diff --git a/dockerfiles/pytorch/pytorch-spr-ssd-resnet34-inference.Dockerfile b/dockerfiles/pytorch/pytorch-spr-ssd-resnet34-inference.Dockerfile
new file mode 100644
index 000000000..92af5b531
--- /dev/null
+++ b/dockerfiles/pytorch/pytorch-spr-ssd-resnet34-inference.Dockerfile
@@ -0,0 +1,79 @@
+# Copyright (c) 2020-2021 Intel 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.
+# ============================================================================
+#
+# THIS IS A GENERATED DOCKERFILE.
+#
+# This file was assembled from multiple pieces, whose use is documented
+# throughout. Please refer to the TensorFlow dockerfiles documentation
+# for more information.
+
+ARG PYTORCH_IMAGE="model-zoo"
+ARG PYTORCH_TAG="pytorch-ipex-spr"
+
+FROM ${PYTORCH_IMAGE}:${PYTORCH_TAG} AS intel-optimized-pytorch
+
+# Build Torch Vision
+ARG TORCHVISION_VERSION=v0.8.0
+
+RUN source ~/anaconda3/bin/activate pytorch && \
+    git clone https://github.com/pytorch/vision && \
+    cd vision && \
+    git checkout ${TORCHVISION_VERSION} && \
+    python setup.py install
+
+RUN source ~/anaconda3/bin/activate pytorch && \
+    pip install matplotlib Pillow pycocotools && \
+    pip install yacs opencv-python cityscapesscripts transformers && \
+    conda install -y libopenblas psutil && \
+    cd /workspace/installs && \
+    wget https://github.com/gperftools/gperftools/releases/download/gperftools-2.7.90/gperftools-2.7.90.tar.gz && \
+    tar -xzf gperftools-2.7.90.tar.gz && \
+    cd gperftools-2.7.90 && \
+    ./configure --prefix=$HOME/.local && \
+    make && \
+    make install && \
+    rm -rf /workspace/installs/
+
+ARG PACKAGE_DIR=model_packages
+
+ARG PACKAGE_NAME="pytorch-spr-ssd-resnet34-inference"
+
+ARG MODEL_WORKSPACE
+
+# ${MODEL_WORKSPACE} and below needs to be owned by root:root rather than the current UID:GID
+# this allows the default user (root) to work in k8s single-node, multi-node
+RUN umask 002 && mkdir -p ${MODEL_WORKSPACE} && chgrp root ${MODEL_WORKSPACE} && chmod g+s+w,o+s+r ${MODEL_WORKSPACE}
+
+ADD --chown=0:0 ${PACKAGE_DIR}/${PACKAGE_NAME}.tar.gz ${MODEL_WORKSPACE}
+
+RUN chown -R root ${MODEL_WORKSPACE}/${PACKAGE_NAME} && chgrp -R root ${MODEL_WORKSPACE}/${PACKAGE_NAME} && chmod -R g+s+w ${MODEL_WORKSPACE}/${PACKAGE_NAME} && find ${MODEL_WORKSPACE}/${PACKAGE_NAME} -type d | xargs chmod o+r+x 
+
+WORKDIR ${MODEL_WORKSPACE}/${PACKAGE_NAME}
+
+FROM intel-optimized-pytorch AS release
+COPY --from=intel-optimized-pytorch /root/anaconda3 /root/anaconda3
+COPY --from=intel-optimized-pytorch /workspace/lib/ /workspace/lib/
+COPY --from=intel-optimized-pytorch /root/.local/ /root/.local/
+
+ENV DNNL_MAX_CPU_ISA="AVX512_CORE_AMX"
+
+ENV PATH="~/anaconda3/bin:${PATH}"
+ENV LD_PRELOAD="/workspace/lib/jemalloc/lib/libjemalloc.so:$LD_PRELOAD"
+ENV MALLOC_CONF="oversize_threshold:1,background_thread:true,metadata_thp:auto,dirty_decay_ms:9000000000,muzzy_decay_ms:9000000000"
+ENV BASH_ENV=/root/.bash_profile
+WORKDIR /workspace/
+RUN yum install -y numactl mesa-libGL && \
+    yum clean all && \
+    echo "source activate pytorch" >> /root/.bash_profile
diff --git a/dockerfiles/pytorch/pytorch-spr-ssd-resnet34-training.Dockerfile b/dockerfiles/pytorch/pytorch-spr-ssd-resnet34-training.Dockerfile
new file mode 100644
index 000000000..1f07444c9
--- /dev/null
+++ b/dockerfiles/pytorch/pytorch-spr-ssd-resnet34-training.Dockerfile
@@ -0,0 +1,88 @@
+# Copyright (c) 2020-2021 Intel 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.
+# ============================================================================
+#
+# THIS IS A GENERATED DOCKERFILE.
+#
+# This file was assembled from multiple pieces, whose use is documented
+# throughout. Please refer to the TensorFlow dockerfiles documentation
+# for more information.
+
+ARG PYTORCH_IMAGE="model-zoo"
+ARG PYTORCH_TAG="pytorch-ipex-spr"
+
+FROM ${PYTORCH_IMAGE}:${PYTORCH_TAG} AS intel-optimized-pytorch
+
+RUN source ~/anaconda3/bin/activate pytorch && \
+    pip install matplotlib Pillow pycocotools && \
+    pip install yacs opencv-python cityscapesscripts transformers && \
+    conda install -y libopenblas psutil && \
+    cd /workspace/installs && \
+    wget https://github.com/gperftools/gperftools/releases/download/gperftools-2.7.90/gperftools-2.7.90.tar.gz && \
+    tar -xzf gperftools-2.7.90.tar.gz && \
+    cd gperftools-2.7.90 && \
+    ./configure --prefix=$HOME/.local && \
+    make && \
+    make install && \
+    rm -rf /workspace/installs/
+
+ARG PACKAGE_DIR=model_packages
+
+ARG PACKAGE_NAME="pytorch-spr-ssd-resnet34-training"
+
+ARG MODEL_WORKSPACE
+
+# ${MODEL_WORKSPACE} and below needs to be owned by root:root rather than the current UID:GID
+# this allows the default user (root) to work in k8s single-node, multi-node
+RUN umask 002 && mkdir -p ${MODEL_WORKSPACE} && chgrp root ${MODEL_WORKSPACE} && chmod g+s+w,o+s+r ${MODEL_WORKSPACE}
+
+ADD --chown=0:0 ${PACKAGE_DIR}/${PACKAGE_NAME}.tar.gz ${MODEL_WORKSPACE}
+
+RUN chown -R root ${MODEL_WORKSPACE}/${PACKAGE_NAME} && chgrp -R root ${MODEL_WORKSPACE}/${PACKAGE_NAME} && chmod -R g+s+w ${MODEL_WORKSPACE}/${PACKAGE_NAME} && find ${MODEL_WORKSPACE}/${PACKAGE_NAME} -type d | xargs chmod o+r+x 
+
+WORKDIR ${MODEL_WORKSPACE}/${PACKAGE_NAME}
+
+RUN source ~/anaconda3/bin/activate pytorch && \
+    pip install --upgrade pip && \
+    pip install --no-cache-dir https://github.com/mlperf/logging/archive/9ea0afa.zip && \
+    pip install --no-cache-dir \
+        Cython==0.28.4 \
+        git+http://github.com/NVIDIA/apex.git@9041a868a1a253172d94b113a963375b9badd030#egg=apex \
+        mlperf-compliance==0.0.10 \
+        cycler==0.10.0 \
+        kiwisolver==1.0.1 \
+        matplotlib==2.2.2 \
+        Pillow==5.2.0 \
+        pyparsing==2.2.0 \
+        python-dateutil==2.7.3 \
+        pytz==2018.5 \
+        six==1.11.0 \
+        torchvision==0.2.1 \
+        pycocotools==2.0.2
+
+FROM intel-optimized-pytorch AS release
+COPY --from=intel-optimized-pytorch /root/anaconda3 /root/anaconda3
+COPY --from=intel-optimized-pytorch /workspace/lib/ /workspace/lib/
+COPY --from=intel-optimized-pytorch /root/.local/ /root/.local/
+
+ENV DNNL_MAX_CPU_ISA="AVX512_CORE_AMX"
+
+ENV PATH="~/anaconda3/bin:${PATH}"
+ENV LD_PRELOAD="/workspace/lib/jemalloc/lib/libjemalloc.so:$LD_PRELOAD"
+ENV MALLOC_CONF="oversize_threshold:1,background_thread:true,metadata_thp:auto,dirty_decay_ms:9000000000,muzzy_decay_ms:9000000000"
+ENV BASH_ENV=/root/.bash_profile
+WORKDIR /workspace/
+RUN yum install -y numactl mesa-libGL && \
+    yum clean all && \
+    echo "source activate pytorch" >> /root/.bash_profile
diff --git a/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/.docs/container_build.md b/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/.docs/container_build.md
new file mode 100644
index 000000000..9b4f9b303
--- /dev/null
+++ b/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/.docs/container_build.md
@@ -0,0 +1,26 @@
+## Build the container
+
+The <model name> <mode> package has scripts and a Dockerfile that are
+used to build a workload container that runs the model. This container
+uses the PyTorch/IPEX container as it's base, so ensure that you have built
+the `pytorch-ipex-spr.tar.gz` container prior to building this model container.
