Prepere for IMZ v2.0.0 release

Author: ashahba

.gitignore

@@ -10,3 +10,4 @@
 test_data/
 download_glue_data.py
 data/
+output/

README.md

@@ -2,6 +2,8 @@
 
 This repository contains **links to pre-trained models, sample scripts, best practices, and step-by-step tutorials** for many popular open-source machine learning models optimized by Intel to run on Intel® Xeon® Scalable processors. 
 
+Model packages and containers for running the Model Zoo's workloads can be found at the [Intel® oneContainer Portal](https://software.intel.com/containers).
+
 ## Purpose of the Model Zoo
 
   - Demonstrate the AI workloads and deep learning models Intel has optimized and validated to run on Intel hardware

SECURITY.md

@@ -0,0 +1,12 @@
+# Security Policy
+
+## Report a Vulnerability
+
+Please report security issues or vulnerabilities to the [Intel® Security Center].
+
+For more information on how Intel® works to resolve security issues, see
+[Vulnerability Handling Guidelines].
+
+[Intel® Security Center]:https://www.intel.com/security
+
+[Vulnerability Handling Guidelines]:https://www.intel.com/content/www/us/en/security-center/vulnerability-handling-guidelines.html

benchmarks/README.md

@@ -23,20 +23,25 @@ dependencies to be installed:
 | Image Recognition      | TensorFlow    | [ResNet 50](https://arxiv.org/pdf/1512.03385.pdf)           | Inference | [Int8](image_recognition/tensorflow/resnet50/README.md#int8-inference-instructions) [FP32](image_recognition/tensorflow/resnet50/README.md#fp32-inference-instructions)|
 | Image Recognition      | TensorFlow    | [ResNet 50v1.5](https://github.com/tensorflow/models/tree/master/official/resnet) | Inference | [Int8](image_recognition/tensorflow/resnet50v1_5/README.md#int8-inference-instructions) [FP32](image_recognition/tensorflow/resnet50v1_5/README.md#fp32-inference-instructions) [BFloat16**](image_recognition/tensorflow/resnet50v1_5/README.md#bfloat16-inference-instructions)|
 | Image Recognition      | TensorFlow    | [ResNet 50v1.5](https://github.com/tensorflow/models/tree/master/official/resnet) | Training | [FP32](image_recognition/tensorflow/resnet50v1_5/README.md#fp32-training-instructions) [BFloat16**](image_recognition/tensorflow/resnet50v1_5/README.md#bfloat16-training-instructions)|
+| Image Segmentation     | TensorFlow    | [UNet](https://arxiv.org/pdf/1505.04597.pdf)                | Inference | [FP32](image_segmentation/tensorflow/unet/README.md#fp32-inference-instructions) |
+| Image Segmentation     | TensorFlow    | [MaskRCNN](https://arxiv.org/abs/1703.06870) | Inference | [FP32](image_segmentation/tensorflow/maskrcnn/README.md#fp32-training-instructions) |
 | Language Modeling      | TensorFlow    | [BERT](https://arxiv.org/pdf/1810.04805.pdf)                | Inference | [FP32](language_modeling/tensorflow/bert_large/README.md#fp32-inference-instructions) [BFloat16**](language_modeling/tensorflow/bert_large/README.md#bfloat16-inference-instructions) |
 | Language Modeling      | TensorFlow    | [BERT](https://arxiv.org/pdf/1810.04805.pdf)                | Training  | [FP32](language_modeling/tensorflow/bert_large/README.md#fp32-training-instructions) [BFloat16**](language_modeling/tensorflow/bert_large/README.md#bfloat16-training-instructions) |
 | Language Translation   | TensorFlow    | [BERT](https://arxiv.org/pdf/1810.04805.pdf)                | Inference | [FP32](language_translation/tensorflow/bert/README.md#fp32-inference-instructions) |
