[str, List[str]]) path to experiment test
-        statistics file.
-    :param kwargs: (dict) parameters for the requested visualizations.
-
-    # Return
-
+    :param test_statistics: (Union[str, List[str]]) path to experiment test statistics file.
+    :param kwargs: (dict) parameters for the requested visualizations.  # Return
     :return None:
     """
     test_stats_per_model = load_data_for_viz("load_json", test_statistics)
@@ -1325,13 +1304,9 @@ def confusion_matrix_cli(test_statistics: Union[str, List[str]], ground_truth_me
 
     # Inputs
 
-    :param test_statistics: (Union[str, List[str]]) path to experiment test
-        statistics file.
+    :param test_statistics: (Union[str, List[str]]) path to experiment test statistics file.
     :param ground_truth_metadata: (str) path to ground truth metadata file.
-    :param kwargs: (dict) parameters for the requested visualizations.
-
-    # Return
-
+    :param kwargs: (dict) parameters for the requested visualizations.  # Return
     :return None:
     """
     test_stats_per_model = load_data_for_viz("load_json", test_statistics)
@@ -1345,13 +1320,9 @@ def frequency_vs_f1_cli(test_statistics: Union[str, List[str]], ground_truth_met
 
     # Inputs
 
-    :param test_statistics: (Union[str, List[str]]) path to experiment test
-        statistics file.
+    :param test_statistics: (Union[str, List[str]]) path to experiment test statistics file.
     :param ground_truth_metadata: (str) path to ground truth metadata file.
-    :param kwargs: (dict) parameters for the requested visualizations.
-
-    # Return
-
+    :param kwargs: (dict) parameters for the requested visualizations.  # Return
     :return None:
     """
     test_stats_per_model = load_data_for_viz("load_json", test_statistics)
diff --git a/pyproject.toml b/pyproject.toml
index 38ac51086cf..bdc2934a345 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -1,3 +1,218 @@
+[build-system]
+requires = ["hatchling", "setuptools>=65.0"]
+build-backend = "hatchling.build"
+
+[project]
+name = "ludwig"
+dynamic = ["version"]
+description = "Declarative machine learning: End-to-end machine learning pipelines using data-driven configurations."
+readme = "README.md"
+license = { text = "Apache 2.0" }
+requires-python = ">=3.10"
+authors = [{ name = "Piero Molino", email = "piero.molino@gmail.com" }]
+keywords = [
+    "computer",
+    "deep",
+    "deep_learning",
+    "language",
+    "learning",
+    "ludwig",
+    "machine",
+    "machine_learning",
+    "natural",
+    "processing",
+    "vision",
+]
+
+dependencies = [
+    "absl-py",
+    "bitsandbytes<0.41.0",
+    "Cython>=0.25,<1.0",
+    "dataclasses-json",
+    "datasets",
+    "filelock",
+    "fsspec[http]<=2023.10.0",
+    "getdaft==0.1.20",
+    "gpustat",
+    "h5py>=2.6,!=3.0.0",
+    "html5lib",
+    "imagecodecs",
+    "jsonschema>=4.5.0,<4.7",
+    "kaggle",
+    "lxml",
+    "marshmallow",
+    "marshmallow-dataclass==8.5.4",
+    "marshmallow-jsonschema",
+    "nltk",
+    "numpy==1.26",
+    "openpyxl>=3.0.7",
+    "packaging",
+    "pandas",
+    "protobuf",
+    "psutil",
+    "py-cpuinfo==9.0.0",
+    "pyarrow<15.0.0",
+    "pydantic<2.0",
+    "pythran>=0.9",
+    "pyxlsb>=1.0.8",
+    "PyYAML==6.0.2",
+    "requests",
+    "retry",
+    "rich~=12.4.4",
+    "sacremoses",
+    "scikit-learn==1.3",
+    "matplotlib==3.9.3,!=3.4.3",
+    #"scipy>=0.18",
+    "scipy==1.14.1",
+    "sentencepiece",
+    "spacy",
+    "tabulate>=0.7",
+    "tensorboard",
+    "tokenizers>=0.15",
+    "torch==2.4.1",
+    "torchaudio==2.4.1",
+    "torchinfo",
+    "torchmetrics>=0.11.0",
+    "torchvision==0.19.1",
+    "tqdm",
+    "transformers>=4.42.3",
+    "urllib3<2",
+    "xlrd>=2.0.1",
+    "xlsxwriter>=1.4.3",
+    "xlwt",
+    "tifffile==2024.9.20",
+    "onnx",
+]
+
+[project.optional-dependencies]
+dev = ["flake8", "flake8-pyproject", "pre-commit", "setuptools"]
+test = [
+    # Core testing
+    "pytest",
+    "pytest-timeout",
+    "pytest-cov",
+    "tifffile",
+    "wget",
+    "six>=1.13.0",
+
+    # Logging and experiment tracking
+    "aim",
+    #"wandb<0.12.11",
+    "wandb",
+    "comet_ml",
+    "mlflow",
+    "sqlalchemy<2",  # Pinned for aimstack compatibility
+
+    # Ray Tune Search Algorithms
+    "hpbandster",            # BOHB algorithm
+    "ConfigSpace==0.7.1",
+    "ax-platform",           # AX algorithm
+    "bayesian-optimization", # Bayesian optimization
+    "flaml[blendsearch]",    # CFO and blendsearch
+    "HEBO",                  # HEBO algorithm
+    "nevergrad",             # Nevergrad algorithm
+    "optuna",                # Optuna algorithm
+    "scikit-optimize",       # SKopt algorithm
+    "zoopt",                 # ZOOpt algorithm
+
+    # Storage
+    "s3fs>=2022.8.2",
+]
+benchmarking = ["s3fs"]
+distributed = [
+    "awscli",
+    "dask[dataframe]<2023.4.0",
+    "deepspeed!=0.11.0,<0.13.0",
+    "getdaft[ray]==0.1.20",
+    "GPUtil",
+    "pyarrow",
+    "ray[default,data,serve,tune]==2.3.1",
+    "tblib",
+    "tensorboardX<2.3",
+]
+explain = ["captum"]
+extra = [
+    "horovod[pytorch]>=0.24.0,!=0.26.0",
+    "modin[ray]",
+    "predibase>=2023.10.2",
+]
+hyperopt = ["hyperopt", "ray[default,tune]>=2.0.0"]
+llm = [
+    "accelerate",
+    "faiss-cpu",
+    "loralib",
+    "peft>=0.10.0",
+    "sentence-transformers",
+]
+serve = [
+    "cartonml-nightly",
+    "fastapi",
+    "httpx",
+    "neuropod==0.3.0rc6 ; platform_system != \"Windows\" and python_version < '3.9'",
+    "python-multipart",
+    "uvicorn",
+    "starlette",
+]