+
+Use `docker images` to verify that you have the base container built. For example:
+```
+$ docker images | grep pytorch-ipex-spr
+model-zoo         pytorch-ipex-spr         fecc7096a11e        40 minutes ago      8.31GB
+```
+
+To build the <model name> <mode> container, extract the package and
+run the `build.sh` script.
+```
+# Extract the package
+tar -xzf <package name>
+cd <package dir>
+
+# Build the container
+./build.sh
+```
+
+After the build completes, you should have a container called
+`<docker image>` that will be used to run the model.
diff --git a/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/.docs/datasets.md b/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/.docs/datasets.md
new file mode 100644
index 000000000..78335650c
--- /dev/null
+++ b/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/.docs/datasets.md
@@ -0,0 +1,33 @@
+## Datasets
+
+### COCO
+
+The [COCO dataset](https://cocodataset.org) is used to run <model name>.
+
+Download and extract the 2017 training/validation images and annotations from the
+[COCO dataset website](https://cocodataset.org/#download) to a `coco` folder
+and unzip the files. After extracting the zip files, your dataset directory
+structure should look something like this:
+```
+coco
+├── annotations
+│   ├── captions_train2017.json
+│   ├── captions_val2017.json
+│   ├── instances_train2017.json
+│   ├── instances_val2017.json
+│   ├── person_keypoints_train2017.json
+│   └── person_keypoints_val2017.json
+├── train2017
+│   ├── 000000454854.jpg
+│   ├── 000000137045.jpg
+│   ├── 000000129582.jpg
+│   └── ...
+└── val2017
+    ├── 000000000139.jpg
+    ├── 000000000285.jpg
+    ├── 000000000632.jpg
+    └── ...
+```
+The parent of the `annotations`, `train2017`, and `val2017` directory (in this example `coco`)
+is the directory that should be used when setting the `DATASET_DIR` environment
+variable for <model name> (for example: `export DATASET_DIR=/home/<user>/coco`).
diff --git a/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/.docs/description.md b/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/.docs/description.md
new file mode 100644
index 000000000..49f899094
--- /dev/null
+++ b/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/.docs/description.md
@@ -0,0 +1,5 @@
+<!-- 10. Description -->
+## Description
+
+This document has instructions for running <model name> <mode> using
+Intel-optimized PyTorch.
diff --git a/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/.docs/docker_spr.md b/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/.docs/docker_spr.md
new file mode 100644
index 000000000..7d3040e45
--- /dev/null
+++ b/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/.docs/docker_spr.md
@@ -0,0 +1,33 @@
+## Run the model
+
+Download the pretrained model weights using the script from the MLPerf repo
+and set the `PRETRAINED_MODEL` environment variable to point to the downloaded file:
+```
+wget https://raw.githubusercontent.com/mlcommons/inference/v0.7/others/cloud/single_stage_detector/download_model.sh
+sh download_model.sh
+export PRETRAINED_MODEL=$(pwd)/pretrained/resnet34-ssd1200.pth
+```
+
+After downloading the pretrained model and following the instructions to
+[build the container](#build-the-container) and [prepare the dataset](#datasets),
+use the `run.sh` script from the container package to run <model name> <mode>
+using docker. Set environment variables to specify the dataset directory,
+precision to run, and an output directory for logs. By default, the `run.sh`
+script will run the `inference_realtime.sh` quickstart script. To run a different
+script, specify the name of the script using the `SCRIPT` environment variable.
+```
+# Navigate to the container package directory
+cd <package dir>
+
+# Set the required environment vars
+export DATASET_DIR=<path to the coco dataset>
+export PRETRAINED_MODEL=<path to the resnet34-ssd1200.pth file>
+export PRECISION=<specify the precision to run>
+export OUTPUT_DIR=<directory where log files will be written>
+
+# Run the container with inference_realtime.sh quickstart script
+./run.sh
+
+# To run a difference quickstart script, us the SCRIPT env var
+SCRIPT=accuracy.sh ./run.sh
+```
diff --git a/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/.docs/license.md b/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/.docs/license.md
new file mode 100644
index 000000000..e547f148d
--- /dev/null
+++ b/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/.docs/license.md
@@ -0,0 +1,4 @@
+<!--- 80. License -->
+## License
+
+Licenses can be found in the model package, in the `licenses` directory.
diff --git a/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/.docs/quickstart.md b/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/.docs/quickstart.md
new file mode 100644
index 000000000..02d01973f
--- /dev/null
+++ b/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/.docs/quickstart.md
@@ -0,0 +1,8 @@
+<!--- 40. Quick Start Scripts -->
+## Quick Start Scripts
+
+| Script name | Description |
+|-------------|-------------|
+| `inference_realtime.sh` | Runs multi instance realtime inference using 4 cores per instance for the specified precision (fp32, int8 or bf16). |
+| `inference_throughput.sh` | Runs multi instance batch inference using 1 instance per socket for the specified precision (fp32, int8 or bf16). |
+| `accuracy.sh` | Measures the inference accuracy (providing a `DATASET_DIR` environment variable is required) for the specified precision (fp32, int8 or bf16). |
diff --git a/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/.docs/title.md b/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/.docs/title.md
new file mode 100644
index 000000000..a4ccbf863
--- /dev/null
+++ b/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/.docs/title.md
@@ -0,0 +1,2 @@
+<!--- 0. Title -->
+# PyTorch <model name> <mode>
diff --git a/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/.docs/wrapper_package.md b/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/.docs/wrapper_package.md
new file mode 100644
index 000000000..0d77199f3
--- /dev/null
+++ b/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/.docs/wrapper_package.md
@@ -0,0 +1,16 @@
+## Model Package
+
+The model package includes the Dockerfile and scripts needed to build and
+run <model name> <mode> in a container.
+```
+<package dir>
+├── README.md
+├── build.sh
+├── licenses
+│   ├── LICENSE
+│   └── third_party
+├── model_packages
+│   └── <package name>
+├── <package dir>.Dockerfile
+└── run.sh
+```
diff --git a/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/README_SPR.md b/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/README_SPR.md
new file mode 100644
index 000000000..f51bb7842
--- /dev/null
+++ b/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/README_SPR.md
@@ -0,0 +1,135 @@
+<!--- 0. Title -->
+# PyTorch ssd-resnet34 inference
+
+<!-- 10. Description -->
+## Description
+
+This document has instructions for running ssd-resnet34 inference using
+Intel-optimized PyTorch.
