diff --git a/.github/workflows/build.yaml b/.github/workflows/build.yaml
new file mode 100644
index 0000000..03ded27
--- /dev/null
+++ b/.github/workflows/build.yaml
@@ -0,0 +1,114 @@
+name: build
+
+on:
+  push:
+    branches:
+      - "branch-*"
+    tags:
+      - v[0-9][0-9].[0-9][0-9].[0-9][0-9]
+  workflow_dispatch:
+    inputs:
+      branch:
+        required: true
+        type: string
+      date:
+        required: true
+        type: string
+      sha:
+        required: true
+        type: string
+      build_type:
+        type: string
+        default: nightly
+
+concurrency:
+  group: ${{ github.workflow }}-${{ github.ref }}-${{ github.event_name }}
+  cancel-in-progress: true
+
+jobs:
+  cpp-build:
+    secrets: inherit
+    uses: rapidsai/shared-workflows/.github/workflows/conda-cpp-build.yaml@branch-24.12
+    with:
+      build_type: ${{ inputs.build_type || 'branch' }}
+      branch: ${{ inputs.branch }}
+      date: ${{ inputs.date }}
+      sha: ${{ inputs.sha }}
+      node_type: cpu32
+  python-build:
+    needs: [cpp-build]
+    secrets: inherit
+    uses: rapidsai/shared-workflows/.github/workflows/conda-python-build.yaml@branch-24.12
+    with:
+      build_type: ${{ inputs.build_type || 'branch' }}
+      branch: ${{ inputs.branch }}
+      date: ${{ inputs.date }}
+      sha: ${{ inputs.sha }}
+  upload-conda:
+    needs: [cpp-build, python-build]
+    secrets: inherit
+    uses: rapidsai/shared-workflows/.github/workflows/conda-upload-packages.yaml@branch-24.12
+    with:
+      build_type: ${{ inputs.build_type || 'branch' }}
+      branch: ${{ inputs.branch }}
+      date: ${{ inputs.date }}
+      sha: ${{ inputs.sha }}
+  wheel-build-cugraph-dgl:
+    needs: wheel-publish-pylibwholegraph
+    secrets: inherit
+    uses: rapidsai/shared-workflows/.github/workflows/wheels-build.yaml@branch-24.12
+    with:
+      build_type: ${{ inputs.build_type || 'branch' }}
+      branch: ${{ inputs.branch }}
+      sha: ${{ inputs.sha }}
+      date: ${{ inputs.date }}
+      script: ci/build_wheel_cugraph-dgl.sh
+  wheel-publish-cugraph-dgl:
+    needs: wheel-build-cugraph-dgl
+    secrets: inherit
+    uses: rapidsai/shared-workflows/.github/workflows/wheels-publish.yaml@branch-24.12
+    with:
+      build_type: ${{ inputs.build_type || 'branch' }}
+      branch: ${{ inputs.branch }}
+      sha: ${{ inputs.sha }}
+      date: ${{ inputs.date }}
+      package-name: cugraph-dgl
+  wheel-build-cugraph-pyg:
+    needs: wheel-publish-pylibwholegraph
+    secrets: inherit
+    uses: rapidsai/shared-workflows/.github/workflows/wheels-build.yaml@branch-24.12
+    with:
+      build_type: ${{ inputs.build_type || 'branch' }}
+      branch: ${{ inputs.branch }}
+      sha: ${{ inputs.sha }}
+      date: ${{ inputs.date }}
+      script: ci/build_wheel_cugraph-pyg.sh
+  wheel-publish-cugraph-pyg:
+    needs: wheel-build-cugraph-pyg
+    secrets: inherit
+    uses: rapidsai/shared-workflows/.github/workflows/wheels-publish.yaml@branch-24.12
+    with:
+      build_type: ${{ inputs.build_type || 'branch' }}
+      branch: ${{ inputs.branch }}
+      sha: ${{ inputs.sha }}
+      date: ${{ inputs.date }}
+      package-name: cugraph-pyg
+  wheel-build-pylibwholegraph:
+    secrets: inherit
+    uses: rapidsai/shared-workflows/.github/workflows/wheels-build.yaml@branch-24.12
+    with:
+      build_type: ${{ inputs.build_type || 'branch' }}
+      branch: ${{ inputs.branch }}
+      sha: ${{ inputs.sha }}
+      date: ${{ inputs.date }}
+      script: ci/build_wheel_pylibwholegraph.sh
+  wheel-publish-pylibwholegraph:
+    needs: wheel-build-pylibwholegraph
+    secrets: inherit
+    uses: rapidsai/shared-workflows/.github/workflows/wheels-publish.yaml@branch-24.12
+    with:
+      build_type: ${{ inputs.build_type || 'branch' }}
+      branch: ${{ inputs.branch }}
+      sha: ${{ inputs.sha }}
+      date: ${{ inputs.date }}
+      package-name: pylibwholegraph
diff --git a/.github/workflows/pr.yaml b/.github/workflows/pr.yaml
index 8628d46..06fee99 100644
--- a/.github/workflows/pr.yaml
+++ b/.github/workflows/pr.yaml
@@ -16,10 +16,17 @@ jobs:
       - checks
       - conda-cpp-build
       - conda-cpp-tests
+      # - conda-cpp-checks
+      - conda-notebook-tests
       - conda-python-build
       - conda-python-tests
       - wheel-build-pylibwholegraph
       - wheel-tests-pylibwholegraph
+      - wheel-build-cugraph-dgl
+      - wheel-tests-cugraph-dgl
+      - wheel-build-cugraph-pyg
+      - wheel-tests-cugraph-pyg
+      - devcontainer
     secrets: inherit
     uses: rapidsai/shared-workflows/.github/workflows/pr-builder.yaml@branch-24.12
     if: always()
@@ -74,6 +81,14 @@ jobs:
     if: fromJSON(needs.changed-files.outputs.changed_file_groups).test_cpp
     with:
       build_type: pull-request
+  # conda-cpp-checks:
+  #   needs: conda-cpp-build
+  #   secrets: inherit
+  #   uses: rapidsai/shared-workflows/.github/workflows/conda-cpp-post-build-checks.yaml@branch-24.12
+  #   with:
+  #     build_type: pull-request
+  #     enable_check_symbols: true
+  #     symbol_exclusions: (cugraph::ops|hornet|void writeEdgeCountsKernel|void markUniqueOffsetsKernel)
   conda-python-build:
     needs: conda-cpp-build
     secrets: inherit
@@ -87,6 +102,17 @@ jobs:
     if: fromJSON(needs.changed-files.outputs.changed_file_groups).test_python
     with:
       build_type: pull-request
+  conda-notebook-tests:
+    needs: [conda-python-build, changed-files]
+    secrets: inherit
+    uses: rapidsai/shared-workflows/.github/workflows/custom-job.yaml@branch-24.12
+    if: fromJSON(needs.changed-files.outputs.changed_file_groups).test_notebooks
+    with:
+      build_type: pull-request
+      node_type: "gpu-v100-latest-1"
+      arch: "amd64"
+      container_image: "rapidsai/ci-conda:cuda11.8.0-ubuntu22.04-py3.10"
+      run_script: "ci/test_notebooks.sh"
   wheel-build-pylibwholegraph:
     needs: checks
     secrets: inherit
@@ -104,3 +130,46 @@ jobs:
       build_type: pull-request
       script: ci/test_wheel_pylibwholegraph.sh
       matrix_filter: map(select(.ARCH == "amd64"))
+  wheel-build-cugraph-dgl:
+    needs: wheel-tests-pylibwholegraph
+    secrets: inherit
+    uses: rapidsai/shared-workflows/.github/workflows/wheels-build.yaml@branch-24.12
+    with:
+      build_type: pull-request
+      script: ci/build_wheel_cugraph-dgl.sh
+  wheel-tests-cugraph-dgl:
+    needs: [wheel-build-cugraph-dgl, changed-files]
+    secrets: inherit
+    uses: rapidsai/shared-workflows/.github/workflows/wheels-test.yaml@branch-24.12
+    if: fromJSON(needs.changed-files.outputs.changed_file_groups).test_python
+    with:
+      build_type: pull-request
+      script: ci/test_wheel_cugraph-dgl.sh
+      matrix_filter: map(select(.ARCH == "amd64"))
+  wheel-build-cugraph-pyg:
+    needs: wheel-tests-pylibwholegraph
+    secrets: inherit
+    uses: rapidsai/shared-workflows/.github/workflows/wheels-build.yaml@branch-24.12
+    with:
+      build_type: pull-request
+      script: ci/build_wheel_cugraph-pyg.sh
+  wheel-tests-cugraph-pyg:
+    needs: [wheel-build-cugraph-pyg, changed-files]
+    secrets: inherit
+    uses: rapidsai/shared-workflows/.github/workflows/wheels-test.yaml@branch-24.12
+    if: fromJSON(needs.changed-files.outputs.changed_file_groups).test_python
+    with:
+      build_type: pull-request
+      script: ci/test_wheel_cugraph-pyg.sh
+      matrix_filter: map(select(.ARCH == "amd64"))
+  devcontainer:
+    secrets: inherit
+    uses: rapidsai/shared-workflows/.github/workflows/build-in-devcontainer.yaml@branch-24.12
+    with:
+      arch: '["amd64"]'
+      cuda: '["12.5"]'
+      node_type: cpu32
+      build_command: |
+        sccache -z;
+        build-all --verbose -j$(nproc --ignore=1);
+        sccache -s;
diff --git a/.github/workflows/test.yaml b/.github/workflows/test.yaml
new file mode 100644
index 0000000..187f4b4
--- /dev/null
+++ b/.github/workflows/test.yaml
@@ -0,0 +1,72 @@
+name: test
+
+on:
+  workflow_dispatch:
+    inputs:
+      branch:
+        required: true
+        type: string
+      date:
+        required: true
+        type: string
+      sha:
+        required: true
+        type: string
+
+jobs:
+  conda-cpp-checks:
+    secrets: inherit
+    uses: rapidsai/shared-workflows/.github/workflows/conda-cpp-post-build-checks.yaml@branch-24.12
+    with:
+      build_type: nightly
+      branch: ${{ inputs.branch }}
+      date: ${{ inputs.date }}
+      sha: ${{ inputs.sha }}
+      enable_check_symbols: true
+      symbol_exclusions: (cugraph::ops|hornet|void writeEdgeCountsKernel|void markUniqueOffsetsKernel)
+  conda-cpp-tests:
+    secrets: inherit
+    uses: rapidsai/shared-workflows/.github/workflows/conda-cpp-tests.yaml@branch-24.12
+    with:
+      build_type: nightly
+      branch: ${{ inputs.branch }}
+      date: ${{ inputs.date }}
+      sha: ${{ inputs.sha }}
+  conda-python-tests:
+    secrets: inherit
+    uses: rapidsai/shared-workflows/.github/workflows/conda-python-tests.yaml@branch-24.12
+    with:
+      build_type: nightly
+      branch: ${{ inputs.branch }}
+      date: ${{ inputs.date }}
+      sha: ${{ inputs.sha }}
+  wheel-tests-pylibwholegraph:
+    secrets: inherit
+    uses: rapidsai/shared-workflows/.github/workflows/wheels-test.yaml@branch-24.12
+    with:
+      build_type: nightly
+      branch: ${{ inputs.branch }}
+      date: ${{ inputs.date }}
+      sha: ${{ inputs.sha }}
+      script: ci/test_wheel_pylibwholegraph.sh
+      matrix_filter: map(select(.ARCH == "amd64"))
+  wheel-tests-cugraph-dgl:
+    secrets: inherit
+    uses: rapidsai/shared-workflows/.github/workflows/wheels-test.yaml@branch-24.12
+    with:
+      build_type: nightly
+      branch: ${{ inputs.branch }}
+      date: ${{ inputs.date }}
+      sha: ${{ inputs.sha }}
+      script: ci/test_wheel_cugraph-dgl.sh
+      matrix_filter: map(select(.ARCH == "amd64"))
+  wheel-tests-cugraph-pyg:
+    secrets: inherit
+    uses: rapidsai/shared-workflows/.github/workflows/wheels-test.yaml@branch-24.12
+    with:
+      build_type: nightly
+      branch: ${{ inputs.branch }}
+      date: ${{ inputs.date }}
+      sha: ${{ inputs.sha }}
+      script: ci/test_wheel_cugraph-pyg.sh
+      matrix_filter: map(select(.ARCH == "amd64"))
diff --git a/build.sh b/build.sh
index edece26..435567b 100755
--- a/build.sh
+++ b/build.sh
@@ -29,12 +29,14 @@ VALIDARGS="
    pylibwholegraph
    libwholegraph
    tests
+   benchmarks
    all
    -v
    -g
    -n
    --pydevelop
    --allgpuarch
+   --compile-cmd
    --clean
    -h
    --help
@@ -45,19 +47,22 @@ HELP="$0 [<target> ...] [<flag> ...]
    clean                      - remove all existing build artifacts and configuration (start over)
    uninstall                  - uninstall libcugraph and cugraph from a prior build/install (see also -n)
    cugraph-pyg                - build the cugraph-pyg Python package
-   cugraph-dgl                - build the cugraph-dgl extensions for DGL
+   cugraph-dgl                - build the cugraph-dgl Python package
    pylibwholegraph            - build the pylibwholegraph Python package
    libwholegraph              - build the libwholegraph library
    tests                      - build the C++ tests
+   benchmarks                 - build benchmarks.
    all                        - build everything
  and <flag> is:
-   -v                         - verbose build mode
-   -g                         - build for debug
-   -n                         - do not install after a successful build (does not affect Python packages)
-   --pydevelop                - install the Python packages in editable mode
-   --allgpuarch               - build for all supported GPU architectures
-   --clean                    - clean an individual target (note: to do a complete rebuild, use the clean target described above)
-   -h                         - print this text
+   -v                          - verbose build mode
+   -g                          - build for debug
+   -n                          - do not install after a successful build (does not affect Python packages)
+   --pydevelop                 - install the Python packages in editable mode
+   --allgpuarch                - build for all supported GPU architectures
+   --enable-nvshmem            - build with nvshmem support (beta).
+   --compile-cmd               - only output compile commands (invoke CMake without build)
+   --clean                     - clean an individual target (note: to do a complete rebuild, use the clean target described above)
+   -h                          - print this text
 