+tree = ["hummingbird-ml>=0.4.8", "lightgbm", "lightgbm-ray"]
+viz = [
+    "hiplot",
+    "matplotlib==3.9.3",
+    "ptitprince",
+    "seaborn>=0.7,<0.12",
+]
+
+[project.urls]
+Download = "https://pypi.org/project/ludwig/"
+Homepage = "https://github.com/ludwig-ai/ludwig"
+Website = "https://ludwig.ai/latest/"
+
+[project.scripts]
+ludwig = "ludwig.cli:main"
+
+[tool.hatch.version]
+path = "ludwig/__about__.py"
+
+[tool.hatch.build.targets.sdist]
+include = ["/ludwig", "/tests"]
+
+# ------- flake8 ----------
+[tool.flake8]
+max-line-length = 120
+exclude = [".tox", "*.egg", "*_pb2.py", "build", "temp"]
+select = ["E", "W", "F"]
+doctests = true
+verbose = 2
+format = "pylint"
+ignore = [
+    "E731",
+    "W503",
+    "E203",
+    "E231",
+    "E241",
+    "E221",
+    "E225",
+    "E226",
+    "E241",
+    "E271",
+    "E275",
+    "E501",
+]
+
+[tool.hatch.envs.lint]
+dependencies = ["flake8", "flake8-pyproject"]
+
+[tool.hatch.envs.lint.scripts]
+style = "flake8 ."
+
+[tool.hatch.envs.default]
+python = "3.11"
+dependencies = ["setuptools>=65.0"]
+
+[tool.hatch.envs.dev]
+python = "3.11"
+dependencies = [".[dev]"]
+
+
 [tool.isort]
 profile = "black"
 line_length = 120
diff --git a/requirements.txt b/requirements.txt
deleted file mode 100644
index d293073e2c7..00000000000
--- a/requirements.txt
+++ /dev/null
@@ -1,69 +0,0 @@
-Cython>=0.25
-h5py>=2.6,!=3.0.0
-numpy>=1.15
-# GitHub Issue for Pinning Pandas < 2.2.0: https://github.com/ludwig-ai/ludwig/issues/3907
-pandas>=1.0,!=1.1.5,<2.2.0
-scipy>=0.18
-tabulate>=0.7
-scikit-learn
-tqdm
-torch>=2.0.0
-torchaudio
-torchtext
-torchvision
-pydantic<2.0
-transformers>=4.42.3
-tifffile
-imagecodecs
-tokenizers>=0.15
-spacy>=2.3
-PyYAML>=3.12,<6.0.1,!=5.4.* #Exlude PyYAML 5.4.* due to incompatibility with awscli
-absl-py
-kaggle
-requests
-fsspec[http]<=2023.10.0
-dataclasses-json
-jsonschema>=4.5.0,<4.7
-marshmallow
-marshmallow-jsonschema
-marshmallow-dataclass==8.5.4
-tensorboard
-nltk    # Required for rouge scores.
-torchmetrics>=0.11.0
-torchinfo
-filelock
-psutil
-protobuf
-py-cpuinfo==9.0.0
-gpustat
-rich~=12.4.4
-packaging
-retry
-
-# required for TransfoXLTokenizer when using transformer_xl
-sacremoses
-sentencepiece
-
-# requirements for daft
-# NOTE: daft needs to be <0.2 because of deprecation of fsspec argument in Daft
-# Pinned for consistency with ludwig-ray docker image.
-getdaft==0.1.20
-
-# requirement for various paged and 8-bit optimizers
-bitsandbytes<0.41.0
-
-# new data format support
-xlwt                # excel
-xlrd>=2.0.1         # excel
-xlsxwriter>=1.4.3   # excel
-openpyxl>=3.0.7     # excel
-pyxlsb>=1.0.8       # excel
-pyarrow<15.0.0      # parquet
-lxml                # html
-html5lib            # html
-
-# requirement for loading hugging face datasets
-datasets
-
-# pin required for torch 2.1.0
-urllib3<2
diff --git a/requirements_benchmarking.txt b/requirements_benchmarking.txt
deleted file mode 100644
index a51391a9488..00000000000
--- a/requirements_benchmarking.txt
+++ /dev/null
@@ -1 +0,0 @@
-s3fs
diff --git a/requirements_distributed.txt b/requirements_distributed.txt
deleted file mode 100644
index e39ee755950..00000000000
--- a/requirements_distributed.txt
+++ /dev/null
@@ -1,17 +0,0 @@
-# requirements for dask
-dask[dataframe]<2023.4.0
-pyarrow
-
-# requirements for ray
-ray[default,data,serve,tune]==2.3.1
-tensorboardX<2.3
-GPUtil
-tblib
-awscli
-
-# https://github.com/microsoft/DeepSpeed/issues/4473
-# https://github.com/ludwig-ai/ludwig/issues/3905
-deepspeed!=0.11.0,<0.13.0
-
-# requirements for daft
-getdaft[ray]==0.1.20
diff --git a/requirements_explain.txt b/requirements_explain.txt
deleted file mode 100644
index 7a4edb90b8e..00000000000
--- a/requirements_explain.txt
+++ /dev/null
@@ -1 +0,0 @@
-captum
diff --git a/requirements_extra.txt b/requirements_extra.txt
deleted file mode 100644
index 26fe48eb998..00000000000
--- a/requirements_extra.txt
+++ /dev/null
@@ -1,8 +0,0 @@
-# requirements for horovod
-horovod[pytorch]>=0.24.0,!=0.26.0
-
-# alternative to Dask
-modin[ray]
-
-# Allows users to upload
-predibase>=2023.10.2
diff --git a/requirements_hyperopt.txt b/requirements_hyperopt.txt
deleted file mode 100644
index 3b85fea598c..00000000000
--- a/requirements_hyperopt.txt
+++ /dev/null
@@ -1,5 +0,0 @@
-ray[default,tune]>=2.0.0
-
-# required for Ray Tune Search Algorithm support for AutoML
-#search_alg: hyperopt
-hyperopt
diff --git a/requirements_llm.txt b/requirements_llm.txt
deleted file mode 100644
index c691bc0bac3..00000000000
--- a/requirements_llm.txt
+++ /dev/null
@@ -1,7 +0,0 @@
-sentence-transformers
-faiss-cpu
-
-accelerate
-loralib
-
-peft>=0.10.0
diff --git a/requirements_serve.txt b/requirements_serve.txt
deleted file mode 100644
index 353adbb3f5c..00000000000
--- a/requirements_serve.txt
+++ /dev/null
@@ -1,6 +0,0 @@
-uvicorn
-httpx
-fastapi
-python-multipart
-neuropod==0.3.0rc6 ; platform_system != "Windows" and python_version < '3.9'
-cartonml-nightly
diff --git a/requirements_test.txt b/requirements_test.txt
deleted file mode 100644
index f42f76db6a7..00000000000
--- a/requirements_test.txt
+++ /dev/null
@@ -1,47 +0,0 @@
-pytest
-pytest-timeout
-tifffile
-wget
-six>=1.13.0
-aim
-wandb<0.12.11
-comet_ml
-mlflow
-
-# For testing optional Ray Tune Search Algorithms
-# search_alg: bohb
-hpbandster
-ConfigSpace==0.7.1
-
-# search_alg: ax
-ax-platform
-
-# Pinning because aimstack does not support 2.x.x - https://github.com/aimhubio/aim/issues/2514
-sqlalchemy<2
-
-# search_alg: bayesopt
-bayesian-optimization
-
-# search_alg: cfo and blendsearch