+
+## Model Package
+
+The model package includes the Dockerfile and scripts needed to build and
+run ssd-resnet34 inference in a container.
+```
+pytorch-spr-ssd-resnet34-inference
+├── README.md
+├── build.sh
+├── licenses
+│   ├── LICENSE
+│   └── third_party
+├── model_packages
+│   └── pytorch-spr-ssd-resnet34-inference.tar.gz
+├── pytorch-spr-ssd-resnet34-inference.Dockerfile
+└── run.sh
+```
+
+<!--- 40. Quick Start Scripts -->
+## Quick Start Scripts
+
+| Script name | Description |
+|-------------|-------------|
+| `inference_realtime.sh` | Runs multi instance realtime inference using 4 cores per instance for the specified precision (fp32, int8 or bf16). |
+| `inference_throughput.sh` | Runs multi instance batch inference using 1 instance per socket for the specified precision (fp32, int8 or bf16). |
+| `accuracy.sh` | Measures the inference accuracy (providing a `DATASET_DIR` environment variable is required) for the specified precision (fp32, int8 or bf16). |
+
+## Datasets
+
+### COCO
+
+The [COCO dataset](https://cocodataset.org) is used to run ssd-resnet34.
+
+Download and extract the 2017 training/validation images and annotations from the
+[COCO dataset website](https://cocodataset.org/#download) to a `coco` folder
+and unzip the files. After extracting the zip files, your dataset directory
+structure should look something like this:
+```
+coco
+├── annotations
+│   ├── captions_train2017.json
+│   ├── captions_val2017.json
+│   ├── instances_train2017.json
+│   ├── instances_val2017.json
+│   ├── person_keypoints_train2017.json
+│   └── person_keypoints_val2017.json
+├── train2017
+│   ├── 000000454854.jpg
+│   ├── 000000137045.jpg
+│   ├── 000000129582.jpg
+│   └── ...
+└── val2017
+    ├── 000000000139.jpg
+    ├── 000000000285.jpg
+    ├── 000000000632.jpg
+    └── ...
+```
+The parent of the `annotations`, `train2017`, and `val2017` directory (in this example `coco`)
+is the directory that should be used when setting the `DATASET_DIR` environment
+variable for ssd-resnet34 (for example: `export DATASET_DIR=/home/<user>/coco`).
+
+## Build the container
+
+The ssd-resnet34 inference package has scripts and a Dockerfile that are
+used to build a workload container that runs the model. This container
+uses the PyTorch/IPEX container as it's base, so ensure that you have built
+the `pytorch-ipex-spr.tar.gz` container prior to building this model container.
+
+Use `docker images` to verify that you have the base container built. For example:
+```
+$ docker images | grep pytorch-ipex-spr
+model-zoo         pytorch-ipex-spr         fecc7096a11e        40 minutes ago      8.31GB
+```
+
+To build the ssd-resnet34 inference container, extract the package and
+run the `build.sh` script.
+```
+# Extract the package
+tar -xzf pytorch-spr-ssd-resnet34-inference.tar.gz
+cd pytorch-spr-ssd-resnet34-inference
+
+# Build the container
+./build.sh
+```
+
+After the build completes, you should have a container called
+`model-zoo:pytorch-spr-ssd-resnet34-inference` that will be used to run the model.
+
+## Run the model
+
+Download the pretrained model weights using the script from the MLPerf repo
+and set the `PRETRAINED_MODEL` environment variable to point to the downloaded file:
+```
+wget https://raw.githubusercontent.com/mlcommons/inference/v0.7/others/cloud/single_stage_detector/download_model.sh
+sh download_model.sh
+export PRETRAINED_MODEL=$(pwd)/pretrained/resnet34-ssd1200.pth
+```
+
+After downloading the pretrained model and following the instructions to
+[build the container](#build-the-container) and [prepare the dataset](#datasets),
+use the `run.sh` script from the container package to run ssd-resnet34 inference
+using docker. Set environment variables to specify the dataset directory,
+precision to run, and an output directory for logs. By default, the `run.sh`
+script will run the `inference_realtime.sh` quickstart script. To run a different
+script, specify the name of the script using the `SCRIPT` environment variable.
+```
+# Navigate to the container package directory
+cd pytorch-spr-ssd-resnet34-inference
+
+# Set the required environment vars
+export DATASET_DIR=<path to the coco dataset>
+export PRETRAINED_MODEL=<path to the resnet34-ssd1200.pth file>
+export PRECISION=<specify the precision to run>
+export OUTPUT_DIR=<directory where log files will be written>
+
+# Run the container with inference_realtime.sh quickstart script
+./run.sh
+
+# To run a difference quickstart script, us the SCRIPT env var
+SCRIPT=accuracy.sh ./run.sh
+```
+
+<!--- 80. License -->
+## License
+
+Licenses can be found in the model package, in the `licenses` directory.
+
diff --git a/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/accuracy.sh b/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/accuracy.sh
new file mode 100755
index 000000000..477aea4ff
--- /dev/null
+++ b/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/accuracy.sh
@@ -0,0 +1,66 @@
+#!/usr/bin/env bash
+#
+# Copyright (c) 2021 Intel 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.
+#
+
+MODEL_DIR=${MODEL_DIR-$PWD}
+
+echo "DATASET_DIR: ${DATASET_DIR}"
+echo "PRETRAINED_MODEL: ${PRETRAINED_MODEL}"
+echo "PRECISION: ${PRECISION}"
+echo "OUTPUT_DIR: ${OUTPUT_DIR}"
+
+if [ -z "${OUTPUT_DIR}" ]; then
+  echo "The required environment variable OUTPUT_DIR has not been set"
+  exit 1
+fi
+
+# Create the output directory in case it doesn't already exist
+mkdir -p ${OUTPUT_DIR}
+
+if [ -z "${DATASET_DIR}" ]; then
+  echo "The required environment variable DATASET_DIR has not been set"
+  exit 1
+fi
+
+if [ ! -d "${DATASET_DIR}" ]; then
+  echo "The DATASET_DIR '${DATASET_DIR}' does not exist"
+  exit 1
+fi
+
+if [ -z "${PRECISION}" ]; then
+  echo "The required environment variable PRECISION has not been set"
+  echo "Please set PRECISION to fp32, int8, or bf16."
+  exit 1
+fi
+
+cd ${MODEL_DIR}/models/ssd/inference/others/cloud/single_stage_detector/pytorch
+
+# Set env vars that the bash script looks for
+export DATA_DIR=${DATASET_DIR}
+export MODEL_DIR=${PRETRAINED_MODEL}
+export work_space=${OUTPUT_DIR}
+
+if [[ $PRECISION == "int8" ]]; then
+    bash run_and_time_accuracy_cpu.sh int8 jit ./pytorch_default_recipe_ssd_configure.json 2>&1 | tee -a ${OUTPUT_DIR}/ssd-resnet34-inference-accuracy-int8.log
+elif [[ $PRECISION == "bf16" ]]; then
+    bash run_and_time_accuracy_cpu.sh bf16 jit 2>&1 | tee -a ${OUTPUT_DIR}/ssd-resnet34-inference-accuracy-bf16.log
+elif [[ $PRECISION == "fp32" ]]; then
+    bash run_and_time_accuracy_cpu.sh fp32 jit 2>&1 | tee -a ${OUTPUT_DIR}/ssd-resnet34-inference-accuracy-fp32.log
+else
+    echo "The specified precision '${PRECISION}' is unsupported."
