Mike at 14:45, 14 January 2020

2020-01-14T14:45:16Z

Mike: Formatted Benchmarking for NVIDIA Deep Learning

2020-01-14T14:44:25Z

Formatted Benchmarking for NVIDIA Deep Learning

New page

The QuickStart guide is 9 steps (out of which only 7 are focused on training) and the details can be found here: https://ngc.nvidia.com/catalog/model-scripts/nvidia:resnet_50_v1_5_for_mxnet/quickStartGuide

===== 1. Clone the repository. =====
<syntaxhighlight>
git clone https://github.com/NVIDIA/DeepLearningExamples
cd DeepLearningExamples/MxNet/Classification/RN50v1.5
</syntaxhighlight>
===== 2. Build the ResNet50 MXNet NGC container. =====
After Docker is setup, you can build the ResNet50 image with:
<syntaxhighlight>
docker build . -t nvidia_rn50_mx
</syntaxhighlight>
===== 3. Start an interactive session in the NGC container to run preprocessing/training/inference. =====
<syntaxhighlight>
nvidia-docker run --rm -it --ipc=host <path to dataset>:/data/imagenet/train-val-recordio-passthrough nvidia_rn50_mx
</syntaxhighlight>
===== 4. Download and preprocess the data. =====
• Download the images from http://image-net.org/download-images.

• Extract the training and validation data:
<syntaxhighlight>
mkdir train && mv ILSVRC2012_img_train.tar train/ && cd train
tar -xvf ILSVRC2012_img_train.tar && rm -f ILSVRC2012_img_train.tar
find . -name "*.tar" | while read NAME ; do mkdir -p "${NAME%.tar}"; tar -xvf "${NAME}" -C "${NAME%.tar}"; rm -f "${NAME}"; done
cd ..
</syntaxhighlight>
===== 5. Extract the validation data and move the images to subfolders. ======
<syntaxhighlight>
mkdir val && mv ILSVRC2012_img_val.tar val/ && cd val && tar -xvf ILSVRC2012_img_val.tar
wget -qO- https://raw.githubusercontent.com/soumith/imagenetloader.torch/master/valprep.sh | bash
</syntaxhighlight>
===== 6. Preprocess the dataset. =====
<syntaxhighlight>
./scripts/prepare_imagenet.sh <path to raw imagenet> <path where processed dataset will be created>
</syntaxhighlight>

===== 7. Start training. =====
<syntaxhighlight>
./runner -n <number of gpus> -b <batch size per GPU (default 192)>
</syntaxhighlight>

@@ Line 1: / Line 1: @@
 The QuickStart guide is 9 steps (out of which only 7 are focused on training) and the details can be found here: https://ngc.nvidia.com/catalog/model-scripts/nvidia:resnet_50_v1_5_for_mxnet/quickStartGuide
-===== 1. Clone the repository. =====
+===== Clone the repository. =====
 <syntaxhighlight>
 git clone https://github.com/NVIDIA/DeepLearningExamples
 cd DeepLearningExamples/MxNet/Classification/RN50v1.5
 </syntaxhighlight>
-===== 2. Build the ResNet50 MXNet NGC container. =====
+===== Build the ResNet50 MXNet NGC container. =====
 After Docker is setup, you can build the ResNet50 image with:
 <syntaxhighlight>
 docker build . -t nvidia_rn50_mx
 </syntaxhighlight>
-===== 3. Start an interactive session in the NGC container to run preprocessing/training/inference. =====
+===== Start an interactive session in the NGC container to run preprocessing/training/inference. =====
 <syntaxhighlight>
 nvidia-docker run --rm -it --ipc=host <path to dataset>:/data/imagenet/train-val-recordio-passthrough nvidia_rn50_mx
 </syntaxhighlight>
-===== 4. Download and preprocess the data. =====
+===== Download and preprocess the data. =====
 •	Download the images from http://image-net.org/download-images.
@@ Line 25: / Line 25: @@
    cd ..
 </syntaxhighlight>
-===== 5. Extract the validation data and move the images to subfolders. ======
+===== Extract the validation data and move the images to subfolders. =====
 <syntaxhighlight>
 mkdir val && mv ILSVRC2012_img_val.tar val/ && cd val && tar -xvf ILSVRC2012_img_val.tar
 wget -qO- https://raw.githubusercontent.com/soumith/imagenetloader.torch/master/valprep.sh | bash
 </syntaxhighlight>
-===== 6. Preprocess the dataset. =====
+===== Preprocess the dataset. =====
 <syntaxhighlight>
 ./scripts/prepare_imagenet.sh <path to raw imagenet> <path where processed dataset will be created>
 </syntaxhighlight>
-===== 7. Start training. =====
+===== Start training. =====
 <syntaxhighlight>
 ./runner -n <number of gpus> -b <batch size per GPU (default 192)>
 </syntaxhighlight>

Benchmarking: Deep Learning - Revision history

Mike at 14:45, 14 January 2020

Mike: Formatted Benchmarking for NVIDIA Deep Learning