 | Language Translation   | TensorFlow    | [GNMT*](https://arxiv.org/pdf/1609.08144.pdf)                | Inference | [FP32](language_translation/tensorflow/mlperf_gnmt/README.md#fp32-inference-instructions) |
 | Language Translation   | TensorFlow    | [Transformer_LT_Official ](https://arxiv.org/pdf/1706.03762.pdf)| Inference | [FP32](language_translation/tensorflow/transformer_lt_official/README.md#fp32-inference-instructions) |
 | Language Translation   | TensorFlow    | [Transformer_LT_mlperf ](https://arxiv.org/pdf/1706.03762.pdf)| Training | [FP32](language_translation/tensorflow/transformer_mlperf/README.md#fp32-training-instructions) [BFloat16**](language_translation/tensorflow/transformer_mlperf/README.md#bfloat16-training-instructions) |
+| Object Detection       | TensorFlow    | [Faster R-CNN](https://arxiv.org/pdf/1506.01497.pdf)        | Inference | [Int8](object_detection/tensorflow/faster_rcnn/README.md#int8-inference-instructions) [FP32](object_detection/tensorflow/faster_rcnn/README.md#fp32-inference-instructions) |
 | Object Detection       | TensorFlow    | [R-FCN](https://arxiv.org/pdf/1605.06409.pdf)               | Inference | [Int8](object_detection/tensorflow/rfcn/README.md#int8-inference-instructions) [FP32](object_detection/tensorflow/rfcn/README.md#fp32-inference-instructions) |
 | Object Detection       | TensorFlow    | [SSD-MobileNet*](https://arxiv.org/pdf/1704.04861.pdf)       | Inference | [Int8](object_detection/tensorflow/ssd-mobilenet/README.md#int8-inference-instructions) [FP32](object_detection/tensorflow/ssd-mobilenet/README.md#fp32-inference-instructions) |
 | Object Detection       | TensorFlow    | [SSD-ResNet34*](https://arxiv.org/pdf/1512.02325.pdf)        | Inference | [Int8](object_detection/tensorflow/ssd-resnet34/README.md#int8-inference-instructions) [FP32](object_detection/tensorflow/ssd-resnet34/README.md#fp32-inference-instructions) |
 | Object Detection       | TensorFlow    | [SSD-ResNet34](https://arxiv.org/pdf/1512.02325.pdf)        | Training  | [FP32](object_detection/tensorflow/ssd-resnet34/README.md#fp32-training-instructions) [BFloat16**](object_detection/tensorflow/ssd-resnet34/README.md#bf16-training-instructions) |
+| Recommendation         | TensorFlow    | [NCF](https://arxiv.org/pdf/1708.05031.pdf)                 | Inference | [FP32](recommendation/tensorflow/ncf/README.md#fp32-inference-instructions) |
 | Recommendation         | TensorFlow    | [Wide & Deep Large Dataset](https://arxiv.org/pdf/1606.07792.pdf)	| Inference | [Int8](recommendation/tensorflow/wide_deep_large_ds/README.md#int8-inference-instructions) [FP32](recommendation/tensorflow/wide_deep_large_ds/README.md#fp32-inference-instructions) |
 | Recommendation         | TensorFlow    | [Wide & Deep Large Dataset](https://arxiv.org/pdf/1606.07792.pdf)	| Training | [FP32](recommendation/tensorflow/wide_deep_large_ds/README.md#fp32-training-instructions) |
 | Recommendation         | TensorFlow    | [Wide & Deep](https://arxiv.org/pdf/1606.07792.pdf)         | Inference | [FP32](recommendation/tensorflow/wide_deep/README.md#fp32-inference-instructions) |
 | Reinforcement          | TensorFlow    | [MiniGo](https://arxiv.org/abs/1712.01815.pdf)              | Training  | [FP32](reinforcement/tensorflow/minigo/README.md#fp32-training-instructions)|
+| Text-to-Speech         | TensorFlow    | [WaveNet](https://arxiv.org/pdf/1609.03499.pdf)             | Inference | [FP32](text_to_speech/tensorflow/wavenet/README.md#fp32-inference-instructions) |
 