+    echo "Supported precisions are: fp32, bf16, and int8"
+    exit 1
+fi
diff --git a/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/build.sh b/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/build.sh
new file mode 100755
index 000000000..a1862546c
--- /dev/null
+++ b/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/build.sh
@@ -0,0 +1,41 @@
+#!/usr/bin/env bash
+#
+# Copyright (c) 2021 Intel 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.
+#
+
+set -e
+
+PACKAGE_NAME=pytorch-spr-ssd-resnet34-inference
+DOCKERFILE=pytorch-spr-ssd-resnet34-inference.Dockerfile
+PYTORCH_BASE_IMAGE=${PYTORCH_BASE_IMAGE:-model-zoo}
+PYTORCH_BASE_TAG=${PYTORCH_BASE_TAG:-pytorch-ipex-spr}
+IMAGE_NAME=${IMAGE_NAME:-model-zoo:pytorch-spr-ssd-resnet34-inference}
+
+if [ "$(docker images -q ${PYTORCH_BASE_IMAGE}:${PYTORCH_BASE_TAG})" == "" ]; then
+  echo "The Intel(R) Extension for PyTorch container (${PYTORCH_BASE_IMAGE}:${PYTORCH_BASE_TAG}) was not found."
+  echo "This container is required, as it is used as the base for building the ssd-resnet34 inference container."
+  echo "Please download the IPEX container package and build the image and then retry this build."
+  exit 1
+fi
+
+docker build --build-arg PYTORCH_IMAGE=${PYTORCH_BASE_IMAGE} \
+             --build-arg PYTORCH_TAG=${PYTORCH_BASE_TAG}\
+             --build-arg PACKAGE_NAME=$PACKAGE_NAME \
+             --build-arg MODEL_WORKSPACE=/workspace \
+             --build-arg http_proxy=$http_proxy \
+             --build-arg https_proxy=$https_proxy  \
+             --build-arg no_proxy=$no_proxy \
+             -t $IMAGE_NAME \
+             -f $DOCKERFILE .
diff --git a/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/inference_realtime.sh b/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/inference_realtime.sh
new file mode 100755
index 000000000..13d0ccaa1
--- /dev/null
+++ b/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/inference_realtime.sh
@@ -0,0 +1,66 @@
+#!/usr/bin/env bash
+#
+# Copyright (c) 2021 Intel Corporation
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#    http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+
+MODEL_DIR=${MODEL_DIR-$PWD}
+
+echo "DATASET_DIR: ${DATASET_DIR}"
+echo "PRETRAINED_MODEL: ${PRETRAINED_MODEL}"
+echo "PRECISION: ${PRECISION}"
+echo "OUTPUT_DIR: ${OUTPUT_DIR}"
+
+if [ -z "${OUTPUT_DIR}" ]; then
+  echo "The required environment variable OUTPUT_DIR has not been set"
+  exit 1
+fi
+
+# Create the output directory in case it doesn't already exist
+mkdir -p ${OUTPUT_DIR}
+
+if [ -z "${DATASET_DIR}" ]; then
+  echo "The required environment variable DATASET_DIR has not been set"
+  exit 1
+fi
+
+if [ ! -d "${DATASET_DIR}" ]; then
+  echo "The DATASET_DIR '${DATASET_DIR}' does not exist"
+  exit 1
+fi
+
+if [ -z "${PRECISION}" ]; then
+  echo "The required environment variable PRECISION has not been set"
+  echo "Please set PRECISION to fp32, int8, or bf16."
+  exit 1
+fi
+
+cd ${MODEL_DIR}/models/ssd/inference/others/cloud/single_stage_detector/pytorch
+
+# Set env vars that the bash script looks for
+export DATA_DIR=${DATASET_DIR}
+export MODEL_DIR=${PRETRAINED_MODEL}
+export work_space=${OUTPUT_DIR}
+
+if [[ $PRECISION == "int8" ]]; then
+    bash run_multi_instance_ipex.sh int8 jit ./pytorch_default_recipe_ssd_configure.json 2>&1 | tee -a ${OUTPUT_DIR}/ssd-resnet34-inference-throughput-int8.log
+elif [[ $PRECISION == "bf16" ]]; then
+    bash run_multi_instance_ipex.sh bf16 jit 2>&1 | tee -a ${OUTPUT_DIR}/ssd-resnet34-inference-throughput-bf16.log
+elif [[ $PRECISION == "fp32" ]]; then
+    bash run_multi_instance_ipex.sh fp32 jit 2>&1 | tee -a ${OUTPUT_DIR}/ssd-resnet34-inference-throughput-fp32.log
+else
+    echo "The specified precision '${PRECISION}' is unsupported."
+    echo "Supported precisions are: fp32, bf16, and int8"
+    exit 1
+fi
diff --git a/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/inference_throughput.sh b/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/inference_throughput.sh
new file mode 100755
index 000000000..944cc082d
--- /dev/null
+++ b/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/inference_throughput.sh
@@ -0,0 +1,66 @@
+#!/usr/bin/env bash
+#
+# Copyright (c) 2021 Intel 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.
+#
+
+MODEL_DIR=${MODEL_DIR-$PWD}
+
+echo "DATASET_DIR: ${DATASET_DIR}"
+echo "PRETRAINED_MODEL: ${PRETRAINED_MODEL}"
+echo "PRECISION: ${PRECISION}"
+echo "OUTPUT_DIR: ${OUTPUT_DIR}"
+
+if [ -z "${OUTPUT_DIR}" ]; then
+  echo "The required environment variable OUTPUT_DIR has not been set"
+  exit 1
+fi
+
+# Create the output directory in case it doesn't already exist
+mkdir -p ${OUTPUT_DIR}
+
+if [ -z "${DATASET_DIR}" ]; then
+  echo "The required environment variable DATASET_DIR has not been set"
+  exit 1
+fi
+
+if [ ! -d "${DATASET_DIR}" ]; then
+  echo "The DATASET_DIR '${DATASET_DIR}' does not exist"
+  exit 1
+fi
+
+if [ -z "${PRECISION}" ]; then
+  echo "The required environment variable PRECISION has not been set"
+  echo "Please set PRECISION to fp32, int8, or bf16."
+  exit 1
+fi
+
+cd ${MODEL_DIR}/models/ssd/inference/others/cloud/single_stage_detector/pytorch
+
+# Set env vars that the bash script looks for
+export DATA_DIR=${DATASET_DIR}
+export MODEL_DIR=${PRETRAINED_MODEL}
+export work_space=${OUTPUT_DIR}
+
+if [[ $PRECISION == "int8" ]]; then
+    bash run_multi_instance_latency_ipex.sh int8 jit ./pytorch_default_recipe_ssd_configure.json 2>&1 | tee -a ${OUTPUT_DIR}/ssd-resnet34-inference-realtime-int8.log
+elif [[ $PRECISION == "bf16" ]]; then
+    bash run_multi_instance_latency_ipex.sh bf16 jit 2>&1 | tee -a ${OUTPUT_DIR}/ssd-resnet34-inference-realtime-bf16.log
+elif [[ $PRECISION == "fp32" ]]; then
+    bash run_multi_instance_latency_ipex.sh fp32 jit 2>&1 | tee -a ${OUTPUT_DIR}/ssd-resnet34-inference-realtime-fp32.log
+else
+    echo "The specified precision '${PRECISION}' is unsupported."
+    echo "Supported precisions are: fp32, bf16, and int8"
+    exit 1
+fi
diff --git a/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/run.sh b/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/run.sh
new file mode 100755
index 000000000..f6f2a7f33
--- /dev/null
+++ b/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/run.sh
@@ -0,0 +1,68 @@
+#!/usr/bin/env bash
+#
+# Copyright (c) 2021 Intel 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.