-flaml[blendsearch]
-
-# Disabling due to numpy installation failure https://github.com/ludwig-ai/ludwig/actions/runs/4737879639/jobs/8411146481
-# search_alg: dragonfly
-# dragonfly-opt
-
-# search_alg: hebo
-HEBO
-
-# search_alg: nevergrad
-nevergrad
-
-# search_alg: optuna
-optuna
-
-# search_alg: skopt
-scikit-optimize
-
-# search_alg: zoopt
-zoopt
-
-s3fs>=2022.8.2
diff --git a/requirements_tree.txt b/requirements_tree.txt
deleted file mode 100644
index f2153b1f3e0..00000000000
--- a/requirements_tree.txt
+++ /dev/null
@@ -1,3 +0,0 @@
-hummingbird-ml>=0.4.8
-lightgbm
-lightgbm-ray
diff --git a/requirements_viz.txt b/requirements_viz.txt
deleted file mode 100644
index a33a1d546f3..00000000000
--- a/requirements_viz.txt
+++ /dev/null
@@ -1,5 +0,0 @@
-matplotlib>3.4,<3.9.0; python_version > '3.6'
-matplotlib>=3.0,<3.4; python_version <= '3.6'
-seaborn>=0.7,<0.12
-hiplot
-ptitprince
diff --git a/setup.cfg b/setup.cfg
deleted file mode 100644
index c9095299645..00000000000
--- a/setup.cfg
+++ /dev/null
@@ -1,37 +0,0 @@
-[flake8]
-max-line-length = 120
-exclude =
-    .tox,
-    *.egg,
-    *_pb2.py,
-    build,
-    temp
-
-select = E,W,F
-doctests = True
-verbose = 2
-# https://pep8.readthedocs.io/en/latest/intro.html#error-codes
-format = pylint
-ignore =
-    # Ignore "Do not assign a lambda expression, use a def"
-    E731
-    # Ignore "Line break occurred before a binary operator"
-    W503
-    # Ignore "whitespace before ':'"
-    E203
-    # Ignore "missing whitespace after ':'"
-    E231
-    # Ignore "multiple spaces after ':'"
-    E241
-    # Ignore "multiple spaces before operator"
-    E221
-    # Ignore "whitespace around operator"
-    E225
-    # Ignore "whitespace around arithmetic operator"
-    E226
-    # Ignore "multiple spaces after ':'"
-    E241
-    # Ignore "multiple spaces after keyword"
-    E271
-    # Ignore "missing whitespace after keyword"
-    E275
diff --git a/setup.py b/setup.py
deleted file mode 100644
index 7276b5fabf7..00000000000
--- a/setup.py
+++ /dev/null
@@ -1,69 +0,0 @@
-"""Ludwig: Data-centric declarative deep learning framework."""
-from codecs import open
-from os import path
-
-from setuptools import find_packages, setup
-
-here = path.abspath(path.dirname(__file__))
-
-# Get the long description from the README.md file
-with open(path.join(here, "README.md"), encoding="utf-8") as f:
-    long_description = f.read()
-
-with open(path.join(here, "requirements.txt"), encoding="utf-8") as f:
-    requirements = [line.strip() for line in f if line]
-
-extra_requirements = {}
-
-with open(path.join(here, "requirements_serve.txt"), encoding="utf-8") as f:
-    extra_requirements["serve"] = [line.strip() for line in f if line]
-
-with open(path.join(here, "requirements_viz.txt"), encoding="utf-8") as f:
-    extra_requirements["viz"] = [line.strip() for line in f if line]
-
-with open(path.join(here, "requirements_distributed.txt"), encoding="utf-8") as f:
-    extra_requirements["distributed"] = [line.strip() for line in f if line]
-
-with open(path.join(here, "requirements_hyperopt.txt"), encoding="utf-8") as f:
-    extra_requirements["hyperopt"] = [line.strip() for line in f if line]
-
-with open(path.join(here, "requirements_tree.txt"), encoding="utf-8") as f:
-    extra_requirements["tree"] = [line.strip() for line in f if line]
-
-with open(path.join(here, "requirements_llm.txt"), encoding="utf-8") as f:
-    extra_requirements["llm"] = [line.strip() for line in f if line]
-
-with open(path.join(here, "requirements_explain.txt"), encoding="utf-8") as f:
-    extra_requirements["explain"] = [line.strip() for line in f if line]
-
-with open(path.join(here, "requirements_benchmarking.txt"), encoding="utf-8") as f:
-    extra_requirements["benchmarking"] = [line.strip() for line in f if line]
-
-extra_requirements["full"] = [item for sublist in extra_requirements.values() for item in sublist]
-
-with open(path.join(here, "requirements_test.txt"), encoding="utf-8") as f:
-    extra_requirements["test"] = extra_requirements["full"] + [line.strip() for line in f if line]
-
-with open(path.join(here, "requirements_extra.txt"), encoding="utf-8") as f:
-    extra_requirements["extra"] = [line.strip() for line in f if line]
-
-setup(
-    name="ludwig",
-    version="0.10.4.dev",
-    description="Declarative machine learning: End-to-end machine learning pipelines using data-driven configurations.",
-    long_description=long_description,
-    long_description_content_type="text/markdown",
-    url="https://github.com/ludwig-ai/ludwig",
-    download_url="https://pypi.org/project/ludwig/",
-    author="Piero Molino",
-    author_email="piero.molino@gmail.com",
-    license="Apache 2.0",
-    keywords="ludwig deep learning deep_learning machine machine_learning natural language processing computer vision",
-    packages=find_packages(exclude=["contrib", "docs", "tests"]),
-    python_requires=">=3.8",
-    include_package_data=True,
-    package_data={"ludwig": ["etc/*", "examples/*.py"]},
-    install_requires=requirements,
-    extras_require=extra_requirements,
-    entry_points={"console_scripts": ["ludwig=ludwig.cli:main"]},
-)
diff --git a/tests/integration_tests/scripts/run_train_comet.py b/tests/integration_tests/scripts/run_train_comet.py
index 52842b40200..2f4b2b73463 100644
--- a/tests/integration_tests/scripts/run_train_comet.py
+++ b/tests/integration_tests/scripts/run_train_comet.py
@@ -28,7 +28,8 @@
 PATH_ROOT = os.path.join(PATH_HERE, "..", "..", "..")
 sys.path.insert(0, os.path.abspath(PATH_ROOT))
 