  default action (no args) is to build and install 'libwholegraph' then 'pylibwholegraph' then 'cugraph-pyg' then 'cugraph-dgl'
 
@@ -139,6 +144,21 @@ fi
 if hasArg --pydevelop; then
     PYTHON_ARGS_FOR_INSTALL="${PYTHON_ARGS_FOR_INSTALL} -e"
 fi
+if hasArg --enable-nvshmem; then
+    BUILD_WITH_NVSHMEM=ON
+else
+    BUILD_WITH_NVSHMEM=OFF
+fi
+if hasArg tests; then
+    BUILD_TESTS=ON
+else
+    BUILD_TESTS=OFF
+fi
+if hasArg benchmarks; then
+    BUILD_BENCHMARKS=ON
+else
+    BUILD_BENCHMARKS=OFF
+fi
 
 if hasArg tests; then
     BUILD_TESTS=ON
@@ -189,6 +209,7 @@ if hasArg clean; then
 fi
 
 ################################################################################
+
 # Build and install the libwholegraph library
 if hasArg libwholegraph || buildDefault || hasArg all ; then
 
@@ -250,7 +271,7 @@ if hasArg cugraph-pyg || buildDefault || hasArg all; then
     fi
 fi
 
-# Install the cugraph-dgl extensions for DGL
+# Build and install the cugraph-dgl Python package
 if hasArg cugraph-dgl || buildDefault ||hasArg all; then
     if hasArg --clean; then
         cleanPythonDir ${REPODIR}/python/cugraph-dgl
diff --git a/ci/build_python.sh b/ci/build_python.sh
index 49d3c12..9f42948 100755
--- a/ci/build_python.sh
+++ b/ci/build_python.sh
@@ -25,4 +25,16 @@ RAPIDS_PACKAGE_VERSION=$(head -1 ./VERSION) rapids-conda-retry mambabuild \
   --channel "${CPP_CHANNEL}" \
   conda/recipes/pylibwholegraph
 
+RAPIDS_PACKAGE_VERSION=$(head -1 ./VERSION) rapids-conda-retry mambabuild \
+  --no-test \
+  --channel "${CPP_CHANNEL}" \
+  --channel "${RAPIDS_CONDA_BLD_OUTPUT_DIR}" \
+  conda/recipes/cugraph-pyg
+
+RAPIDS_PACKAGE_VERSION=$(head -1 ./VERSION) rapids-conda-retry mambabuild \
+  --no-test \
+  --channel "${CPP_CHANNEL}" \
+  --channel "${RAPIDS_CONDA_BLD_OUTPUT_DIR}" \
+  conda/recipes/cugraph-dgl
+
 rapids-upload-conda-to-s3 python
diff --git a/ci/run_cugraph_dgl_pytests.sh b/ci/run_cugraph_dgl_pytests.sh
index 83c26a5..e85eb04 100755
--- a/ci/run_cugraph_dgl_pytests.sh
+++ b/ci/run_cugraph_dgl_pytests.sh
@@ -4,6 +4,6 @@
 set -euo pipefail
 
 # Support invoking run_cugraph_dgl_pytests.sh outside the script directory
-cd "$(dirname "$(realpath "${BASH_SOURCE[0]}")")"/../python/cugraph-dgl/tests
+cd "$(dirname "$(realpath "${BASH_SOURCE[0]}")")"/../python/cugraph-dgl/cugraph_dgl
 
 pytest --cache-clear --ignore=mg "$@" .
diff --git a/ci/test_notebooks.sh b/ci/test_notebooks.sh
index 31ec560..59b60e4 100755
--- a/ci/test_notebooks.sh
+++ b/ci/test_notebooks.sh
@@ -5,6 +5,8 @@ set -Eeuo pipefail
 
 . /opt/conda/etc/profile.d/conda.sh
 
+RAPIDS_VERSION="$(rapids-version)"
+
 rapids-logger "Generate notebook testing dependencies"
 rapids-dependency-file-generator \
   --output conda \
@@ -27,7 +29,9 @@ PYTHON_CHANNEL=$(rapids-download-conda-from-s3 python)
 rapids-mamba-retry install \
   --channel "${CPP_CHANNEL}" \
   --channel "${PYTHON_CHANNEL}" \
-  libcugraph pylibcugraph cugraph
+  "libcugraph=${RAPIDS_VERSION}" \
+  "pylibcugraph=${RAPIDS_VERSION}" \
+  "cugraph=${RAPIDS_VERSION}"
 