+#
+
+if [ -z "${OUTPUT_DIR}" ]; then
+  echo "The required environment variable OUTPUT_DIR has not been set"
+  exit 1
+fi
+
+if [ -z "${PRECISION}" ]; then
+  echo "The required environment variable PRECISION has not been set"
+  exit 1
+fi
+
+if [ -z "${DATASET_DIR}" ]; then
+  echo "The required environment variable DATASET_DIR has not been set"
+  exit 1
+fi
+
+if [ -z "${PRETRAINED_MODEL}" ]; then
+  echo "The required environment variable PRETRAINED_MODEL has not been set"
+  exit 1
+fi
+
+IMAGE_NAME=${IMAGE_NAME:-model-zoo:pytorch-spr-ssd-resnet34-inference}
+DOCKER_ARGS=${DOCKER_ARGS:---privileged --init -it}
+WORKDIR=/workspace/pytorch-spr-ssd-resnet34-inference
+
+# inference scripts:
+# inference_realtime.sh
+# inference_throughput.sh
+# accuracy.sh
+export SCRIPT="${SCRIPT:-inference_realtime.sh}"
+
+if [[ ${SCRIPT} != quickstart* ]]; then
+  SCRIPT="quickstart/$SCRIPT"
+fi
+
+docker run --rm \
+  ${dataset_env} \
+  --env DATASET_DIR=${DATASET_DIR} \
+  --env PRECISION=${PRECISION} \
+  --env PRETRAINED_MODEL=${PRETRAINED_MODEL} \
+  --env OUTPUT_DIR=${OUTPUT_DIR} \
+  --env http_proxy=${http_proxy} \
+  --env https_proxy=${https_proxy} \
+  --env no_proxy=${no_proxy} \
+  --volume ${DATASET_DIR}:${DATASET_DIR} \
+  --volume ${OUTPUT_DIR}:${OUTPUT_DIR} \
+  --volume ${PRETRAINED_MODEL}:${PRETRAINED_MODEL} \
+  --shm-size 8G \
+  -w ${WORKDIR} \
+  ${DOCKER_ARGS} \
+  $IMAGE_NAME \
+  /bin/bash $SCRIPT
diff --git a/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/.docs/container_build.md b/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/.docs/container_build.md
new file mode 100644
index 000000000..9b4f9b303
--- /dev/null
+++ b/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/.docs/container_build.md
@@ -0,0 +1,26 @@
+## Build the container
+
+The <model name> <mode> package has scripts and a Dockerfile that are
+used to build a workload container that runs the model. This container
+uses the PyTorch/IPEX container as it's base, so ensure that you have built
+the `pytorch-ipex-spr.tar.gz` container prior to building this model container.
+
+Use `docker images` to verify that you have the base container built. For example:
+```
+$ docker images | grep pytorch-ipex-spr
+model-zoo         pytorch-ipex-spr         fecc7096a11e        40 minutes ago      8.31GB
+```
+
+To build the <model name> <mode> container, extract the package and
+run the `build.sh` script.
+```
+# Extract the package
+tar -xzf <package name>
+cd <package dir>
+
+# Build the container
+./build.sh
+```
+
+After the build completes, you should have a container called
+`<docker image>` that will be used to run the model.
diff --git a/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/.docs/datasets.md b/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/.docs/datasets.md
new file mode 100644
index 000000000..78335650c
--- /dev/null
+++ b/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/.docs/datasets.md
@@ -0,0 +1,33 @@
+## Datasets
+
+### COCO
+
+The [COCO dataset](https://cocodataset.org) is used to run <model name>.
+
+Download and extract the 2017 training/validation images and annotations from the
+[COCO dataset website](https://cocodataset.org/#download) to a `coco` folder
+and unzip the files. After extracting the zip files, your dataset directory
+structure should look something like this:
+```
+coco
+├── annotations
+│   ├── captions_train2017.json
+│   ├── captions_val2017.json
+│   ├── instances_train2017.json
+│   ├── instances_val2017.json
+│   ├── person_keypoints_train2017.json
+│   └── person_keypoints_val2017.json
+├── train2017
+│   ├── 000000454854.jpg
+│   ├── 000000137045.jpg
+│   ├── 000000129582.jpg
+│   └── ...
+└── val2017
+    ├── 000000000139.jpg
+    ├── 000000000285.jpg
+    ├── 000000000632.jpg
+    └── ...
+```
+The parent of the `annotations`, `train2017`, and `val2017` directory (in this example `coco`)
+is the directory that should be used when setting the `DATASET_DIR` environment
+variable for <model name> (for example: `export DATASET_DIR=/home/<user>/coco`).
diff --git a/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/.docs/description.md b/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/.docs/description.md
new file mode 100644
index 000000000..49f899094
--- /dev/null
+++ b/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/.docs/description.md
@@ -0,0 +1,5 @@
+<!-- 10. Description -->
+## Description
+
+This document has instructions for running <model name> <mode> using
+Intel-optimized PyTorch.
diff --git a/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/.docs/docker_spr.md b/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/.docs/docker_spr.md
new file mode 100644
index 000000000..012bb3997
--- /dev/null
+++ b/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/.docs/docker_spr.md
@@ -0,0 +1,32 @@
+## Run the model
+
+Download the backbone weights and set the `BACKBONE_WEIGHTS` environment variable
+to point to the downloaded file:
+```
+curl -O https://download.pytorch.org/models/resnet34-333f7ec4.pth
+export BACKBONE_WEIGHTS=$(pwd)/resnet34-333f7ec4.pth
+```
+
+After you've downloaded the backbone weights and followed the instructions to
+[build the container](#build-the-container) and [prepare the dataset](#datasets),
+use the `run.sh` script from the container package to run <model name> <mode>
+using docker. Set environment variables to point to the COCO dataset directory,
+weights, precision, and an output directory for logs. By default, the `run.sh`
+script will run the `train_performance.sh` quickstart script. To run the `train_accuracy.sh`
+script instead, specify that script name using the `SCRIPT` environment variable.
+```
+# Navigate to the container package directory
+cd <package dir>
+
+# Set the required environment vars
+export DATASET_DIR=<path to the COCO dataset directory>
+export BACKBONE_WEIGHTS=<path to the downloaded weights file>
+export PRECISION=<specify the precision to run>
+export OUTPUT_DIR=<directory where log files will be written>
+
+# Run the container with train_performance.sh quickstart script
+./run.sh
+
+# Run a different script by specifying the SCRIPT env var
+SCRIPT=train_accuracy.sh ./run.sh
+```
diff --git a/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/.docs/license.md b/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/.docs/license.md
new file mode 100644
index 000000000..e547f148d
--- /dev/null
+++ b/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/.docs/license.md
@@ -0,0 +1,4 @@
+<!--- 80. License -->
+## License
+
+Licenses can be found in the model package, in the `licenses` directory.
diff --git a/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/.docs/quickstart.md b/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/.docs/quickstart.md
new file mode 100644
index 000000000..57f5f9e43
--- /dev/null
+++ b/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/.docs/quickstart.md
@@ -0,0 +1,7 @@
+<!--- 40. Quick Start Scripts -->
+## Quick Start Scripts
+
+| Script name | Description |
+|-------------|-------------|
+| `train_performance.sh` | Tests the training performance for SSD-ResNet34 for the specified precision (fp32 or bf16). |
+| `train_accuracy.sh` | Tests the training accuracy for SSD-ResNet34 for the specified precision (fp32 or bf16). |
diff --git a/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/.docs/title.md b/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/.docs/title.md
new file mode 100644
index 000000000..a4ccbf863
--- /dev/null
+++ b/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/.docs/title.md
@@ -0,0 +1,2 @@
+<!--- 0. Title -->
+# PyTorch <model name> <mode>
diff --git a/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/.docs/wrapper_package.md b/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/.docs/wrapper_package.md
new file mode 100644
index 000000000..0d77199f3
--- /dev/null
+++ b/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/.docs/wrapper_package.md
@@ -0,0 +1,16 @@
+## Model Package
+
+The model package includes the Dockerfile and scripts needed to build and
+run <model name> <mode> in a container.