-from tests.integration_tests.utils import category_feature, generate_data, image_feature  # noqa
+from tests.integration_tests.utils import category_feature  # noqa
+from tests.integration_tests.utils import generate_data, image_feature
 
 parser = argparse.ArgumentParser()
 parser.add_argument("--csv-filename", required=True)
diff --git a/tests/integration_tests/scripts/run_train_horovod.py b/tests/integration_tests/scripts/run_train_horovod.py
index a40beabe5ef..8cf7182f52d 100644
--- a/tests/integration_tests/scripts/run_train_horovod.py
+++ b/tests/integration_tests/scripts/run_train_horovod.py
@@ -82,5 +82,5 @@ def test_horovod_intent_classification(rel_path, input_features, output_features
         args.rel_path,
         json.loads(args.input_features),
         json.loads(args.output_features),
-        **json.loads(args.ludwig_kwargs)
+        **json.loads(args.ludwig_kwargs),
     )
diff --git a/tests/integration_tests/scripts/run_train_wandb.py b/tests/integration_tests/scripts/run_train_wandb.py
index 5496a4ebdb4..116624114db 100644
--- a/tests/integration_tests/scripts/run_train_wandb.py
+++ b/tests/integration_tests/scripts/run_train_wandb.py
@@ -18,7 +18,8 @@
 PATH_ROOT = os.path.join(PATH_HERE, "..", "..", "..")
 sys.path.insert(0, os.path.abspath(PATH_ROOT))
 
-from tests.integration_tests.utils import category_feature, generate_data, image_feature, run_experiment  # noqa
+from tests.integration_tests.utils import category_feature  # noqa
+from tests.integration_tests.utils import generate_data, image_feature, run_experiment
 
 parser = argparse.ArgumentParser()
 parser.add_argument("--csv-filename", required=True)
diff --git a/tests/integration_tests/test_explain.py b/tests/integration_tests/test_explain.py
index d0b183734c0..9e1541bb476 100644
--- a/tests/integration_tests/test_explain.py
+++ b/tests/integration_tests/test_explain.py
@@ -166,7 +166,7 @@ def run_test_explainer_api(
     tmpdir,
     input_features=None,
     batch_size=128,
-    **kwargs
+    **kwargs,
 ):
     image_dest_folder = os.path.join(tmpdir, "generated_images")
 
diff --git a/tests/integration_tests/test_hyperopt.py b/tests/integration_tests/test_hyperopt.py
index 42f4e73c645..9db188ca279 100644
--- a/tests/integration_tests/test_hyperopt.py
+++ b/tests/integration_tests/test_hyperopt.py
@@ -62,7 +62,8 @@
 
 ray = pytest.importorskip("ray")
 
-from ludwig.hyperopt.execution import get_build_hyperopt_executor, RayTuneExecutor  # noqa
+from ludwig.hyperopt.execution import RayTuneExecutor  # noqa
+from ludwig.hyperopt.execution import get_build_hyperopt_executor
 
 pytestmark = [pytest.mark.distributed, pytest.mark.integration_tests_a]
 
diff --git a/tests/integration_tests/test_torchscript.py b/tests/integration_tests/test_torchscript.py
index 198089fed88..bb06371f733 100644
--- a/tests/integration_tests/test_torchscript.py
+++ b/tests/integration_tests/test_torchscript.py
@@ -1,3 +1,4 @@
+# flake8: noqa: E501
 # Copyright (c) 2023 Predibase, Inc., 2019 Uber Technologies, Inc.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
@@ -21,7 +22,6 @@
 import pandas as pd
 import pytest
 import torch
-import torchtext
 
 from ludwig.api import LudwigModel
 from ludwig.backend import RAY
@@ -408,32 +408,7 @@ def test_torchscript_e2e_text(tmpdir, csv_filename):
     validate_torchscript_outputs(tmpdir, config, backend, training_data_csv_path)
 