 NBTEST="$(realpath "$(dirname "$0")/utils/nbtest.sh")"
 NOTEBOOK_LIST="$(realpath "$(dirname "$0")/notebook_list.py")"
diff --git a/ci/test_python.sh b/ci/test_python.sh
index dd5f539..68cdf2c 100755
--- a/ci/test_python.sh
+++ b/ci/test_python.sh
@@ -27,10 +27,113 @@ RAPIDS_TESTS_DIR=${RAPIDS_TESTS_DIR:-"${PWD}/test-results"}
 RAPIDS_COVERAGE_DIR=${RAPIDS_COVERAGE_DIR:-"${PWD}/coverage-results"}
 mkdir -p "${RAPIDS_TESTS_DIR}" "${RAPIDS_COVERAGE_DIR}"
 
+# RAPIDS_DATASET_ROOT_DIR is used by test scripts
+export RAPIDS_DATASET_ROOT_DIR="$(realpath datasets)"
+mkdir -p "${RAPIDS_DATASET_ROOT_DIR}"
+pushd "${RAPIDS_DATASET_ROOT_DIR}"
+./get_test_data.sh --benchmark
+popd
+
 EXITCODE=0
 trap "EXITCODE=1" ERR
 set +e
 
+# Test runs that include tests that use dask require
+# --import-mode=append. Those tests start a LocalCUDACluster that inherits
+# changes from pytest's modifications to PYTHONPATH (which defaults to
+# prepending source tree paths to PYTHONPATH).  This causes the
+# LocalCUDACluster subprocess to import cugraph from the source tree instead of
+# the install location, and in most cases, the source tree does not have
+# extensions built in-place and will result in ImportErrors.
+#
+# FIXME: TEMPORARILY disable MG PropertyGraph tests (experimental) tests and
+# bulk sampler IO tests (hangs in CI)
+
+if [[ "${RUNNER_ARCH}" != "ARM64" ]]; then
+  rapids-mamba-retry env create --yes -f env.yaml -n test_cugraph_dgl
+
+  # activate test_cugraph_dgl environment for dgl
+  set +u
+  conda activate test_cugraph_dgl
+  set -u
+
+  rapids-mamba-retry install \
+    --channel "${CPP_CHANNEL}" \
+    --channel "${PYTHON_CHANNEL}" \
+    --channel pytorch \
+    --channel conda-forge \
+    --channel dglteam/label/th23_cu118 \
+    --channel nvidia \
+    "pylibwholegraph=${RAPIDS_VERSION}" \
+    "pylibcugraphops=${RAPIDS_VERSION}" \
+    "cugraph=${RAPIDS_VERSION}" \
+    "cugraph-dgl=${RAPIDS_VERSION}" \
+    'pytorch::pytorch>=2.3,<2.4' \
+    'cuda-version=11.8' \
+    "ogb"
+
+  rapids-print-env
+
+  rapids-logger "Check GPU usage"
+  nvidia-smi
+
+  rapids-logger "pytest cugraph_dgl (single GPU)"
+  ./ci/run_cugraph_dgl_pytests.sh \
+    --junitxml="${RAPIDS_TESTS_DIR}/junit-cugraph-dgl.xml" \
+    --cov-config=../../.coveragerc \
+    --cov=cugraph_dgl \
+    --cov-report=xml:"${RAPIDS_COVERAGE_DIR}/cugraph-dgl-coverage.xml" \
+    --cov-report=term
+
+  # Reactivate the test environment back
+  set +u
+  conda deactivate
+  set -u
+else
+  rapids-logger "skipping cugraph_dgl pytest on ARM64"
+fi
+
+if [[ "${RUNNER_ARCH}" != "ARM64" ]]; then
+  rapids-mamba-retry env create --yes -f env.yaml -n test_cugraph_pyg
+
+  # Temporarily allow unbound variables for conda activation.
+  set +u
+  conda activate test_cugraph_pyg
+  set -u
+
+  # Will automatically install built dependencies of cuGraph-PyG
+  rapids-mamba-retry install \
+    --channel "${CPP_CHANNEL}" \
+    --channel "${PYTHON_CHANNEL}" \
+    --channel pytorch \
+    "pylibwholegraph=${RAPIDS_VERSION}" \
+    "pylibcugraphops=${RAPIDS_VERSION}" \
+    "cugraph=${RAPIDS_VERSION}" \
+    "cugraph-pyg=${RAPIDS_VERSION}" \
+    'pytorch::pytorch>=2.3,<2.4' \
+    'ogb'
+
+  rapids-print-env
+
+  rapids-logger "Check GPU usage"
+  nvidia-smi
+
+  rapids-logger "pytest cugraph_pyg (single GPU)"
+  ./ci/run_cugraph_pyg_pytests.sh \
+    --junitxml="${RAPIDS_TESTS_DIR}/junit-cugraph-pyg.xml" \
+    --cov-config=../../.coveragerc \
+    --cov=cugraph_pyg \
+    --cov-report=xml:"${RAPIDS_COVERAGE_DIR}/cugraph-pyg-coverage.xml" \
+    --cov-report=term
+
+  # Reactivate the test environment back
+  set +u
+  conda deactivate
+  set -u
+else
+  rapids-logger "skipping cugraph_pyg pytest on ARM64"
+fi
+
 if [[ "${RUNNER_ARCH}" != "ARM64" ]]; then
   rapids-mamba-retry env create --yes -f env.yaml -n test_pylibwholegraph
 
diff --git a/ci/test_wheel_cugraph-dgl.sh b/ci/test_wheel_cugraph-dgl.sh
index 564b46c..d709145 100755
--- a/ci/test_wheel_cugraph-dgl.sh
+++ b/ci/test_wheel_cugraph-dgl.sh
@@ -4,24 +4,15 @@
 set -eoxu pipefail
 
 package_name="cugraph-dgl"
-package_dir="python/cugraph-dgl"
-
-python_package_name=$(echo ${package_name}|sed 's/-/_/g')
 
 mkdir -p ./dist
 RAPIDS_PY_CUDA_SUFFIX="$(rapids-wheel-ctk-name-gen ${RAPIDS_CUDA_VERSION})"
 
-# Download wheels built during this job.
-RAPIDS_PY_WHEEL_NAME="pylibcugraph_${RAPIDS_PY_CUDA_SUFFIX}" rapids-download-wheels-from-s3 ./local-deps
-RAPIDS_PY_WHEEL_NAME="cugraph_${RAPIDS_PY_CUDA_SUFFIX}" rapids-download-wheels-from-s3 ./local-deps
-python -m pip install ./local-deps/*.whl
-
-# use 'ls' to expand wildcard before adding `[extra]` requires for pip
+# Download the pylibwholegraph and cugraph-dgl built in the previous step
+RAPIDS_PY_WHEEL_NAME="pylibwholegraph_${RAPIDS_PY_CUDA_SUFFIX}" rapids-download-wheels-from-s3 ./local-deps
 RAPIDS_PY_WHEEL_NAME="${package_name}_${RAPIDS_PY_CUDA_SUFFIX}" RAPIDS_PY_WHEEL_PURE="1" rapids-download-wheels-from-s3 ./dist
-# pip creates wheels using python package names
-python -m pip install $(ls ./dist/${python_package_name}*.whl)[test]
-
 
+# determine pytorch and DGL sources
 PKG_CUDA_VER="$(echo ${CUDA_VERSION} | cut -d '.' -f1,2 | tr -d '.')"
 PKG_CUDA_VER_MAJOR=${PKG_CUDA_VER:0:2}
 if [[ "${PKG_CUDA_VER_MAJOR}" == "12" ]]; then
@@ -30,20 +21,16 @@ else
   PYTORCH_CUDA_VER=$PKG_CUDA_VER
 fi
 PYTORCH_URL="https://download.pytorch.org/whl/cu${PYTORCH_CUDA_VER}"
-DGL_URL="https://data.dgl.ai/wheels/cu${PYTORCH_CUDA_VER}/repo.html"
-
-# Starting from 2.2, PyTorch wheels depend on nvidia-nccl-cuxx>=2.19 wheel and
-# dynamically link to NCCL. RAPIDS CUDA 11 CI images have an older NCCL version that
-# might shadow the newer NCCL required by PyTorch during import (when importing
-# `cupy` before `torch`).
-if [[ "${NCCL_VERSION}" < "2.19" ]]; then
-  PYTORCH_VER="2.1.0"
-else
-  PYTORCH_VER="2.3.0"
-fi
-
-rapids-logger "Installing PyTorch and DGL"
-rapids-retry python -m pip install "torch==${PYTORCH_VER}" --index-url ${PYTORCH_URL}
-rapids-retry python -m pip install dgl==2.0.0 --find-links ${DGL_URL}
-
-python -m pytest python/cugraph-dgl/tests
+DGL_URL="https://data.dgl.ai/wheels/torch-2.3/cu${PYTORCH_CUDA_VER}/repo.html"
+
+# echo to expand wildcard before adding `[extra]` requires for pip
+python -m pip install \
+    -v \
+    --extra-index-url "${PYTORCH_URL}" \
+    --find-links "${DGL_URL}" \
+    "$(echo ./local-deps/pylibwholegraph_${RAPIDS_PY_CUDA_SUFFIX}*.whl)" \
+    "$(echo ./dist/cugraph_dgl_${RAPIDS_PY_CUDA_SUFFIX}*.whl)[test]" \
+    'dgl==2.4.0' \
+    'torch>=2.3.0,<2.4'
+
+python -m pytest python/cugraph-dgl/cugraph_dgl/tests
diff --git a/ci/test_wheel_cugraph-pyg.sh b/ci/test_wheel_cugraph-pyg.sh
index 1004063..3a4bf5f 100755
--- a/ci/test_wheel_cugraph-pyg.sh
+++ b/ci/test_wheel_cugraph-pyg.sh
@@ -4,46 +4,42 @@
 set -eoxu pipefail
 
 package_name="cugraph-pyg"
-package_dir="python/cugraph-pyg"
-
-python_package_name=$(echo ${package_name}|sed 's/-/_/g')
 
 mkdir -p ./dist
 RAPIDS_PY_CUDA_SUFFIX="$(rapids-wheel-ctk-name-gen ${RAPIDS_CUDA_VERSION})"
 