+```
+<package dir>
+├── README.md
+├── build.sh
+├── licenses
+│   ├── LICENSE
+│   └── third_party
+├── model_packages
+│   └── <package name>
+├── <package dir>.Dockerfile
+└── run.sh
+```
diff --git a/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/README_SPR.md b/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/README_SPR.md
new file mode 100644
index 000000000..c47a9047b
--- /dev/null
+++ b/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/README_SPR.md
@@ -0,0 +1,133 @@
+<!--- 0. Title -->
+# PyTorch ssd-resnet34 training
+
+<!-- 10. Description -->
+## Description
+
+This document has instructions for running ssd-resnet34 training using
+Intel-optimized PyTorch.
+
+## Model Package
+
+The model package includes the Dockerfile and scripts needed to build and
+run ssd-resnet34 training in a container.
+```
+pytorch-spr-ssd-resnet34-training
+├── README.md
+├── build.sh
+├── licenses
+│   ├── LICENSE
+│   └── third_party
+├── model_packages
+│   └── pytorch-spr-ssd-resnet34-training.tar.gz
+├── pytorch-spr-ssd-resnet34-training.Dockerfile
+└── run.sh
+```
+
+<!--- 40. Quick Start Scripts -->
+## Quick Start Scripts
+
+| Script name | Description |
+|-------------|-------------|
+| `train_performance.sh` | Tests the training performance for SSD-ResNet34 for the specified precision (fp32 or bf16). |
+| `train_accuracy.sh` | Tests the training accuracy for SSD-ResNet34 for the specified precision (fp32 or bf16). |
+
+## Datasets
+
+### COCO
+
+The [COCO dataset](https://cocodataset.org) is used to run ssd-resnet34.
+
+Download and extract the 2017 training/validation images and annotations from the
+[COCO dataset website](https://cocodataset.org/#download) to a `coco` folder
+and unzip the files. After extracting the zip files, your dataset directory
+structure should look something like this:
+```
+coco
+├── annotations
+│   ├── captions_train2017.json
+│   ├── captions_val2017.json
+│   ├── instances_train2017.json
+│   ├── instances_val2017.json
+│   ├── person_keypoints_train2017.json
+│   └── person_keypoints_val2017.json
+├── train2017
+│   ├── 000000454854.jpg
+│   ├── 000000137045.jpg
+│   ├── 000000129582.jpg
+│   └── ...
+└── val2017
+    ├── 000000000139.jpg
+    ├── 000000000285.jpg
+    ├── 000000000632.jpg
+    └── ...
+```
+The parent of the `annotations`, `train2017`, and `val2017` directory (in this example `coco`)
+is the directory that should be used when setting the `DATASET_DIR` environment
+variable for ssd-resnet34 (for example: `export DATASET_DIR=/home/<user>/coco`).
+
+## Build the container
+
+The ssd-resnet34 training package has scripts and a Dockerfile that are
+used to build a workload container that runs the model. This container
+uses the PyTorch/IPEX container as it's base, so ensure that you have built
+the `pytorch-ipex-spr.tar.gz` container prior to building this model container.
+
+Use `docker images` to verify that you have the base container built. For example:
+```
+$ docker images | grep pytorch-ipex-spr
+model-zoo         pytorch-ipex-spr         fecc7096a11e        40 minutes ago      8.31GB
+```
+
+To build the ssd-resnet34 training container, extract the package and
+run the `build.sh` script.
+```
+# Extract the package
+tar -xzf pytorch-spr-ssd-resnet34-training.tar.gz
+cd pytorch-spr-ssd-resnet34-training
+
+# Build the container
+./build.sh
+```
+
+After the build completes, you should have a container called
+`model-zoo:pytorch-spr-ssd-resnet34-training` that will be used to run the model.
+
+## Run the model
+
+Download the backbone weights and set the `BACKBONE_WEIGHTS` environment variable
+to point to the downloaded file:
+```
+curl -O https://download.pytorch.org/models/resnet34-333f7ec4.pth
+export BACKBONE_WEIGHTS=$(pwd)/resnet34-333f7ec4.pth
+```
+
+After you've downloaded the backbone weights and followed the instructions to
+[build the container](#build-the-container) and [prepare the dataset](#datasets),
+use the `run.sh` script from the container package to run ssd-resnet34 training
+using docker. Set environment variables to point to the COCO dataset directory,
+weights, precision, and an output directory for logs. By default, the `run.sh`
+script will run the `train_performance.sh` quickstart script. To run the `train_accuracy.sh`
+script instead, specify that script name using the `SCRIPT` environment variable.
+```
+# Navigate to the container package directory
+cd pytorch-spr-ssd-resnet34-training
+
+# Set the required environment vars
+export DATASET_DIR=<path to the COCO dataset directory>
+export BACKBONE_WEIGHTS=<path to the downloaded weights file>
+export PRECISION=<specify the precision to run>
+export OUTPUT_DIR=<directory where log files will be written>
+
+# Run the container with train_performance.sh quickstart script
+./run.sh
+
+# Run a different script by specifying the SCRIPT env var
+SCRIPT=train_accuracy.sh ./run.sh
+```
+
+<!--- 80. License -->
+## License
+
+Licenses can be found in the model package, in the `licenses` directory.
+
diff --git a/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/build.sh b/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/build.sh
new file mode 100755
index 000000000..b858529b2
--- /dev/null
+++ b/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/build.sh
@@ -0,0 +1,41 @@
+#!/usr/bin/env bash
+#
+# Copyright (c) 2021 Intel 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.
+#
+
+set -e
+
+PACKAGE_NAME=pytorch-spr-ssd-resnet34-training
+DOCKERFILE=pytorch-spr-ssd-resnet34-training.Dockerfile
+PYTORCH_BASE_IMAGE=${PYTORCH_BASE_IMAGE:-model-zoo}
+PYTORCH_BASE_TAG=${PYTORCH_BASE_TAG:-pytorch-ipex-spr}
+IMAGE_NAME=${IMAGE_NAME:-model-zoo:pytorch-spr-ssd-resnet34-training}
+
+if [ "$(docker images -q ${PYTORCH_BASE_IMAGE}:${PYTORCH_BASE_TAG})" == "" ]; then
+  echo "The Intel(R) Extension for PyTorch container (${PYTORCH_BASE_IMAGE}:${PYTORCH_BASE_TAG}) was not found."
+  echo "This container is required, as it is used as the base for building the ssd-resnet34 training container."
+  echo "Please download the IPEX container package and build the image and then retry this build."