 
-@pytest.mark.skipif(
-    torch.torch_version.TorchVersion(torchtext.__version__) < (0, 14, 0),
-    reason="requires torchtext 0.14.0 or higher",
-)
-@pytest.mark.integration_tests_e
-def test_torchscript_e2e_text_hf_tokenizer(tmpdir, csv_filename):
-    data_csv_path = os.path.join(tmpdir, csv_filename)
-    input_features = [text_feature(encoder={"vocab_size": 3, "type": "bert"})]
-    output_features = [
-        category_feature(),
-    ]
-    backend = LocalTestBackend()
-    config = {
-        "input_features": input_features,
-        "output_features": output_features,
-        TRAINER: {"epochs": 2, BATCH_SIZE: 128, EVAL_BATCH_SIZE: 128},
-    }
-    training_data_csv_path = generate_data(input_features, output_features, data_csv_path)
-
-    validate_torchscript_outputs(tmpdir, config, backend, training_data_csv_path)
-
-
-@pytest.mark.skipif(
-    torch.torch_version.TorchVersion(torchtext.__version__) < (0, 14, 0),
-    reason="requires torchtext 0.14.0 or higher",
-)
+@pytest.mark.skip()
 @pytest.mark.integration_tests_e
 def test_torchscript_e2e_text_hf_tokenizer_truncated_sequence(tmpdir, csv_filename):
     data_csv_path = os.path.join(tmpdir, csv_filename)
diff --git a/tests/integration_tests/test_visualization.py b/tests/integration_tests/test_visualization.py
index 35893f625c7..0060e322b90 100644
--- a/tests/integration_tests/test_visualization.py
+++ b/tests/integration_tests/test_visualization.py
@@ -80,8 +80,7 @@ def get_output_feature_name(experiment_dir, output_feature=0):
 
     :param experiment_dir: Path to the experiment directory
     :param output_feature: position of the output feature the description.json
-    :return output_feature_name: name of the first output feature name
-                        from the experiment
+    :return output_feature_name: name of the first output feature name from the experiment
     """
     description_file = os.path.join(experiment_dir, DESCRIPTION_FILE_NAME)
     with open(description_file, "rb") as f:
@@ -179,8 +178,7 @@ def test_visualization_confusion_matrix_output_saved(csv_filename):
 def test_visualization_compare_performance_output_saved(csv_filename):
     """Ensure pdf and png figures from the experiments can be saved.
 
-    Compare performance between two models. To reduce test complexity
-    one model is compared to it self.
+    Compare performance between two models. To reduce test complexity one model is compared to it self.
 
     :param csv_filename: csv fixture from tests.conftest.csv_filename
     :return: None
diff --git a/tests/integration_tests/test_visualization_api.py b/tests/integration_tests/test_visualization_api.py
index fac28182be1..0f10bc5c922 100644
--- a/tests/integration_tests/test_visualization_api.py
+++ b/tests/integration_tests/test_visualization_api.py
@@ -118,9 +118,7 @@ def _create_model(self):
 def obtain_df_splits(data_csv):
     """Split input data csv file in to train, validation and test dataframes.
 
-    :param data_csv: Input data CSV file.
-    :return test_df, train_df, val_df: Train, validation and test dataframe
-            splits
+    :param data_csv: Input data CSV file. :return test_df, train_df, val_df: Train, validation and test dataframe splits
     """
     data_df = read_csv(data_csv)
     # Obtain data split array mapping data rows to split type
@@ -134,8 +132,7 @@ def obtain_df_splits(data_csv):
 def test_learning_curves_vis_api(experiment_to_use, training_only):
     """Ensure pdf and png figures can be saved via visualization API call.
 
-    :param experiment_to_use: Object containing trained model and results to
-        test visualization
+    :param experiment_to_use: Object containing trained model and results to test visualization
     :return: None
     """
     experiment = experiment_to_use
@@ -158,8 +155,7 @@ def test_learning_curves_vis_api(experiment_to_use, training_only):
 def test_compare_performance_vis_api(experiment_to_use):
     """Ensure pdf and png figures can be saved via visualization API call.
 
-    :param experiment_to_use: Object containing trained model and results to
-        test visualization
+    :param experiment_to_use: Object containing trained model and results to test visualization
     :return: None
     """
     experiment = experiment_to_use
@@ -183,8 +179,7 @@ def test_compare_performance_vis_api(experiment_to_use):
 def test_compare_classifier_performance_from_prob_vis_api(experiment_to_use):
     """Ensure pdf and png figures can be saved via visualization API call.
 
-    :param experiment_to_use: Object containing trained model and results to
-        test visualization
+    :param experiment_to_use: Object containing trained model and results to test visualization
     :return: None
     """
     experiment = experiment_to_use
@@ -211,8 +206,7 @@ def test_compare_classifier_performance_from_prob_vis_api(experiment_to_use):
 def test_compare_classifier_performance_from_pred_vis_api(experiment_to_use):
     """Ensure pdf and png figures can be saved via visualization API call.
 
-    :param experiment_to_use: Object containing trained model and results to
-        test visualization
+    :param experiment_to_use: Object containing trained model and results to test visualization
     :return: None
     """
     experiment = experiment_to_use
@@ -238,8 +232,7 @@ def test_compare_classifier_performance_from_pred_vis_api(experiment_to_use):
 def test_compare_classifiers_performance_subset_vis_api(experiment_to_use):
     """Ensure pdf and png figures can be saved via visualization API call.
 
-    :param experiment_to_use: Object containing trained model and results to
-        test visualization
+    :param experiment_to_use: Object containing trained model and results to test visualization
     :return: None
     """
     experiment = experiment_to_use
@@ -267,8 +260,7 @@ def test_compare_classifiers_performance_subset_vis_api(experiment_to_use):
 def test_compare_classifiers_performance_changing_k_vis_api(experiment_to_use):
     """Ensure pdf and png figures can be saved via visualization API call.
 
-    :param experiment_to_use: Object containing trained model and results to
-        test visualization
+    :param experiment_to_use: Object containing trained model and results to test visualization
     :return: None
     """
     experiment = experiment_to_use
@@ -295,8 +287,7 @@ def test_compare_classifiers_performance_changing_k_vis_api(experiment_to_use):
 def test_compare_classifiers_multiclass_multimetric_vis_api(experiment_to_use):
     """Ensure pdf and png figures can be saved via visualization API call.
 