-# Download wheels built during this job.
-RAPIDS_PY_WHEEL_NAME="pylibcugraph_${RAPIDS_PY_CUDA_SUFFIX}" rapids-download-wheels-from-s3 ./local-deps
-RAPIDS_PY_WHEEL_NAME="cugraph_${RAPIDS_PY_CUDA_SUFFIX}" rapids-download-wheels-from-s3 ./local-deps
-python -m pip install ./local-deps/*.whl
-
-# use 'ls' to expand wildcard before adding `[extra]` requires for pip
+# Download the pylibwholegraph and cugraph-pyg built in the previous step
+RAPIDS_PY_WHEEL_NAME="pylibwholegraph_${RAPIDS_PY_CUDA_SUFFIX}" rapids-download-wheels-from-s3 ./local-deps
 RAPIDS_PY_WHEEL_NAME="${package_name}_${RAPIDS_PY_CUDA_SUFFIX}" RAPIDS_PY_WHEEL_PURE="1" rapids-download-wheels-from-s3 ./dist
-# pip creates wheels using python package names
-python -m pip install $(ls ./dist/${python_package_name}*.whl)[test]
-
-# RAPIDS_DATASET_ROOT_DIR is used by test scripts
-export RAPIDS_DATASET_ROOT_DIR="$(realpath datasets)"
-
-# Used to skip certain examples in CI due to memory limitations
-export CI_RUN=1
 
+# determine pytorch and pyg sources
 if [[ "${CUDA_VERSION}" == "11.8.0" ]]; then
   PYTORCH_URL="https://download.pytorch.org/whl/cu118"
-  PYG_URL="https://data.pyg.org/whl/torch-2.1.0+cu118.html"
+  PYG_URL="https://data.pyg.org/whl/torch-2.3.0+cu118.html"
 else
   PYTORCH_URL="https://download.pytorch.org/whl/cu121"
-  PYG_URL="https://data.pyg.org/whl/torch-2.1.0+cu121.html"
+  PYG_URL="https://data.pyg.org/whl/torch-2.3.0+cu121.html"
 fi
-rapids-logger "Installing PyTorch and PyG dependencies"
-rapids-retry python -m pip install torch==2.1.0 --index-url ${PYTORCH_URL}
-rapids-retry python -m pip install "torch-geometric>=2.5,<2.6"
-rapids-retry python -m pip install \
-  ogb \
-  pyg_lib \
-  torch_scatter \
-  torch_sparse \
-  tensordict \
-  -f ${PYG_URL}
+
+# echo to expand wildcard before adding `[extra]` requires for pip
+python -m pip install \
+    -v \
+    --extra-index-url "${PYTORCH_URL}" \
+    --find-links "${PYG_URL}" \
+    "$(echo ./local-deps/pylibwholegraph_${RAPIDS_PY_CUDA_SUFFIX}*.whl)" \
+    "$(echo ./dist/cugraph_pyg_${RAPIDS_PY_CUDA_SUFFIX}*.whl)[test]" \
+    'ogb' \
+    'pyg_lib' \
+    'torch>=2.3.0,<2.4' \
+    'torch-geometric>=2.5,<2.6' \
+    'torch_scatter' \
+    'torch_sparse'
+
+# RAPIDS_DATASET_ROOT_DIR is used by test scripts
+export RAPIDS_DATASET_ROOT_DIR="$(realpath datasets)"
+
+# Used to skip certain examples in CI due to memory limitations
+export CI_RUN=1
 
 rapids-logger "pytest cugraph-pyg (single GPU)"
 pushd python/cugraph-pyg/cugraph_pyg
diff --git a/ci/utils/nbtest.sh b/ci/utils/nbtest.sh
index 91af633..faf7d28 100755
--- a/ci/utils/nbtest.sh
+++ b/ci/utils/nbtest.sh
@@ -60,7 +60,7 @@ for nb in $*; do
     echo --------------------------------------------------------------------------------
     echo STARTING: ${NBNAME}
     echo --------------------------------------------------------------------------------
-    jupyter nbconvert --to script ${NBFILENAME} --output ${NBTMPDIR}/${NBNAME}-test
+    jupyter nbconvert --to python ${NBFILENAME} --output ${NBTMPDIR}/${NBNAME}-test
     echo "${MAGIC_OVERRIDE_CODE}" > ${NBTMPDIR}/tmpfile
     cat ${NBTESTSCRIPT} >> ${NBTMPDIR}/tmpfile
     mv ${NBTMPDIR}/tmpfile ${NBTESTSCRIPT}
diff --git a/conda/environments/all_cuda-118_arch-x86_64.yaml b/conda/environments/all_cuda-118_arch-x86_64.yaml
index 6383b57..56438bf 100644
--- a/conda/environments/all_cuda-118_arch-x86_64.yaml
+++ b/conda/environments/all_cuda-118_arch-x86_64.yaml
@@ -9,19 +9,25 @@ channels:
 - nvidia
 dependencies:
 - breathe
+- c-compiler
 - cmake>=3.26.4,!=3.30.0
 - cuda-nvtx
 - cuda-version=11.8
 - cudatoolkit
 - cudf==24.12.*,>=0.0.0a0
+- cudnn=8.8
 - cugraph==24.12.*,>=0.0.0a0
 - cupy>=12.0.0
+- cxx-compiler
 - cython>=3.0.0
 - dask-cudf==24.12.*,>=0.0.0a0
-- dglteam/label/th21_cu118::dgl
 - doxygen
 - graphviz
 - ipython
+- libraft-headers==24.12.*,>=0.0.0a0
+- librmm==24.12.*,>=0.0.0a0
+- nanobind>=0.2.0
+- nbconvert>=7.16
 - nbsphinx
 - nccl>=2.19
 - ninja
@@ -33,7 +39,6 @@ dependencies:
 - pre-commit
 - pydantic
 - pydata-sphinx-theme
-- pyg::pyg
 - pylibcugraphops==24.12.*,>=0.0.0a0
 - pylibraft==24.12.*,>=0.0.0a0
 - pytest
@@ -41,13 +46,13 @@ dependencies:
 - pytest-cov
 - pytest-forked
 - pytest-xdist
-- pytorch-cuda=11.8
-- pytorch::pytorch>=2.0,<2.4.0a0
+- pytorch::pytorch>=2.3,<2.4.0a0
 - raft-dask==24.12.*,>=0.0.0a0
 - recommonmark
 - rmm==24.12.*,>=0.0.0a0
 - scikit-build-core>=0.10.0
 - scipy
+- setuptools
 - sphinx-copybutton
 - sphinx-markdown-tables
 - sphinx<6
diff --git a/conda/environments/all_cuda-121_arch-x86_64.yaml b/conda/environments/all_cuda-121_arch-x86_64.yaml
index 7429b0d..95aba20 100644
--- a/conda/environments/all_cuda-121_arch-x86_64.yaml
+++ b/conda/environments/all_cuda-121_arch-x86_64.yaml
@@ -9,17 +9,19 @@ channels:
 - nvidia
 dependencies:
 - breathe
+- c-compiler
 - cmake>=3.26.4,!=3.30.0
 - cuda-cudart-dev
 - cuda-nvtx-dev
 - cuda-profiler-api
 - cuda-version=12.1
 - cudf==24.12.*,>=0.0.0a0
+- cudnn=8.8
 - cugraph==24.12.*,>=0.0.0a0
 - cupy>=12.0.0
+- cxx-compiler
 - cython>=3.0.0
 - dask-cudf==24.12.*,>=0.0.0a0
-- dglteam/label/th21_cu121::dgl
 - doxygen
 - graphviz
 - ipython
@@ -27,6 +29,10 @@ dependencies:
 - libcurand-dev
 - libcusolver-dev
 - libcusparse-dev
+- libraft-headers==24.12.*,>=0.0.0a0
+- librmm==24.12.*,>=0.0.0a0
+- nanobind>=0.2.0
+- nbconvert>=7.16
 - nbsphinx
 - nccl>=2.19
 - ninja
@@ -38,7 +44,6 @@ dependencies:
 - pre-commit
 - pydantic
 - pydata-sphinx-theme
-- pyg::pyg
 - pylibcugraphops==24.12.*,>=0.0.0a0
 - pylibraft==24.12.*,>=0.0.0a0
 - pytest
@@ -46,13 +51,13 @@ dependencies:
 - pytest-cov
 - pytest-forked
 - pytest-xdist
-- pytorch-cuda=12.1
-- pytorch::pytorch>=2.0,<2.4.0a0
+- pytorch::pytorch>=2.3,<2.4.0a0
 - raft-dask==24.12.*,>=0.0.0a0
 - recommonmark
 - rmm==24.12.*,>=0.0.0a0
 - scikit-build-core>=0.10.0
 - scipy
+- setuptools
 - sphinx-copybutton
 - sphinx-markdown-tables
 - sphinx<6
diff --git a/conda/environments/all_cuda-124_arch-x86_64.yaml b/conda/environments/all_cuda-124_arch-x86_64.yaml
index 08c0ee6..214688f 100644
--- a/conda/environments/all_cuda-124_arch-x86_64.yaml
+++ b/conda/environments/all_cuda-124_arch-x86_64.yaml
@@ -9,17 +9,19 @@ channels:
 - nvidia
 dependencies:
 - breathe
+- c-compiler
 - cmake>=3.26.4,!=3.30.0
 - cuda-cudart-dev
 - cuda-nvtx-dev
 - cuda-profiler-api
 - cuda-version=12.4
 - cudf==24.12.*,>=0.0.0a0
+- cudnn=8.8
 - cugraph==24.12.*,>=0.0.0a0
 - cupy>=12.0.0
+- cxx-compiler
 - cython>=3.0.0
 - dask-cudf==24.12.*,>=0.0.0a0
-- dglteam/label/th21_cu121::dgl
 - doxygen
 - graphviz
 - ipython
@@ -27,6 +29,10 @@ dependencies:
 - libcurand-dev
 - libcusolver-dev
 - libcusparse-dev
+- libraft-headers==24.12.*,>=0.0.0a0
+- librmm==24.12.*,>=0.0.0a0
+- nanobind>=0.2.0
+- nbconvert>=7.16
 - nbsphinx
 - nccl>=2.19
 - ninja
@@ -38,7 +44,6 @@ dependencies:
 - pre-commit
 - pydantic
 - pydata-sphinx-theme
-- pyg::pyg
 - pylibcugraphops==24.12.*,>=0.0.0a0
 - pylibraft==24.12.*,>=0.0.0a0
 - pytest
@@ -46,13 +51,13 @@ dependencies:
 - pytest-cov
 - pytest-forked
 - pytest-xdist
-- pytorch-cuda=12.1
-- pytorch::pytorch>=2.0,<2.4.0a0
+- pytorch::pytorch>=2.3,<2.4.0a0
 - raft-dask==24.12.*,>=0.0.0a0
 - recommonmark
 - rmm==24.12.*,>=0.0.0a0
 - scikit-build-core>=0.10.0
 - scipy
+- setuptools
 - sphinx-copybutton
 - sphinx-markdown-tables
 - sphinx<6
diff --git a/conda/recipes/cugraph-dgl/meta.yaml b/conda/recipes/cugraph-dgl/meta.yaml
index 3c90075..f08c831 100644
--- a/conda/recipes/cugraph-dgl/meta.yaml
+++ b/conda/recipes/cugraph-dgl/meta.yaml
@@ -21,15 +21,17 @@ build:
 requirements:
   host:
     - python
+    - rapids-build-backend>=0.3.1,<0.4.0.dev0
+    - setuptools>=61.0.0
   run:
     - cugraph ={{ version }}
-    - dgl >=1.1.0.cu*
+    - dgl >=2.4.0.th23.cu*
     - numba >=0.57
     - numpy >=1.23,<3.0a0
     - pylibcugraphops ={{ minor_version }}
     - tensordict >=0.1.2
     - python
-    - pytorch >=2.0
+    - pytorch >=2.3,<2.4.0a0
     - cupy >=12.0.0
 