 ## TensorFlow Serving Use Cases
 

benchmarks/common/tensorflow/start.sh

@@ -90,8 +90,8 @@ if [[ ${NOINSTALL} != "True" ]]; then
     export HOROVOD_WITHOUT_PYTORCH=1
     export HOROVOD_WITHOUT_MXNET=1
     export HOROVOD_WITH_TENSORFLOW=1
-    # lock horovod==0.19.1 release commit/version
-    pip install --no-cache-dir horovod==0.19.1
+    # lock horovod==0.20.0 release commit/version
+    pip install --no-cache-dir horovod==0.20.0
   fi 
 fi
 
@@ -500,6 +500,51 @@ function densenet169() {
   fi
 }
 
+# Faster R-CNN (ResNet50) model
+function faster_rcnn() {
+    export PYTHONPATH=$PYTHONPATH:${MOUNT_EXTERNAL_MODELS_SOURCE}/research:${MOUNT_EXTERNAL_MODELS_SOURCE}/research/slim
+    original_dir=$(pwd)
+
+    if [ ${NOINSTALL} != "True" ]; then
+      # install dependencies
+      pip install -r "${MOUNT_BENCHMARK}/object_detection/tensorflow/faster_rcnn/requirements.txt"
+      cd "${MOUNT_EXTERNAL_MODELS_SOURCE}/research"
+      # install protoc v3.3.0, if necessary, then compile protoc files
+      install_protoc "https://github.com/google/protobuf/releases/download/v3.3.0/protoc-3.3.0-linux-x86_64.zip"
+
+      # Install git so that we can apply the patch
+      apt-get update && apt-get install -y git
+    fi
+
+    # Apply the patch to the tensorflow/models repo with fixes for the accuracy
+    # script and for running with python 3
+    cd ${MOUNT_EXTERNAL_MODELS_SOURCE}
+    git apply ${MOUNT_INTELAI_MODELS_SOURCE}/${MODE}/${PRECISION}/faster_rcnn.patch
+
+    if [ ${PRECISION} == "fp32" ]; then
+      if [ -n "${config_file}" ]; then
+        CMD="${CMD} --config_file=${config_file}"
+      fi
+
+      if [[ -z "${config_file}" ]] && [ ${BENCHMARK_ONLY} == "True" ]; then
+        echo "Fast R-CNN requires -- config_file arg to be defined"
+        exit 1
+      fi
+
+    elif [ ${PRECISION} == "int8" ]; then
+      number_of_steps_arg=""
+      if [ -n "${number_of_steps}" ] && [ ${BENCHMARK_ONLY} == "True" ]; then
+        CMD="${CMD} --number-of-steps=${number_of_steps}"
+      fi
+    else
+      echo "PRECISION=${PRECISION} is not supported for ${MODEL_NAME}"
+      exit 1
+    fi
+    cd $original_dir
+    PYTHONPATH=${PYTHONPATH} CMD=${CMD} run_model
+}
+
+
 # inceptionv4 model
 function inceptionv4() {
   # For accuracy, dataset location is required
@@ -644,6 +689,23 @@ function minigo() {
     fi
   else
     echo "MODE=${MODE} PRECISION=${PRECISION} is not supported for ${MODEL_NAME}"
+        exit 1
+  fi
+}
+# Mask R-CNN model
+function maskrcnn() {
+  if [ ${PRECISION} == "fp32" ]; then
+    original_dir=$(pwd)
+
+    if [ ${NOINSTALL} != "True" ]; then
+      # install dependencies
+      pip3 install -r ${MOUNT_BENCHMARK}/image_segmentation/tensorflow/maskrcnn/inference/fp32/requirements.txt 
+    fi
+    export PYTHONPATH=${PYTHONPATH}:${MOUNT_EXTERNAL_MODELS_SOURCE}:${MOUNT_EXTERNAL_MODELS_SOURCE}/mrcnn
+    CMD="${CMD} --data-location=${DATASET_LOCATION}"
+    PYTHONPATH=${PYTHONPATH} CMD=${CMD} run_model
+  else
+    echo "PRECISION=${PRECISION} is not supported for ${MODEL_NAME}"
     exit 1
   fi
 }
@@ -692,30 +754,16 @@ function mtcc() {
 