-    :param experiment_to_use: Object containing trained model and results to
-        test visualization
+    :param experiment_to_use: Object containing trained model and results to test visualization
     :return: None
     """
     experiment = experiment_to_use
@@ -322,8 +313,7 @@ def test_compare_classifiers_multiclass_multimetric_vis_api(experiment_to_use):
 def test_compare_classifiers_predictions_vis_api(experiment_to_use):
     """Ensure pdf and png figures can be saved via visualization API call.
 
-    :param experiment_to_use: Object containing trained model and results to
-        test visualization
+    :param experiment_to_use: Object containing trained model and results to test visualization
     :return: None
     """
     experiment = experiment_to_use
@@ -349,8 +339,7 @@ def test_compare_classifiers_predictions_vis_api(experiment_to_use):
 def test_compare_classifiers_predictions_distribution_vis_api(experiment_to_use):
     """Ensure pdf and png figures can be saved via visualization API call.
 
-    :param experiment_to_use: Object containing trained model and results to
-        test visualization
+    :param experiment_to_use: Object containing trained model and results to test visualization
     :return: None
     """
     experiment = experiment_to_use
@@ -376,8 +365,7 @@ def test_compare_classifiers_predictions_distribution_vis_api(experiment_to_use)
 def test_confidence_thresholding_vis_api(experiment_to_use):
     """Ensure pdf and png figures can be saved via visualization API call.
 
-    :param experiment_to_use: Object containing trained model and results to
-        test visualization
+    :param experiment_to_use: Object containing trained model and results to test visualization
     :return: None
     """
     experiment = experiment_to_use
@@ -403,8 +391,7 @@ def test_confidence_thresholding_vis_api(experiment_to_use):
 def test_confidence_thresholding_data_vs_acc_vis_api(experiment_to_use):
     """Ensure pdf and png figures can be saved via visualization API call.
 
-    :param experiment_to_use: Object containing trained model and results to
-        test visualization
+    :param experiment_to_use: Object containing trained model and results to test visualization
     :return: None
     """
     experiment = experiment_to_use
@@ -430,8 +417,7 @@ def test_confidence_thresholding_data_vs_acc_vis_api(experiment_to_use):
 def test_confidence_thresholding_data_vs_acc_subset_vis_api(experiment_to_use):
     """Ensure pdf and png figures can be saved via visualization API call.
 
-    :param experiment_to_use: Object containing trained model and results to
-        test visualization
+    :param experiment_to_use: Object containing trained model and results to test visualization
     :return: None
     """
     experiment = experiment_to_use
@@ -459,8 +445,7 @@ def test_confidence_thresholding_data_vs_acc_subset_vis_api(experiment_to_use):
 def test_confidence_thresholding_data_vs_acc_subset_per_class_vis_api(experiment_to_use):
     """Ensure pdf and png figures can be saved via visualization API call.
 
-    :param experiment_to_use: Object containing trained model and results to
-        test visualization
+    :param experiment_to_use: Object containing trained model and results to test visualization
     :return: None
     """
     experiment = experiment_to_use
@@ -635,8 +620,7 @@ def test_confidence_thresholding_2thresholds_3d_vis_api(csv_filename):
 def test_binary_threshold_vs_metric_vis_api(experiment_to_use):
     """Ensure pdf and png figures can be saved via visualization API call.
 
-    :param experiment_to_use: Object containing trained model and results to
-        test visualization
+    :param experiment_to_use: Object containing trained model and results to test visualization
     :return: None
     """
     experiment = experiment_to_use
@@ -665,8 +649,7 @@ def test_binary_threshold_vs_metric_vis_api(experiment_to_use):
 def test_precision_recall_curves_vis_api(experiment_to_use):
     """Ensure pdf and png figures can be saved via visualization API call.
 
-    :param experiment_to_use: Object containing trained model and results to
-        test visualization
+    :param experiment_to_use: Object containing trained model and results to test visualization
     :return: None
     """
     experiment = experiment_to_use
@@ -727,8 +710,7 @@ def test_precision_recall_curves_from_test_statistics_vis_api(csv_filename):
 def test_roc_curves_vis_api(experiment_to_use):
     """Ensure pdf and png figures can be saved via visualization API call.
 
-    :param experiment_to_use: Object containing trained model and results to
-        test visualization
+    :param experiment_to_use: Object containing trained model and results to test visualization
     :return: None
     """
     experiment = experiment_to_use
@@ -789,8 +771,7 @@ def test_roc_curves_from_test_statistics_vis_api(csv_filename):
 def test_calibration_1_vs_all_vis_api(experiment_to_use):
     """Ensure pdf and png figures can be saved via visualization API call.
 
-    :param experiment_to_use: Object containing trained model and results to
-        test visualization
+    :param experiment_to_use: Object containing trained model and results to test visualization
     :return: None
     """
     experiment = experiment_to_use
@@ -817,8 +798,7 @@ def test_calibration_1_vs_all_vis_api(experiment_to_use):
 def test_calibration_multiclass_vis_api(experiment_to_use):
     """Ensure pdf and png figures can be saved via visualization API call.
 
-    :param experiment_to_use: Object containing trained model and results to
-        test visualization
+    :param experiment_to_use: Object containing trained model and results to test visualization
     :return: None
     """
     experiment = experiment_to_use
@@ -844,8 +824,7 @@ def test_calibration_multiclass_vis_api(experiment_to_use):
 def test_confusion_matrix_vis_api(experiment_to_use):
     """Ensure pdf and png figures can be saved via visualization API call.
 
-    :param experiment_to_use: Object containing trained model and results to
-        test visualization
+    :param experiment_to_use: Object containing trained model and results to test visualization
     :return: None
     """
     experiment = experiment_to_use
@@ -872,8 +851,7 @@ def test_confusion_matrix_vis_api(experiment_to_use):
 def test_frequency_vs_f1_vis_api(experiment_to_use):
     """Ensure pdf and png figures can be saved via visualization API call.
 