+  exit 1
+fi
+
+docker build --build-arg PYTORCH_IMAGE=${PYTORCH_BASE_IMAGE} \
+             --build-arg PYTORCH_TAG=${PYTORCH_BASE_TAG} \
+             --build-arg PACKAGE_NAME=$PACKAGE_NAME \
+             --build-arg MODEL_WORKSPACE=/workspace \
+             --build-arg http_proxy=$http_proxy \
+             --build-arg https_proxy=$https_proxy  \
+             --build-arg no_proxy=$no_proxy \
+             -t $IMAGE_NAME \
+             -f $DOCKERFILE .
diff --git a/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/run.sh b/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/run.sh
new file mode 100755
index 000000000..29961c1b1
--- /dev/null
+++ b/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/run.sh
@@ -0,0 +1,67 @@
+#!/usr/bin/env bash
+#
+# Copyright (c) 2021 Intel Corporation
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#    http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+
+if [ -z "${OUTPUT_DIR}" ]; then
+  echo "The required environment variable OUTPUT_DIR has not been set"
+  exit 1
+fi
+
+if [ -z "${PRECISION}" ]; then
+  echo "The required environment variable PRECISION has not been set"
+  exit 1
+fi
+
+if [ -z "${DATASET_DIR}" ]; then
+  echo "The required environment variable DATASET_DIR has not been set"
+  exit 1
+fi
+
+if [ -z "${BACKBONE_WEIGHTS}" ]; then
+  echo "The required environment variable BACKBONE_WEIGHTS has not been set"
+  exit 1
+fi
+
+IMAGE_NAME=${IMAGE_NAME:-model-zoo:pytorch-spr-ssd-resnet34-training}
+DOCKER_ARGS=${DOCKER_ARGS:---privileged --init -it}
+WORKDIR=/workspace/pytorch-spr-ssd-resnet34-training
+
+# training scripts:
+# train_performance.sh
+# train_accuracy.sh
+export SCRIPT="${SCRIPT:-train_performance.sh}"
+
+if [[ ${SCRIPT} != quickstart* ]]; then
+  SCRIPT="quickstart/$SCRIPT"
+fi
+
+docker run --rm \
+  ${dataset_env} \
+  --env PRECISION=${PRECISION} \
+  --env OUTPUT_DIR=${OUTPUT_DIR} \
+  --env DATASET_DIR=${DATASET_DIR} \
+  --env BACKBONE_WEIGHTS=${BACKBONE_WEIGHTS} \
+  --env http_proxy=${http_proxy} \
+  --env https_proxy=${https_proxy} \
+  --env no_proxy=${no_proxy} \
+  --volume ${DATASET_DIR}:${DATASET_DIR} \
+  --volume ${BACKBONE_WEIGHTS}:${BACKBONE_WEIGHTS} \
+  --volume ${OUTPUT_DIR}:${OUTPUT_DIR} \
+  --shm-size 8G \
+  -w ${WORKDIR} \
+  ${DOCKER_ARGS} \
+  $IMAGE_NAME \
+  /bin/bash $SCRIPT
diff --git a/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/train_accuracy.sh b/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/train_accuracy.sh
new file mode 100755
index 000000000..8d42e6cfb
--- /dev/null
+++ b/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/train_accuracy.sh
@@ -0,0 +1,69 @@
+#!/usr/bin/env bash
+#
+# Copyright (c) 2021 Intel Corporation
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#    http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+
+MODEL_DIR=${MODEL_DIR-$PWD}
+
+echo "DATASET_DIR: ${DATASET_DIR}"
+echo "BACKBONE_WEIGHTS: ${BACKBONE_WEIGHTS}"
+echo "PRECISION: ${PRECISION}"
+echo "OUTPUT_DIR: ${OUTPUT_DIR}"
+
+if [ -z "${OUTPUT_DIR}" ]; then
+  echo "The required environment variable OUTPUT_DIR has not been set"
+  exit 1
+fi
+
+# Create the output directory in case it doesn't already exist
+mkdir -p ${OUTPUT_DIR}
+
+if [ -z "${DATASET_DIR}" ]; then
+  echo "The required environment variable DATASET_DIR has not been set"
+  exit 1
+fi
+
+if [ ! -d "${DATASET_DIR}" ]; then
+  echo "The DATASET_DIR '${DATASET_DIR}' does not exist"
+  exit 1
+fi
+
+if [ -z "${PRECISION}" ]; then
+  echo "The required environment variable PRECISION has not been set"
+  echo "Please set PRECISION to fp32 or bf16."
+  exit 1
+fi
+
+if [ ! -f "${BACKBONE_WEIGHTS}" ]; then
+  echo "The BACKBONE_WEIGHTS '${BACKBONE_WEIGHTS}' file does not exist"
+  exit 1
+fi
+
+cd ${MODEL_DIR}/models/ssd-training/training/single_stage_detector/ssd
+
+# Set env vars that the bash script looks for
+export DATA_DIR=${DATASET_DIR}
+export MODEL_DIR=${BACKBONE_WEIGHTS}
+export work_space=${OUTPUT_DIR}
+
+if [[ $PRECISION == "bf16" ]]; then
+    bash run_accuracy.sh bf16 2>&1 | tee -a ${OUTPUT_DIR}/ssd-resnet34-training-accuracy-bf16.log
+elif [[ $PRECISION == "fp32" ]]; then
+    bash run_accuracy.sh fp32 fp32 2>&1 | tee -a ${OUTPUT_DIR}/ssd-resnet34-training-accuracy-fp32.log
+else
+    echo "The specified precision '${PRECISION}' is unsupported."
+    echo "Supported precisions are: fp32 and bf16"
+    exit 1
+fi
diff --git a/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/train_performance.sh b/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/train_performance.sh
new file mode 100755
index 000000000..a95876395
--- /dev/null
+++ b/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/train_performance.sh
@@ -0,0 +1,69 @@
+#!/usr/bin/env bash
+#
+# Copyright (c) 2021 Intel 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.
+#
+
+MODEL_DIR=${MODEL_DIR-$PWD}
+
+echo "DATASET_DIR: ${BACKBONE_WEIGHTS}"
+echo "BACKBONE_WEIGHTS: ${PRETRAINED_MODEL}"
+echo "PRECISION: ${PRECISION}"
+echo "OUTPUT_DIR: ${OUTPUT_DIR}"
+
+if [ -z "${OUTPUT_DIR}" ]; then
+  echo "The required environment variable OUTPUT_DIR has not been set"
+  exit 1
+fi
+
+# Create the output directory in case it doesn't already exist
+mkdir -p ${OUTPUT_DIR}
+
+if [ -z "${DATASET_DIR}" ]; then
+  echo "The required environment variable DATASET_DIR has not been set"
+  exit 1
+fi
+
+if [ ! -d "${DATASET_DIR}" ]; then
+  echo "The DATASET_DIR '${DATASET_DIR}' does not exist"
+  exit 1
+fi
+
+if [ -z "${PRECISION}" ]; then
+  echo "The required environment variable PRECISION has not been set"
+  echo "Please set PRECISION to fp32 or bf16."
+  exit 1
+fi
+
+if [ ! -f "${BACKBONE_WEIGHTS}" ]; then
+  echo "The BACKBONE_WEIGHTS '${BACKBONE_WEIGHTS}' file does not exist"
+  exit 1
+fi
+
+cd ${MODEL_DIR}/models/ssd-training/training/single_stage_detector/ssd
+
+# Set env vars that the bash script looks for
+export DATA_DIR=${DATASET_DIR}
+export MODEL_DIR=${BACKBONE_WEIGHTS}
+export work_space=${OUTPUT_DIR}
+
+if [[ $PRECISION == "bf16" ]]; then
+    bash run_performance.sh bf16 2>&1 | tee -a ${OUTPUT_DIR}/ssd-resnet34-training-performance-bf16.log
+elif [[ $PRECISION == "fp32" ]]; then
+    bash run_performance.sh fp32 2>&1 | tee -a ${OUTPUT_DIR}/ssd-resnet34-training-performance-fp32.log
+else
+    echo "The specified precision '${PRECISION}' is unsupported."