 tests:
@@ -41,4 +43,4 @@ about:
   dev_url: https://github.com/rapidsai/cugraph
   license: Apache-2.0
   license_file: ../../../LICENSE
-  summary: cuGraph library
+  summary: cuGraph-dgl library
diff --git a/conda/recipes/cugraph-pyg/conda_build_config.yaml b/conda/recipes/cugraph-pyg/conda_build_config.yaml
index 2d7aff1..47d98b4 100644
--- a/conda/recipes/cugraph-pyg/conda_build_config.yaml
+++ b/conda/recipes/cugraph-pyg/conda_build_config.yaml
@@ -10,7 +10,7 @@ cuda_compiler:
   - nvcc
 
 cmake_version:
-  - ">=3.26.4"
+  - ">=3.26.4,!=3.30.0"
 
 c_stdlib:
   - sysroot
diff --git a/conda/recipes/cugraph-pyg/meta.yaml b/conda/recipes/cugraph-pyg/meta.yaml
index 5bce780..7d3e503 100644
--- a/conda/recipes/cugraph-pyg/meta.yaml
+++ b/conda/recipes/cugraph-pyg/meta.yaml
@@ -24,18 +24,19 @@ requirements:
   host:
     - cython >=3.0.0
     - python
-    - scikit-build-core >=0.7.0
+    - rapids-build-backend>=0.3.1,<0.4.0.dev0
+    - setuptools>=61.0.0
   run:
     - rapids-dask-dependency ={{ minor_version }}
     - numba >=0.57
     - numpy >=1.23,<3.0a0
     - python
-    - pytorch >=2.0
+    - pytorch >=2.3,<2.4.0a0
     - cupy >=12.0.0
     - cugraph ={{ version }}
     - pylibcugraphops ={{ minor_version }}
     - tensordict >=0.1.2
-    - pyg >=2.5,<2.6
+    - pytorch_geometric >=2.5,<2.6
 