-    :param experiment_to_use: Object containing trained model and results to
-        test visualization
+    :param experiment_to_use: Object containing trained model and results to test visualization
     :return: None
     """
     experiment = experiment_to_use
diff --git a/tests/integration_tests/utils.py b/tests/integration_tests/utils.py
index 6eb6f23f564..fd7886199c3 100644
--- a/tests/integration_tests/utils.py
+++ b/tests/integration_tests/utils.py
@@ -23,7 +23,8 @@
 import tempfile
 import traceback
 import uuid
-from distutils.util import strtobool
+
+# from distutils.util import strtobool
 from typing import Any, Dict, List, Optional, Set, Tuple, TYPE_CHECKING, Union
 
 import cloudpickle
@@ -124,6 +125,16 @@ def train(self, *args, save_path=MODEL_FILE_NAME, **kwargs):
             return super().train(*args, save_path=tmpdir, **kwargs)
 
 
+def str2bool(val):
+    val = val.lower()
+    if val in ("y", "yes", "t", "true", "on", "1"):
+        return 1
+    elif val in ("n", "no", "f", "false", "off", "0"):
+        return 0
+    else:
+        raise ValueError(f"invalid truth value {val!r}")
+
+
 def parse_flag_from_env(key, default=False):
     try:
         value = os.environ[key]
@@ -135,7 +146,7 @@ def parse_flag_from_env(key, default=False):
         try:
             if isinstance(value, bool):
                 return 1 if value else 0
-            _value = strtobool(value)
+            _value = str2bool(value)
         except ValueError:
             # More values are supported, but let's keep the message simple.
             raise ValueError(f"If set, {key} must be yes or no.")
diff --git a/tests/ludwig/automl/test_base_config.py b/tests/ludwig/automl/test_base_config.py
index ad42c9deed5..f10a09fb015 100644
--- a/tests/ludwig/automl/test_base_config.py
+++ b/tests/ludwig/automl/test_base_config.py
@@ -8,8 +8,8 @@
 
 ray = pytest.importorskip("ray")  # noqa
 
-from ludwig.automl.base_config import (  # noqa
-    get_dataset_info,
+from ludwig.automl.base_config import get_dataset_info  # noqa
+from ludwig.automl.base_config import (
     get_dataset_info_from_source,
     get_field_metadata,
     get_reference_configs,
@@ -18,7 +18,8 @@
 from ludwig.data.dataframe.dask import DaskEngine  # noqa
 from ludwig.data.dataframe.pandas import PandasEngine  # noqa
 from ludwig.schema.model_types.base import ModelConfig  # noqa
-from ludwig.utils.automl.data_source import DataframeSource, wrap_data_source  # noqa
+from ludwig.utils.automl.data_source import DataframeSource  # noqa
+from ludwig.utils.automl.data_source import wrap_data_source
 
 pytestmark = pytest.mark.distributed
 
diff --git a/tests/ludwig/config_validation/test_checks.py b/tests/ludwig/config_validation/test_checks.py
index c614a195191..7082cce0bce 100644
--- a/tests/ludwig/config_validation/test_checks.py
+++ b/tests/ludwig/config_validation/test_checks.py
@@ -2,9 +2,7 @@
 
 Note that all testing should be done with the public API, rather than individual checks.
 
-```
-ModelConfig.from_dict(config)
-```
+``` ModelConfig.from_dict(config) ```
 """
 
 import contextlib
diff --git a/tests/ludwig/data/test_ray_data.py b/tests/ludwig/data/test_ray_data.py
index a71c8ae910b..bd7cb20b871 100644
--- a/tests/ludwig/data/test_ray_data.py
+++ b/tests/ludwig/data/test_ray_data.py
@@ -9,7 +9,8 @@
 ray = pytest.importorskip("ray")  # noqa
 dask = pytest.importorskip("dask")  # noqa
 
-from ludwig.data.dataset.ray import RayDatasetBatcher, read_remote_parquet  # noqa
+from ludwig.data.dataset.ray import RayDatasetBatcher  # noqa
+from ludwig.data.dataset.ray import read_remote_parquet
 
 # Mark the entire module as distributed
 pytestmark = pytest.mark.distributed
diff --git a/tests/ludwig/features/test_sequence_features.py b/tests/ludwig/features/test_sequence_features.py
index ed158475aea..e4e84d7ef39 100644
--- a/tests/ludwig/features/test_sequence_features.py
+++ b/tests/ludwig/features/test_sequence_features.py
@@ -1,9 +1,9 @@
+# flake8: noqa: E501
 from typing import List, Tuple
 
 import numpy as np
 import pytest
 import torch
-import torchtext
 
 from ludwig.constants import ENCODER_OUTPUT, LAST_HIDDEN, LOGITS, SEQUENCE, TEXT, TYPE
 from ludwig.features.sequence_feature import _SequencePreprocessing, SequenceInputFeature, SequenceOutputFeature
@@ -192,9 +192,7 @@ def test_text_preproc_module_space_punct_tokenizer():
     )
 
 
-@pytest.mark.skipif(
-    torch.torch_version.TorchVersion(torchtext.__version__) < (0, 12, 0), reason="requires torchtext 0.12.0 or higher"
-)
+@pytest.mark.skip()
 def test_sequence_preproc_module_sentencepiece_tokenizer():
     metadata = {
         "preprocessing": {
@@ -227,9 +225,7 @@ def test_sequence_preproc_module_sentencepiece_tokenizer():
     )
 
 
-@pytest.mark.skipif(
-    torch.torch_version.TorchVersion(torchtext.__version__) < (0, 12, 0), reason="requires torchtext 0.12.0 or higher"
-)
+@pytest.mark.skip()
 def test_sequence_preproc_module_clip_tokenizer():
     metadata = {
         "preprocessing": {
@@ -260,9 +256,7 @@ def test_sequence_preproc_module_clip_tokenizer():
     )
 
 
-@pytest.mark.skipif(
-    torch.torch_version.TorchVersion(torchtext.__version__) < (0, 12, 0), reason="requires torchtext 0.12.0 or higher"
-)
+@pytest.mark.skip()
 def test_sequence_preproc_module_gpt2bpe_tokenizer():
     metadata = {
         "preprocessing": {
@@ -296,9 +290,7 @@ def test_sequence_preproc_module_gpt2bpe_tokenizer():
     )
 