 # NCF model
 function ncf() {
-  if [[ -n "${clean}" ]]; then
-    CMD="${CMD} --clean"
-  fi
-
-  # NCF supports different datasets including ml-1m and ml-20m.
-  if [[ -n "${dataset}" && ${dataset} != "" ]]; then
-    CMD="${CMD} --dataset=${dataset}"
-  fi
-
-  if [[ -n "${te}" && ${te} != "" ]]; then
-    CMD="${CMD} -te=${te}"
-  fi
-
-  if [ ${PRECISION} == "fp32" -o ${PRECISION} == "bfloat16" ]; then
-    # For ncf, if dataset location is empty, script downloads dataset at given location.
+  if [ ${PRECISION} == "fp32" ]; then
+    # For nfc, if dataset location is empty, script downloads dataset at given location.
     if [ ! -d "${DATASET_LOCATION}" ]; then
-      mkdir -p ./dataset
-      CMD="${CMD} --data-location=./dataset"
+      mkdir -p /dataset
     fi
 
     export PYTHONPATH=${PYTHONPATH}:${MOUNT_EXTERNAL_MODELS_SOURCE}
 
     if [ ${NOINSTALL} != "True" ]; then
-      pip install -r ${MOUNT_EXTERNAL_MODELS_SOURCE}/official/requirements.txt
+      pip install -r ${MOUNT_BENCHMARK}/recommendation/tensorflow/ncf/inference/requirements.txt
     fi
 
     PYTHONPATH=${PYTHONPATH} CMD=${CMD} run_model
@@ -999,8 +1047,12 @@ function ssd_vgg16() {
 # UNet model
 function unet() {
   if [ ${PRECISION} == "fp32" ]; then
+    if [[ ${NOINSTALL} != "True" ]]; then
+      pip install -r "${MOUNT_BENCHMARK}/${USE_CASE}/${FRAMEWORK}/${MODEL_NAME}/requirements.txt"
+    fi
+
     if [[ -z "${checkpoint_name}" ]]; then
-      echo "wavenet requires -- checkpoint_name arg to be defined"
+      echo "UNet requires -- checkpoint_name arg to be defined"
       exit 1
     fi
     if [ ${ACCURACY_ONLY} == "True" ]; then
@@ -1116,7 +1168,7 @@ function wavenet() {
     export PYTHONPATH=${PYTHONPATH}:${MOUNT_EXTERNAL_MODELS_SOURCE}
 
     if [ ${NOINSTALL} != "True" ]; then
-      pip install -r ${MOUNT_EXTERNAL_MODELS_SOURCE}/requirements.txt
+      pip install librosa==0.5
     fi
 
     CMD="${CMD} --checkpoint_name=${checkpoint_name} \
@@ -1177,7 +1229,7 @@ function wide_deep_large_ds() {
     TCMALLOC_LIB="libtcmalloc.so.4"
     LIBTCMALLOC="$(ldconfig -p | grep $TCMALLOC_LIB | tr ' ' '\n' | grep /)"
 