 tests:
   imports:
diff --git a/datasets/get_test_data.sh b/datasets/get_test_data.sh
new file mode 100755
index 0000000..6778166
--- /dev/null
+++ b/datasets/get_test_data.sh
@@ -0,0 +1,128 @@
+# Copyright (c) 2021-2024, NVIDIA CORPORATION.
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+#!/bin/bash
+set -e
+set -o pipefail
+
+# Ensure we're in the cugraph/datasets dir
+cd "$( cd "$( dirname "$(realpath -m "${BASH_SOURCE[0]}")" )" && pwd )";
+
+# Update this to add/remove/change a dataset, using the following format:
+#
+#  comment about the dataset
+#  dataset download URL
+#  destination dir to untar to
+#  blank line separator
+#
+# FIXME: some test data needs to be extracted to "benchmarks", which is
+# confusing now that there's dedicated datasets for benchmarks.
+CPP_CI_DATASET_DATA="
+# ~10s download
+https://data.rapids.ai/cugraph/test/cpp_ci_datasets.tgz
+test
+"
+
+BASE_DATASET_DATA="
+# ~22s download
+https://data.rapids.ai/cugraph/test/datasets.tgz
+test
+
+# ~14s download
+https://data.rapids.ai/cugraph/test/ref/pagerank.tgz
+test/ref
+
+# ~1s download
+https://data.rapids.ai/cugraph/test/ref/sssp.tgz
+test/ref
+
+# ~15s download
+https://data.rapids.ai/cugraph/benchmark/hibench/hibench_1_large.tgz
+benchmark
+
+# ~1s download
+https://data.rapids.ai/cugraph/benchmark/hibench/hibench_1_small.tgz
+benchmark
+
+# ~0.6s download
+https://data.rapids.ai/cugraph/test/tsplib/datasets.tar.gz
+tsplib
+"
+
+EXTENDED_DATASET_DATA="
+# ~42s download - tests using this dataset are currently not run in test.sh with --quick
+https://data.rapids.ai/cugraph/benchmark/hibench/hibench_1_huge.tgz
+benchmark
+"
+
+BENCHMARK_DATASET_DATA="
+# ~90s download - these are used for benchmarks runs (code in <cugraph root>/benchmarks)
+https://data.rapids.ai/cugraph/benchmark/benchmark_csv_data.tgz
+csv
+"
+
+SELF_LOOPS_DATASET_DATA="
+# ~1s download
+https://data.rapids.ai/cugraph/benchmark/benchmark_csv_data_self_loops.tgz
+self_loops
+"
+################################################################################
+# Do not change the script below this line if only adding/updating a dataset
+
+NUMARGS=$#
+ARGS=$*
+function hasArg {
+    (( ${NUMARGS} != 0 )) && (echo " ${ARGS} " | grep -q " $1 ")
+}
+
+if hasArg -h || hasArg --help; then
+    echo "$0 [--subset | --benchmark | --self_loops]"
+    exit 0
+fi
+
+# Select the datasets to install
+if hasArg "--benchmark"; then
+    DATASET_DATA="${BENCHMARK_DATASET_DATA}"
+elif hasArg "--subset"; then
+    DATASET_DATA="${BASE_DATASET_DATA}"
+elif hasArg "--cpp_ci_subset"; then
+    DATASET_DATA="${CPP_CI_DATASET_DATA}"
+elif hasArg "--self_loops"; then
+    DATASET_DATA="${SELF_LOOPS_DATASET_DATA}"
+# Do not include benchmark datasets by default - too big
+else
+    DATASET_DATA="${BASE_DATASET_DATA} ${EXTENDED_DATASET_DATA}"
+fi
+
+URLS=($(echo "$DATASET_DATA"|awk '{if (NR%4 == 3) print $0}'))  # extract 3rd fields to a bash array
+DESTDIRS=($(echo "$DATASET_DATA"|awk '{if (NR%4 == 0) print $0}'))  # extract 4th fields to a bash array
+
+echo Downloading ...
+
+# Download all tarfiles to a tmp dir
+mkdir -p tmp
+cd tmp
+for url in ${URLS[*]}; do
+   time wget -N --progress=dot:giga ${url}
+done
+cd ..
+
+# create the destination dirs
+mkdir -p "${DESTDIRS[@]}"
+
+# Iterate over the arrays and untar the nth tarfile to the nth dest directory.
+# The tarfile name is derived from the download url.
+echo Decompressing ...
+for index in ${!DESTDIRS[*]}; do
+    echo "tmp/$(basename "${URLS[$index]}") -C ${DESTDIRS[$index]}" | tr '\n' '\0'
+done | xargs -0 -t -r -n1 -P$(nproc --all) sh -c 'tar -xzvf $0 --overwrite'
diff --git a/datasets/karate.csv b/datasets/karate.csv
new file mode 100644
index 0000000..4ed9f43
--- /dev/null
+++ b/datasets/karate.csv
@@ -0,0 +1,156 @@
+1 0 1.0
+2 0 1.0
+3 0 1.0
+4 0 1.0
+5 0 1.0
+6 0 1.0
+7 0 1.0
+8 0 1.0
+10 0 1.0
+11 0 1.0
+12 0 1.0
+13 0 1.0
+17 0 1.0
+19 0 1.0
+21 0 1.0
+31 0 1.0
+2 1 1.0
+3 1 1.0
+7 1 1.0
+13 1 1.0
+17 1 1.0
+19 1 1.0
+21 1 1.0
+30 1 1.0
+3 2 1.0
+7 2 1.0
+8 2 1.0
+9 2 1.0
+13 2 1.0
+27 2 1.0
+28 2 1.0
+32 2 1.0
+7 3 1.0
+12 3 1.0
+13 3 1.0
+6 4 1.0
+10 4 1.0
+6 5 1.0
+10 5 1.0
+16 5 1.0
+16 6 1.0
+30 8 1.0
+32 8 1.0
+33 8 1.0
+33 9 1.0
+33 13 1.0
+32 14 1.0
+33 14 1.0
+32 15 1.0
+33 15 1.0
+32 18 1.0
+33 18 1.0
+33 19 1.0
+32 20 1.0
+33 20 1.0
+32 22 1.0
+33 22 1.0
+25 23 1.0
+27 23 1.0
+29 23 1.0
+32 23 1.0
+33 23 1.0
+25 24 1.0
+27 24 1.0
+31 24 1.0
+31 25 1.0
+29 26 1.0
+33 26 1.0
+33 27 1.0
+31 28 1.0
+33 28 1.0
+32 29 1.0
+33 29 1.0
+32 30 1.0
+33 30 1.0
+32 31 1.0
+33 31 1.0
+33 32 1.0
+0 1 1.0
+0 2 1.0
+0 3 1.0
+0 4 1.0
+0 5 1.0
+0 6 1.0
+0 7 1.0
+0 8 1.0
+0 10 1.0
+0 11 1.0
+0 12 1.0
+0 13 1.0
+0 17 1.0
+0 19 1.0
+0 21 1.0
+0 31 1.0
+1 2 1.0
+1 3 1.0
+1 7 1.0
+1 13 1.0
+1 17 1.0
+1 19 1.0
+1 21 1.0
+1 30 1.0
+2 3 1.0
+2 7 1.0
+2 8 1.0
+2 9 1.0
+2 13 1.0
+2 27 1.0
+2 28 1.0
+2 32 1.0
+3 7 1.0
+3 12 1.0
+3 13 1.0
+4 6 1.0
+4 10 1.0
+5 6 1.0
+5 10 1.0
+5 16 1.0
+6 16 1.0
+8 30 1.0
+8 32 1.0
+8 33 1.0
+9 33 1.0
+13 33 1.0
+14 32 1.0
+14 33 1.0
+15 32 1.0
+15 33 1.0
+18 32 1.0
+18 33 1.0
+19 33 1.0
+20 32 1.0
+20 33 1.0
+22 32 1.0
+22 33 1.0
+23 25 1.0
+23 27 1.0
+23 29 1.0
+23 32 1.0
+23 33 1.0
+24 25 1.0
+24 27 1.0
+24 31 1.0
+25 31 1.0
+26 29 1.0
+26 33 1.0
+27 33 1.0
+28 31 1.0
+28 33 1.0
+29 32 1.0
+29 33 1.0
+30 32 1.0
+30 33 1.0
+31 32 1.0
+31 33 1.0
+32 33 1.0
diff --git a/dependencies.yaml b/dependencies.yaml
index e0e74d2..021e749 100644
--- a/dependencies.yaml
+++ b/dependencies.yaml
@@ -24,8 +24,6 @@ files:
       - depends_on_pylibcugraphops
       - depends_on_cupy
       - depends_on_pytorch
-      - depends_on_dgl
-      - depends_on_pyg
       - python_run_cugraph_dgl
       - python_run_cugraph_pyg
       - test_cpp
@@ -56,6 +54,7 @@ files:
     output: none
     includes:
       - cuda_version
+      - depends_on_pytorch
       - py_version
       - test_notebook
       - test_python_common
@@ -107,6 +106,7 @@ files:
     extras:
       table: build-system
     includes:
+      - python_build_rapids
       - python_build_wheel
   py_run_cugraph_dgl:
     output: pyproject
@@ -115,6 +115,7 @@ files:
       table: project
     includes:
       - python_run_cugraph_dgl
+      - depends_on_cugraph
       - depends_on_pylibcugraphops
   py_test_cugraph_dgl:
     output: pyproject
@@ -126,12 +127,14 @@ files:
       - test_python_common
       - test_python_cugraph_dgl
       - depends_on_pylibwholegraph
+      - depends_on_pytorch
   py_build_cugraph_pyg:
     output: pyproject
     pyproject_dir: python/cugraph-pyg
     extras:
       table: build-system
     includes:
+      - python_build_rapids
       - python_build_wheel
   py_run_cugraph_pyg:
     output: pyproject
@@ -140,6 +143,7 @@ files:
       table: project
     includes:
       - python_run_cugraph_pyg
+      - depends_on_cugraph
       - depends_on_pylibcugraphops
   py_test_cugraph_pyg:
     output: pyproject
@@ -151,6 +155,7 @@ files:
       - test_python_common
       - test_python_cugraph_pyg
       - depends_on_pylibwholegraph
+      - depends_on_pytorch
 
 
   cugraph_dgl_dev:
@@ -254,10 +259,20 @@ dependencies:
               - cuda-nvtx
   common_build:
     common:
-      - output_types: [conda, pyproject]
+      - output_types: [requirements, pyproject]
         packages:
-          - &cmake_ver cmake>=3.26.4,!=3.30.0
           - ninja
+      - output_types: [conda]
+        packages:
+          - &cmake_ver cmake>=3.26.4,!=3.30.0
+          - cudnn=8.8
+          - c-compiler
+          - cxx-compiler
+          - libraft-headers==24.12.*,>=0.0.0a0
+          - librmm==24.12.*,>=0.0.0a0
+          - nanobind>=0.2.0
+          - &cython cython>=3.0.0
+          - &nccl nccl>=2.19
 
   docs:
     common:
@@ -294,6 +309,11 @@ dependencies:
           - matrix:
             packages:
               - python>=3.10,<3.13
+  python_build_rapids:
+    common:
+      - output_types: [conda, pyproject, requirements]
+        packages:
+          - rapids-build-backend>=0.3.1,<0.4.0.dev0
   python_build_wheel:
     common:
       - output_types: [conda, pyproject, requirements]
@@ -301,11 +321,12 @@ dependencies:
           - cmake>=3.26.4,!=3.30.0
           - cython>=3.0.0
           - ninja
+          - setuptools
   python_build_cythonize:
     common:
       - output_types: [conda, pyproject, requirements]
         packages:
-          - cython>=3.0.0
+          - *cython
       - output_types: conda
         packages:
           - scikit-build-core>=0.10.0
@@ -323,18 +344,12 @@ dependencies:
         packages:
           - &numba numba>=0.57
           - *numpy
-      - output_types: [pyproject]
-        packages:
-          - &cugraph cugraph==24.12.*,>=0.0.0a0
   python_run_cugraph_pyg:
     common:
       - output_types: [conda, pyproject]
         packages:
           - *numba
           - *numpy
-      - output_types: [pyproject]
-        packages:
-          - *cugraph
   rapids_build_skbuild:
     common:
       - output_types: [conda, requirements, pyproject]
@@ -357,10 +372,17 @@ dependencies:
       - output_types: [conda, requirements]
         packages:
           - ipython
+          - nbconvert>=7.16
           - notebook>=0.5.0
       - output_types: [conda]
         packages:
           - wget
+  test_cpp:
+    common:
+      - output_types: conda
+        packages:
+          - *cmake_ver
+          - *nccl
   test_python_common:
     common:
       - output_types: [conda, pyproject]
@@ -368,6 +390,7 @@ dependencies:
           - pytest
           - pytest-benchmark
           - pytest-cov
+          - pytest-forked
           - pytest-xdist
   test_python_cugraph_dgl:
     common:
@@ -389,24 +412,22 @@ dependencies:
     common:
       - output_types: [conda]
         packages:
-          - cugraph==24.12.*,>=0.0.0a0
-          - pytorch::pytorch>=2.0
-          - dgl>=1.1.0.cu*
+          - &pytorch_conda pytorch::pytorch>=2.3,<2.4.0a0
+          - dgl>=2.4.0.cu*
           - &tensordict tensordict>=0.1.2
   cugraph_pyg_dev:
     common:
       - output_types: [conda]
         packages:
-          - cugraph==24.12.*,>=0.0.0a0
-          - pytorch::pytorch>=2.0
+          - *pytorch_conda
           - *tensordict
-          - pyg>=2.5,<2.6
+          - pytorch_geometric>=2.5,<2.6
 