 
-@pytest.mark.skipif(
-    torch.torch_version.TorchVersion(torchtext.__version__) < (0, 13, 0), reason="requires torchtext 0.13.0 or higher"
-)
+@pytest.mark.skip()
 def test_sequence_preproc_module_bert_tokenizer():
     metadata = {
         "preprocessing": {
diff --git a/tests/ludwig/marshmallow/test_marshmallow_misc.py b/tests/ludwig/marshmallow/test_marshmallow_misc.py
index 42aecf769f6..78ba8326459 100644
--- a/tests/ludwig/marshmallow/test_marshmallow_misc.py
+++ b/tests/ludwig/marshmallow/test_marshmallow_misc.py
@@ -8,7 +8,7 @@
 
 @dataclass
 class CustomTestSchema(BaseMarshmallowConfig):
-    """sample docstring."""
+    """Sample docstring."""
 
     foo: int = 5
     "foo (default: 5)"
diff --git a/tests/ludwig/utils/test_hyperopt_ray_utils.py b/tests/ludwig/utils/test_hyperopt_ray_utils.py
index 5d37d0aa292..0b4f180f4e8 100644
--- a/tests/ludwig/utils/test_hyperopt_ray_utils.py
+++ b/tests/ludwig/utils/test_hyperopt_ray_utils.py
@@ -78,7 +78,7 @@ def test_grid_strategy(key):
         "minimize",
         "validation",
         search_alg={TYPE: "variant_generator"},
-        **{"type": "ray", "num_samples": 2, "scheduler": {"type": "fifo"}}
+        **{"type": "ray", "num_samples": 2, "scheduler": {"type": "fifo"}},
     )
 
     search_space = hyperopt_executor.search_space
diff --git a/tests/ludwig/utils/test_tokenizers.py b/tests/ludwig/utils/test_tokenizers.py
index 82f6d86bdff..0fa8104ed10 100644
--- a/tests/ludwig/utils/test_tokenizers.py
+++ b/tests/ludwig/utils/test_tokenizers.py
@@ -1,58 +1,9 @@
-import os
-
-import pytest
-import torch
-import torchtext
-
-from ludwig.utils.tokenizers import EnglishLemmatizeFilterTokenizer, NgramTokenizer, StringSplitTokenizer
-
-TORCHTEXT_0_14_0_HF_NAMES = [
-    "bert-base-uncased",
-    "distilbert-base-uncased",
-    "google/electra-small-discriminator",
-    "dbmdz/bert-base-italian-cased",  # Community model
-    "nreimers/MiniLM-L6-H384-uncased",  # Community model
-    "emilyalsentzer/Bio_ClinicalBERT",  # Community model
-    "bionlp/bluebert_pubmed_mimic_uncased_L-12_H-768_A-12",  # Community model
-]
-
-
-@pytest.mark.parametrize(
-    "pretrained_model_name_or_path",
-    [
-        pytest.param(
-            model_name,
-            marks=[
-                pytest.mark.skipif(
-                    torch.torch_version.TorchVersion(torchtext.__version__) < (0, 14, 0),
-                    reason="requires torchtext 0.14.0 or higher",
-                ),
-            ],
-        )
-        for model_name in TORCHTEXT_0_14_0_HF_NAMES
-    ],
+from ludwig.utils.tokenizers import (
+    EnglishLemmatizeFilterTokenizer,
+    get_tokenizer_from_registry,
+    NgramTokenizer,
+    StringSplitTokenizer,
 )
-def test_bert_hf_tokenizer_parity(tmpdir, pretrained_model_name_or_path):
-    """Tests the BERTTokenizer implementation.
-
-    Asserts both tokens and token IDs are the same by initializing the BERTTokenizer as a standalone tokenizer and as a
-    HF tokenizer.
-    """
-    from ludwig.utils.tokenizers import get_hf_tokenizer, HFTokenizer
-
-    inputs = "Hello, ``I'm'' ónë of 1,205,000 sentences!"
-    hf_tokenizer = HFTokenizer(pretrained_model_name_or_path)
-    torchtext_tokenizer = get_hf_tokenizer(pretrained_model_name_or_path)
-
-    # Ensure that the tokenizer is scriptable
-    tokenizer_path = os.path.join(tmpdir, "tokenizer.pt")
-    torch.jit.script(torchtext_tokenizer).save(tokenizer_path)
-    torchtext_tokenizer = torch.jit.load(tokenizer_path)
-
-    token_ids_expected = hf_tokenizer(inputs)
-    token_ids = torchtext_tokenizer(inputs)
-
-    assert token_ids_expected == token_ids
 
 
 def test_ngram_tokenizer():
@@ -85,3 +36,46 @@ def test_english_lemmatize_filter_tokenizer():
     tokenizer = EnglishLemmatizeFilterTokenizer()
     tokens = tokenizer(inputs)
     assert len(tokens) > 0
+
+
+def test_sentence_piece_tokenizer():
+    inputs = "This is a sentence. And this is another one."
+    tokenizer = get_tokenizer_from_registry("sentencepiece")()
+    tokens = tokenizer(inputs)
+    assert tokens == ["▁This", "▁is", "▁a", "▁sentence", ".", "▁And", "▁this", "▁is", "▁another", "▁one", "."]
+
+
+def test_clip_tokenizer():
+    inputs = "This is a sentence. And this is another one."
+    tokenizer = get_tokenizer_from_registry("clip")()
+    tokens = tokenizer(inputs)
+    print(tokens)
+    assert tokens == [
+        "this</w>",
+        "is</w>",
+        "a</w>",
+        "sentence</w>",
+        ".</w>",
+        "and</w>",
+        "this</w>",
+        "is</w>",
+        "another</w>",
+        "one</w>",
+        ".</w>",
+    ]
+
+
+def test_gpt2_bpe_tokenizer():
+    inputs = "This is a sentence. And this is another one."
+    tokenizer = get_tokenizer_from_registry("gpt2bpe")()
+    tokens = tokenizer(inputs)
+    print(tokens)
+    assert tokens == ["This", "Ġis", "Ġa", "Ġsentence", ".", "ĠAnd", "Ġthis", "Ġis", "Ġanother", "Ġone", "."]
+
+
+def test_bert_tokenizer():
+    inputs = "This is a sentence. And this is another one."
+    tokenizer = get_tokenizer_from_registry("bert")()
+    tokens = tokenizer(inputs)
+    print(tokens)
+    assert tokens == ["this", "is", "a", "sentence", ".", "and", "this", "is", "another", "one", "."]