-    if [[ -z "${LIBTCMALLOC}" ]]; then
+    if [[ -z $LIBTCMALLOC ]] && [[ $NOINSTALL != True ]]; then
       echo "libtcmalloc.so.4 not found, trying to install"
       apt-get update
       apt-get install google-perftools --fix-missing -y
@@ -1196,7 +1248,7 @@ function wide_deep_large_ds() {
       exit 1
     fi
     if [ ${MODE} == "training" ]; then
-      if [ ${steps} != None ]; then
+      if [[ ! -z $steps ]]; then
         CMD="${CMD} --steps=${steps}"
       fi
       if [ ${PRECISION} == "fp32" ]; then
@@ -1256,12 +1308,16 @@ elif [ ${MODEL_NAME} == "faster_rcnn" ]; then
   faster_rcnn
 elif [ ${MODEL_NAME} == "mlperf_gnmt" ]; then
   mlperf_gnmt
+elif [ ${MODEL_NAME} == "ncf" ]; then
+  ncf
 elif [ ${MODEL_NAME} == "inceptionv3" ]; then
   resnet101_inceptionv3
 elif [ ${MODEL_NAME} == "inceptionv4" ]; then
   inceptionv4
 elif [ ${MODEL_NAME} == "minigo" ]; then
   minigo
+elif [ ${MODEL_NAME} == "maskrcnn" ]; then
+  maskrcnn
 elif [ ${MODEL_NAME} == "mobilenet_v1" ]; then
   mobilenet_v1
 elif [ ${MODEL_NAME} == "resnet101" ]; then
@@ -1280,6 +1336,8 @@ elif [ ${MODEL_NAME} == "transformer_lt_official" ]; then
   transformer_lt_official
 elif [ ${MODEL_NAME} == "transformer_mlperf" ]; then
   transformer_mlperf
+elif [ ${MODEL_NAME} == "unet" ]; then
+  unet
 elif [ ${MODEL_NAME} == "wavenet" ]; then
   wavenet
 elif [ ${MODEL_NAME} == "wide_deep" ]; then

benchmarks/common/tensorflow_serving/start.sh

@@ -161,13 +161,13 @@ function resnet50v1_5(){
 
 function ssd_mobilenet(){
     # Install protofbuf and other requirement
-
-    pip install Cython \
-                contextlib2 \
-                pillow \
-                lxml \
-                absl-py \
-                tf_slim
+    pip install \
+        Cython \
+        absl-py \
+        contextlib2 \
+        lxml \
+        pillow>=7.1.0 \
+        tf_slim
 
     cd ${WORKSPACE}
     rm -rf tensorflow-models

benchmarks/image_recognition/tensorflow/densenet169/README.md

@@ -10,34 +10,17 @@ following modes/precisions:
 
     This step is required only for running accuracy, for running the model for performance we do not need to provide dataset.
 
-    Register and download the ImageNet dataset. Once you have the raw ImageNet dataset downloaded, we need to convert
-    it to the TFRecord format. The TensorFlow models repo provides
-    [scripts and instructions](https://github.com/tensorflow/models/tree/master/research/slim#an-automated-script-for-processing-imagenet-data)
-    to download, process and convert the ImageNet dataset to the TF records format. After converting data, you should have a directory
-    with the sharded dataset something like below, we only need `validation-*` files, discard `train-*` files:
-    ```
-    $ ll /home/myuser/datasets/ImageNet_TFRecords
-    -rw-r--r--. 1 user 143009929 Jun 20 14:53 train-00000-of-01024
-    -rw-r--r--. 1 user 144699468 Jun 20 14:53 train-00001-of-01024
-    -rw-r--r--. 1 user 138428833 Jun 20 14:53 train-00002-of-01024
-    ...
-    -rw-r--r--. 1 user 143137777 Jun 20 15:08 train-01022-of-01024
-    -rw-r--r--. 1 user 143315487 Jun 20 15:08 train-01023-of-01024
-    -rw-r--r--. 1 user  52223858 Jun 20 15:08 validation-00000-of-00128
-    -rw-r--r--. 1 user  51019711 Jun 20 15:08 validation-00001-of-00128
-    -rw-r--r--. 1 user  51520046 Jun 20 15:08 validation-00002-of-00128
-    ...
-    -rw-r--r--. 1 user  52508270 Jun 20 15:09 validation-00126-of-00128
-    -rw-r--r--. 1 user  55292089 Jun 20 15:09 validation-00127-of-00128
-    ```
+    Download and preprocess the ImageNet dataset using the [instructions here](/datasets/imagenet/README.md).
+    After running the conversion script you should have a directory with the
+    ImageNet dataset in the TF records format.
 