   depends_on_pytorch:
     common:
       - output_types: [conda]
         packages:
-          - pytorch::pytorch>=2.0,<2.4.0a0
+          - *pytorch_conda
           - torchdata
           - pydantic
     specific:
@@ -423,70 +444,13 @@ dependencies:
         matrices:
           - matrix: {cuda: "12.*"}
             packages:
-              - &pytorch_pip torch>=2.0,<2.4.0a0
+              - &pytorch_pip torch>=2.3,<2.4.0a0
               - *tensordict
           - matrix: {cuda: "11.*"}
             packages:
               - *pytorch_pip
               - *tensordict
           - {matrix: null, packages: [*pytorch_pip, *tensordict]}
-      - output_types: [conda]
-        matrices:
-          - matrix: {cuda: "12.*"}
-            packages:
-              - pytorch-cuda=12.1
-          - matrix: {cuda: "11.*"}
-            packages:
-              - pytorch-cuda=11.8
-
-  depends_on_dgl:
-    specific:
-      - output_types: [conda]
-        matrices:
-          - matrix: {cuda: "12.*"}
-            packages:
-              - "dglteam/label/th21_cu121::dgl"
-          - matrix: {cuda: "11.*"}
-            packages:
-              - "dglteam/label/th21_cu118::dgl"
-          - {matrix: null, packages: ["dglteam/label/th23_cu121::dgl"]}
-
-  depends_on_pyg:
-    specific:
-      - output_types: [conda]
-        matrices:
-          - matrix: {cuda: "12.*"}
-            packages:
-              - "pyg::pyg"
-          - matrix: {cuda: "11.*"}
-            packages:
-              - "pyg::pyg"
-          - {matrix: null, packages: ["pyg::pyg"]}
-
-  depends_on_pylibwholegraph:
-    common:
-      - output_types: conda
-        packages:
-          - &pylibwholegraph_unsuffixed pylibwholegraph==24.12.*,>=0.0.0a0
-      - output_types: requirements
-        packages:
-          # pip recognizes the index as a global option for the requirements.txt file
-          - --extra-index-url=https://pypi.nvidia.com
-          - --extra-index-url=https://pypi.anaconda.org/rapidsai-wheels-nightly/simple
-    specific:
-      - output_types: [requirements, pyproject]
-        matrices:
-          - matrix:
-              cuda: "12.*"
-              cuda_suffixed: "true"
-            packages:
-              - pylibwholegraph-cu12==24.12.*,>=0.0.0a0
-          - matrix:
-              cuda: "11.*"
-              cuda_suffixed: "true"
-            packages:
-              - pylibwholegraph-cu11==24.12.*,>=0.0.0a0
-          - {matrix: null, packages: [*pylibwholegraph_unsuffixed]}
 
   depends_on_rmm:
     common:
@@ -706,3 +670,55 @@ dependencies:
             packages: &cupy_packages_cu11
               - cupy-cuda11x>=12.0.0
           - {matrix: null, packages: *cupy_packages_cu11}
+
+  depends_on_dgl:
+    specific:
+      - output_types: [conda]
+        matrices:
+          - matrix: {cuda: "12.1"}
+            packages:
+              - "dglteam/label/th23_cu121::dgl"
+          - matrix: {cuda: "12.4"}
+            packages:
+              - "dglteam/label/th23_cu124::dgl"
+          - matrix: {cuda: "11.*"}
+            packages:
+              - "dglteam/label/th23_cu118::dgl"
+          - {matrix: null, packages: ["dglteam/label/th23_cu121::dgl"]}
+
+  depends_on_pyg:
+    specific:
+      - output_types: [conda]
+        matrices:
+          - matrix: {cuda: "12.*"}
+            packages:
+              - "pyg::pyg"
+          - matrix: {cuda: "11.*"}
+            packages:
+              - "pyg::pyg"
+          - {matrix: null, packages: ["pyg::pyg"]}
+
+  depends_on_pylibwholegraph:
+    common:
+      - output_types: conda
+        packages:
+          - &pylibwholegraph_unsuffixed pylibwholegraph==24.12.*,>=0.0.0a0
+      - output_types: requirements
+        packages:
+          # pip recognizes the index as a global option for the requirements.txt file
+          - --extra-index-url=https://pypi.nvidia.com
+          - --extra-index-url=https://pypi.anaconda.org/rapidsai-wheels-nightly/simple
+    specific:
+      - output_types: [requirements, pyproject]
+        matrices:
+          - matrix:
+              cuda: "12.*"
+              cuda_suffixed: "true"
+            packages:
+              - pylibwholegraph-cu12==24.12.*,>=0.0.0a0
+          - matrix:
+              cuda: "11.*"
+              cuda_suffixed: "true"
+            packages:
+              - pylibwholegraph-cu11==24.12.*,>=0.0.0a0
+          - {matrix: null, packages: [*pylibwholegraph_unsuffixed]}
diff --git a/python/cugraph-dgl/conda/cugraph_dgl_dev_cuda-118.yaml b/python/cugraph-dgl/conda/cugraph_dgl_dev_cuda-118.yaml
index 9f0070f..632157f 100644
--- a/python/cugraph-dgl/conda/cugraph_dgl_dev_cuda-118.yaml
+++ b/python/cugraph-dgl/conda/cugraph_dgl_dev_cuda-118.yaml
@@ -9,8 +9,8 @@ channels:
 - nvidia
 dependencies:
 - cugraph==24.12.*,>=0.0.0a0
-- dgl>=1.1.0.cu*
-- dglteam/label/th21_cu118::dgl
+- dgl>=2.4.0.cu*
+- dglteam/label/th23_cu118::dgl
 - pandas
 - pre-commit
 - pydantic
@@ -18,10 +18,9 @@ dependencies:
 - pytest
 - pytest-benchmark
 - pytest-cov
+- pytest-forked
 - pytest-xdist
-- pytorch-cuda=11.8
-- pytorch::pytorch>=2.0
-- pytorch::pytorch>=2.0,<2.4.0a0
+- pytorch::pytorch>=2.3,<2.4.0a0
 - tensordict>=0.1.2
 - torchdata
 name: cugraph_dgl_dev_cuda-118
diff --git a/python/cugraph-dgl/cugraph_dgl/_version.py b/python/cugraph-dgl/cugraph_dgl/_version.py
index f2210fd..e8adcc3 100644
--- a/python/cugraph-dgl/cugraph_dgl/_version.py
+++ b/python/cugraph-dgl/cugraph_dgl/_version.py
@@ -15,12 +15,17 @@
 
 import importlib.resources
 
-# Read VERSION file from the module that is symlinked to VERSION file
-# in the root of the repo at build time or copied to the moudle at
-# installation. VERSION is a separate file that allows CI build-time scripts
-# to update version info (including commit hashes) without modifying
-# source files.
 __version__ = (
-    importlib.resources.files("cugraph_dgl").joinpath("VERSION").read_text().strip()
+    importlib.resources.files(__package__).joinpath("VERSION").read_text().strip()
 )
-__git_commit__ = ""
+try:
+    __git_commit__ = (
+        importlib.resources.files(__package__)
+        .joinpath("GIT_COMMIT")
+        .read_text()
+        .strip()
+    )
+except FileNotFoundError:
+    __git_commit__ = ""
+
+__all__ = ["__git_commit__", "__version__"]
diff --git a/python/cugraph-dgl/cugraph_dgl/dataloading/neighbor_sampler.py b/python/cugraph-dgl/cugraph_dgl/dataloading/neighbor_sampler.py
index 4ec513c..ecc5100 100644
--- a/python/cugraph-dgl/cugraph_dgl/dataloading/neighbor_sampler.py
+++ b/python/cugraph-dgl/cugraph_dgl/dataloading/neighbor_sampler.py
@@ -197,10 +197,8 @@ def sample(
 
         if g.is_homogeneous:
             indices = torch.concat(list(indices))
-            ds.sample_from_nodes(indices.long(), batch_size=batch_size)
-            return HomogeneousSampleReader(
-                ds.get_reader(), self.output_format, self.edge_dir
-            )
+            reader = ds.sample_from_nodes(indices.long(), batch_size=batch_size)
+            return HomogeneousSampleReader(reader, self.output_format, self.edge_dir)
 
         raise ValueError(
             "Sampling heterogeneous graphs is currently"
diff --git a/python/cugraph-dgl/cugraph_dgl/tests/test_graph.py b/python/cugraph-dgl/cugraph_dgl/tests/test_graph.py
index a60db97..2b9f46d 100644
--- a/python/cugraph-dgl/cugraph_dgl/tests/test_graph.py
+++ b/python/cugraph-dgl/cugraph_dgl/tests/test_graph.py
@@ -53,10 +53,11 @@ def test_graph_make_homogeneous_graph(direction):
         graph.nodes() == torch.arange(num_nodes, dtype=torch.int64, device="cuda")
     ).all()
 
-    assert graph.nodes[None]["x"] is not None
-    assert (graph.nodes[None]["x"] == torch.as_tensor(node_x, device="cuda")).all()
+    emb = graph.nodes[None]["x"]
+    assert emb is not None
+    assert (emb() == torch.as_tensor(node_x, device="cuda")).all()
     assert (
-        graph.nodes[None]["num"]
+        graph.nodes[None]["num"]()
         == torch.arange(num_nodes, dtype=torch.int64, device="cuda")
     ).all()
 