+    echo "Supported precisions are: fp32 and bf16"
+    exit 1
+fi
diff --git a/tools/docker/partials/common/pytorch/models/ssd-resnet34-train-dependencies.partial.Dockerfile b/tools/docker/partials/common/pytorch/models/ssd-resnet34-train-dependencies.partial.Dockerfile
new file mode 100644
index 000000000..d47b8d054
--- /dev/null
+++ b/tools/docker/partials/common/pytorch/models/ssd-resnet34-train-dependencies.partial.Dockerfile
@@ -0,0 +1,17 @@
+RUN source ~/anaconda3/bin/activate pytorch && \
+    pip install --upgrade pip && \
+    pip install --no-cache-dir https://github.com/mlperf/logging/archive/9ea0afa.zip && \
+    pip install --no-cache-dir \
+        Cython==0.28.4 \
+        git+http://github.com/NVIDIA/apex.git@9041a868a1a253172d94b113a963375b9badd030#egg=apex \
+        mlperf-compliance==0.0.10 \
+        cycler==0.10.0 \
+        kiwisolver==1.0.1 \
+        matplotlib==2.2.2 \
+        Pillow==5.2.0 \
+        pyparsing==2.2.0 \
+        python-dateutil==2.7.3 \
+        pytz==2018.5 \
+        six==1.11.0 \
+        torchvision==0.2.1 \
+        pycocotools==2.0.2
diff --git a/tools/docker/specs/centos/pytorch/spr-ssd-resnet34-inference_spec.yml b/tools/docker/specs/centos/pytorch/spr-ssd-resnet34-inference_spec.yml
new file mode 100644
index 000000000..e942bece1
--- /dev/null
+++ b/tools/docker/specs/centos/pytorch/spr-ssd-resnet34-inference_spec.yml
@@ -0,0 +1,69 @@
+releases:
+  versioned:
+    tag_specs:
+    - '{pytorch-multistage}{spr-ssd-resnet34-inference}'
+slice_sets:
+  spr-ssd-resnet34-inference:
+    - add_to_name: -spr-ssd-resnet34-inference
+      dockerfile_subdirectory: pytorch
+      args:
+        - PYTORCH_IMAGE=model-zoo
+        - PYTORCH_TAG=pytorch-ipex-spr
+        - PACKAGE_NAME=pytorch-spr-ssd-resnet34-inference
+      partials:
+        - pytorch/torch-vision-from-source
+        - pytorch/spr-model-installs
+        - model_package
+        - pytorch/spr-release-conda-env
+        - pytorch/spr-dnnl-max-var
+        - pytorch/spr-entrypoint
+      files:
+        - source: tools/docker/models/cpu-models/ssd-ww32
+          destination: models/ssd
+        - source: quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/inference_realtime.sh
+          destination: quickstart/inference_realtime.sh
+        - source: quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/inference_throughput.sh
+          destination: quickstart/inference_throughput.sh
+        - source: quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/accuracy.sh
+          destination: quickstart/accuracy.sh
+      wrapper_package_files:
+        - source: output/pytorch-spr-ssd-resnet34-inference.tar.gz
+          destination: model_packages/pytorch-spr-ssd-resnet34-inference.tar.gz
+        - source: quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/build.sh
+          destination: build.sh
+        - source: quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/run.sh
+          destination: run.sh
+        - source: dockerfiles/pytorch/pytorch-spr-ssd-resnet34-inference.Dockerfile
+          destination: pytorch-spr-ssd-resnet34-inference.Dockerfile
+        - source: LICENSE
+          destination: licenses/LICENSE
+        - source: third_party
+          destination: licenses/third_party
+        - source: quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/README_SPR.md
+          destination: README.md
+      documentation:
+        - docs:
+          - name: Title
+            uri: models/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/.docs/title.md
+          - name: Description
+            uri: models/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/.docs/description.md
+          - name: Model Package
+            uri: models/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/.docs/wrapper_package.md
+          - name: Quickstart
+            uri: models/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/.docs/quickstart.md
+          - name: Datasets
+            uri: models/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/.docs/datasets.md
+          - name: Container build
+            uri: models/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/.docs/container_build.md
+          - name: Docker
+            uri: models/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/.docs/docker_spr.md
+          - name: License
+            uri: models/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu/.docs/license.md
+          name: README_SPR.md
+          text_replace:
+            <model name>: ssd-resnet34
+            <mode>: inference
+            <package name>: pytorch-spr-ssd-resnet34-inference.tar.gz
+            <package dir>: pytorch-spr-ssd-resnet34-inference
+            <docker image>: model-zoo:pytorch-spr-ssd-resnet34-inference
+          uri: models/quickstart/object_detection/pytorch/ssd-resnet34/inference/cpu
diff --git a/tools/docker/specs/centos/pytorch/spr-ssd-resnet34-training_spec.yml b/tools/docker/specs/centos/pytorch/spr-ssd-resnet34-training_spec.yml
new file mode 100644
index 000000000..06f65e235
--- /dev/null
+++ b/tools/docker/specs/centos/pytorch/spr-ssd-resnet34-training_spec.yml
@@ -0,0 +1,68 @@
+releases:
+  versioned:
+    tag_specs:
+    - '{pytorch-multistage}{spr-ssd-resnet34-training}'
+slice_sets:
+  spr-ssd-resnet34-training:
+    - add_to_name: -spr-ssd-resnet34-training
+      dockerfile_subdirectory: pytorch
+      args:
+        - PYTORCH_IMAGE=model-zoo
+        - PYTORCH_TAG=pytorch-ipex-spr
+        - PACKAGE_NAME=pytorch-spr-ssd-resnet34-training
+        - SSD_RESNET34_TRAIN_DIR=/workspace/pytorch-spr-ssd-resnet34-training/models/ssd-training
+      partials:
+        - pytorch/spr-model-installs
+        - model_package
+        - pytorch/models/ssd-resnet34-train-dependencies
+        - pytorch/spr-release-conda-env
+        - pytorch/spr-dnnl-max-var
+        - pytorch/spr-entrypoint
+      files:
+        - source: tools/docker/models/cpu-models/ssd-training-ww32
+          destination: models/ssd-training
+        - source: quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/train_performance.sh
+          destination: quickstart/train_performance.sh
+        - source: quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/train_accuracy.sh
+          destination: quickstart/train_accuracy.sh
+      wrapper_package_files:
+        - source: output/pytorch-spr-ssd-resnet34-training.tar.gz
+          destination: model_packages/pytorch-spr-ssd-resnet34-training.tar.gz
+        - source: quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/build.sh
+          destination: build.sh
+        - source: quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/run.sh
+          destination: run.sh
+        - source: dockerfiles/pytorch/pytorch-spr-ssd-resnet34-training.Dockerfile
+          destination: pytorch-spr-ssd-resnet34-training.Dockerfile
+        - source: LICENSE
+          destination: licenses/LICENSE
+        - source: third_party
+          destination: licenses/third_party
+        - source: quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/README_SPR.md
+          destination: README.md
+      documentation:
+        - docs:
+          - name: Title
+            uri: models/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/.docs/title.md
+          - name: Description
+            uri: models/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/.docs/description.md
+          - name: Model Package
+            uri: models/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/.docs/wrapper_package.md
+          - name: Quickstart
+            uri: models/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/.docs/quickstart.md
+          - name: Datasets
+            uri: models/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/.docs/datasets.md
+          - name: Container build
+            uri: models/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/.docs/container_build.md
+          - name: Docker
+            uri: models/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/.docs/docker_spr.md
+          - name: License
+            uri: models/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu/.docs/license.md
+          name: README_SPR.md
+          text_replace:
+            <model name>: ssd-resnet34
+            <mode>: training
+            <package name>: pytorch-spr-ssd-resnet34-training.tar.gz
+            <package dir>: pytorch-spr-ssd-resnet34-training
+            <docker image>: model-zoo:pytorch-spr-ssd-resnet34-training
+          uri: models/quickstart/object_detection/pytorch/ssd-resnet34/training/cpu