 2. Download the pretrained model:
    ```
    $ wget https://storage.googleapis.com/intel-optimized-tensorflow/models/v1_8/densenet169_fp32_pretrained_model.pb
    ```
 
 3. Clone the [intelai/models](https://github.com/intelai/models) repo
-    and then run the model scripts for either online or batch inference or accuracy. For --dataset-location in accuracy run, please use the ImageNet validation data path from step 1.
+    and then run the model scripts for either online or batch inference or accuracy. For --data-location in accuracy run, please use the ImageNet validation data path from step 1.
     Each model run has user configurable arguments separated from regular arguments by '--' at the end of the command.
     Unless configured, these arguments will run with default values. Below are the example codes for each use case:
 

benchmarks/image_recognition/tensorflow/inceptionv3/README.md

@@ -41,33 +41,9 @@ $ wget https://storage.googleapis.com/intel-optimized-tensorflow/models/v1_8/inc
 4. If you would like to run Inception V3 inference with real data or test for
 accuracy, you will need the ImageNet dataset.
 
-Register and download the
-[ImageNet dataset](http://image-net.org/download-images).
-
-Once you have the raw ImageNet dataset downloaded, we need to convert
-it to the TFRecord format. This is done using the
-[build_imagenet_data.py](https://github.com/tensorflow/models/blob/master/research/inception/inception/data/build_imagenet_data.py)
-script. There are instructions in the header of the script explaining
-its usage.
-
-After the script has completed, you should have a directory with the
-sharded dataset something like:
-
-```
-$ ll /home/<user>/datasets/ImageNet_TFRecords
--rw-r--r--. 1 user 143009929 Jun 20 14:53 train-00000-of-01024
--rw-r--r--. 1 user 144699468 Jun 20 14:53 train-00001-of-01024
--rw-r--r--. 1 user 138428833 Jun 20 14:53 train-00002-of-01024
-...
--rw-r--r--. 1 user 143137777 Jun 20 15:08 train-01022-of-01024
--rw-r--r--. 1 user 143315487 Jun 20 15:08 train-01023-of-01024
--rw-r--r--. 1 user  52223858 Jun 20 15:08 validation-00000-of-00128
--rw-r--r--. 1 user  51019711 Jun 20 15:08 validation-00001-of-00128
--rw-r--r--. 1 user  51520046 Jun 20 15:08 validation-00002-of-00128
-...
--rw-r--r--. 1 user  52508270 Jun 20 15:09 validation-00126-of-00128
--rw-r--r--. 1 user  55292089 Jun 20 15:09 validation-00127-of-00128
-```
+Download and preprocess the ImageNet dataset using the [instructions here](/datasets/imagenet/README.md).
+After running the conversion script you should have a directory with the
+ImageNet dataset in the TF records format.
 
 5. Next, navigate to the `benchmarks` directory in your local clone of
 the [intelai/models](https://github.com/IntelAI/models) repo from step 1.
@@ -230,11 +206,11 @@ $ wget https://storage.googleapis.com/intel-optimized-tensorflow/models/v1_8/inc
 ```
 
 3. If you would like to run Inception V3 FP32 inference and test for
-accuracy, you will need the ImageNet dataset. Running for online
-and batch inference do not require the ImageNet dataset. Instructions for
-downloading the dataset and converting it to the TF Records format can
-be found in the TensorFlow documentation
-[here](https://github.com/tensorflow/models/tree/master/research/slim#an-automated-script-for-processing-imagenet-data).
+accuracy, you will need the ImageNet dataset.
+
+Download and preprocess the ImageNet dataset using the [instructions here](/datasets/imagenet/README.md).
+After running the conversion script you should have a directory with the
+ImageNet dataset in the TF records format.
 
 4. Navigate to the `benchmarks` directory in your local clone of
 the [intelai/models](https://github.com/IntelAI/models) repo from step 1.