@@ -64,7 +65,7 @@ def test_graph_make_homogeneous_graph(direction):
         graph.edges("eid", device="cuda")
         == torch.arange(len(df), dtype=torch.int64, device="cuda")
     ).all()
-    assert (graph.edges[None]["weight"] == torch.as_tensor(wgt, device="cuda")).all()
+    assert (graph.edges[None]["weight"]() == torch.as_tensor(wgt, device="cuda")).all()
 
     plc_expected_graph = pylibcugraph.SGGraph(
         pylibcugraph.ResourceHandle(),
diff --git a/python/cugraph-dgl/cugraph_dgl/tests/test_version.py b/python/cugraph-dgl/cugraph_dgl/tests/test_version.py
new file mode 100644
index 0000000..343e4fb
--- /dev/null
+++ b/python/cugraph-dgl/cugraph_dgl/tests/test_version.py
@@ -0,0 +1,12 @@
+# Copyright (c) 2024, NVIDIA CORPORATION.
+
+import cugraph_dgl
+
+
+def test_version_constants_are_populated():
+    # __git_commit__ will only be non-empty in a built distribution
+    assert isinstance(cugraph_dgl.__git_commit__, str)
+
+    # __version__ should always be non-empty
+    assert isinstance(cugraph_dgl.__version__, str)
+    assert len(cugraph_dgl.__version__) > 0
diff --git a/python/cugraph-dgl/cugraph_dgl/view.py b/python/cugraph-dgl/cugraph_dgl/view.py
index dbc53e7..7c4d95f 100644
--- a/python/cugraph-dgl/cugraph_dgl/view.py
+++ b/python/cugraph-dgl/cugraph_dgl/view.py
@@ -11,6 +11,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
+import warnings
 
 from collections import defaultdict
 from collections.abc import MutableMapping
@@ -20,11 +21,53 @@
 
 import cugraph_dgl
 from cugraph_dgl.typing import TensorType
+from cugraph_dgl.utils.cugraph_conversion_utils import _cast_to_torch_tensor
 
 torch = import_optional("torch")
 dgl = import_optional("dgl")
 
 
+class EmbeddingView:
+    def __init__(self, storage: "dgl.storages.base.FeatureStorage", ld: int):
+        self.__ld = ld
+        self.__storage = storage
+
+    def __getitem__(self, u: TensorType) -> "torch.Tensor":
+        u = _cast_to_torch_tensor(u)
+        try:
+            return self.__storage.fetch(
+                u,
+                "cuda",
+            )
+        except RuntimeError as ex:
+            warnings.warn(
+                "Got error accessing data, trying again with index on device: "
+                + str(ex)
+            )
+            return self.__storage.fetch(
+                u.cuda(),
+                "cuda",
+            )
+
+    def __call__(self):
+        warnings.warn(
+            "Getting an entire embedding tensor is not recommended "
+            " as it wastes memory.  Consider indexing to get only the "
+            "required elements of the embedding tensor."
+        )
+        return self[torch.arange(self.__ld, dtype=torch.int64)]
+
+    @property
+    def shape(self) -> "torch.Size":
+        try:
+            f = self.__storage.fetch(torch.tensor([0]), "cpu")
+        except RuntimeError:
+            f = self.__storage.fetch(torch.tensor([0], device="cuda"), "cuda")
+        sz = [s for s in f.shape]
+        sz[0] = self.__ld
+        return torch.Size(tuple(sz))
+
+
 class HeteroEdgeDataView(MutableMapping):
     """
     Duck-typed version of DGL's HeteroEdgeDataView.
diff --git a/python/cugraph-dgl/pyproject.toml b/python/cugraph-dgl/pyproject.toml
index 0ca636d..8af2524 100644
--- a/python/cugraph-dgl/pyproject.toml
+++ b/python/cugraph-dgl/pyproject.toml
@@ -6,8 +6,10 @@ requires = [
     "cmake>=3.26.4,!=3.30.0",
     "cython>=3.0.0",
     "ninja",
+    "rapids-build-backend>=0.3.1,<0.4.0.dev0",
+    "setuptools",
 ] # This list was generated by `rapids-dependency-file-generator`. To make changes, edit ../../dependencies.yaml and run `rapids-dependency-file-generator`.
-build-backend = "setuptools.build_meta"
+build-backend = "rapids_build_backend.build"
 
 [project]
 name = "cugraph-dgl"
@@ -37,7 +39,10 @@ test = [
     "pytest",
     "pytest-benchmark",
     "pytest-cov",
+    "pytest-forked",
     "pytest-xdist",
+    "tensordict>=0.1.2",
+    "torch>=2.3,<2.4.0a0",
 ] # This list was generated by `rapids-dependency-file-generator`. To make changes, edit ../../dependencies.yaml and run `rapids-dependency-file-generator`.
 
 [project.urls]
@@ -54,3 +59,8 @@ version = {file = "cugraph_dgl/VERSION"}
 include = [
     "cugraph_dgl*",
 ]
+
+[tool.rapids-build-backend]
+build-backend = "setuptools.build_meta"
+dependencies-file = "../../dependencies.yaml"
+matrix-entry = "cuda_suffixed=true"
diff --git a/python/cugraph-pyg/conda/cugraph_pyg_dev_cuda-118.yaml b/python/cugraph-pyg/conda/cugraph_pyg_dev_cuda-118.yaml
index a9f520c..5609cb5 100644
--- a/python/cugraph-pyg/conda/cugraph_pyg_dev_cuda-118.yaml
+++ b/python/cugraph-pyg/conda/cugraph_pyg_dev_cuda-118.yaml
@@ -12,15 +12,14 @@ dependencies:
 - pandas
 - pre-commit
 - pydantic
-- pyg>=2.5,<2.6
 - pylibcugraphops==24.12.*,>=0.0.0a0
 - pytest
 - pytest-benchmark
 - pytest-cov
+- pytest-forked
 - pytest-xdist
-- pytorch-cuda=11.8
-- pytorch::pytorch>=2.0
-- pytorch::pytorch>=2.0,<2.4.0a0
+- pytorch::pytorch>=2.3,<2.4.0a0
+- pytorch_geometric>=2.5,<2.6
 - tensordict>=0.1.2
 - torchdata
 name: cugraph_pyg_dev_cuda-118
diff --git a/python/cugraph-pyg/cugraph_pyg/tests/test_version.py b/python/cugraph-pyg/cugraph_pyg/tests/test_version.py
new file mode 100644
index 0000000..4ea0f98
--- /dev/null
+++ b/python/cugraph-pyg/cugraph_pyg/tests/test_version.py
@@ -0,0 +1,12 @@
+# Copyright (c) 2024, NVIDIA CORPORATION.
+
+import cugraph_pyg
+
+
+def test_version_constants_are_populated():
+    # __git_commit__ will only be non-empty in a built distribution
+    assert isinstance(cugraph_pyg.__git_commit__, str)
+
+    # __version__ should always be non-empty
+    assert isinstance(cugraph_pyg.__version__, str)
+    assert len(cugraph_pyg.__version__) > 0
diff --git a/python/cugraph-pyg/pyproject.toml b/python/cugraph-pyg/pyproject.toml
index d0bfe58..f096b58 100644
--- a/python/cugraph-pyg/pyproject.toml
+++ b/python/cugraph-pyg/pyproject.toml
@@ -6,7 +6,10 @@ requires = [
     "cmake>=3.26.4,!=3.30.0",
     "cython>=3.0.0",
     "ninja",
+    "rapids-build-backend>=0.3.1,<0.4.0.dev0",
+    "setuptools",
 ] # This list was generated by `rapids-dependency-file-generator`. To make changes, edit ../../dependencies.yaml and run `rapids-dependency-file-generator`.
+build-backend = "rapids_build_backend.build"
 
 [tool.pytest.ini_options]
 testpaths = ["cugraph_pyg/tests"]
@@ -45,7 +48,10 @@ test = [
     "pytest",
     "pytest-benchmark",
     "pytest-cov",
+    "pytest-forked",
     "pytest-xdist",
+    "tensordict>=0.1.2",
+    "torch>=2.3,<2.4.0a0",
 ] # This list was generated by `rapids-dependency-file-generator`. To make changes, edit ../../dependencies.yaml and run `rapids-dependency-file-generator`.
 
 [tool.setuptools]
@@ -59,3 +65,8 @@ include = [
     "cugraph_pyg*",
     "cugraph_pyg.*",
 ]
+
+[tool.rapids-build-backend]
+build-backend = "setuptools.build_meta"
+dependencies-file = "../../dependencies.yaml"
+matrix-entry = "cuda_suffixed=true"
diff --git a/python/pylibwholegraph/pyproject.toml b/python/pylibwholegraph/pyproject.toml
index 3eed9dc..f1bcfb3 100644
--- a/python/pylibwholegraph/pyproject.toml
+++ b/python/pylibwholegraph/pyproject.toml
@@ -58,6 +58,7 @@ requires = [
     "cmake>=3.26.4,!=3.30.0",
     "cython>=3.0.0",
     "ninja",
+    "setuptools",
 ] # This list was generated by `rapids-dependency-file-generator`. To make changes, edit ../../dependencies.yaml and run `rapids-dependency-file-generator`.
 
 [tool.scikit-build]