vibhatha / AISC-Benchmarks-PyTorch

DNN training benchmarks in PyTorch

AISC Training Benchmarks - PyTorch

This repository contains the scripts and setup instructions for training models.

Setup

1. Install PyTorch and other utilities

Install Git, VIM, Python, PyTorch and other required packages with the following commands.

# run the following two commands for LCOW only
apt-get update
apt-get install -y sudo

# run the following commands for both Azure VM and LCOW
sudo apt-get install -y git
sudo apt-get install -y vim
sudo apt-get -y install python3
sudo apt-get -y install python3-pip
python3 -m pip install torch
python3 -m pip install torchvision
python3 -m pip install tensorboardX

2. Clone the demo repository

git clone https://github.com/rimman/AISC-Benchmarks-PyTorch.git
cd AISC-Benchmarks-PyTorch

PyTorch Models

1. MNIST

See the Readme for description and execution.

2. Fashion MNIST

See the Readme for description and execution.

3. Image Classifier

See the Readme for description and execution.

4. ImageNet

Models: alexnet, densenet121, densenet161, densenet169, densenet201, resnet101, resnet152, resnet18, resnet34, resnet50, squeezenet1_0, squeezenet1_1, vgg11, vgg11_bn, vgg13, vgg13_bn, vgg16, vgg16_bn, vgg19

See the Readme for description and execution.

5. Toy

See the Readme for description and execution.

6. Distributed MNIST

See the Readme for description and execution.

7. PyramidNet

See the Readme for description and execution.

8. Word-level language RNN

See the Readme for description and execution.

MLPerf Benchmark

This directory contains reference implementations for the MLPerf benchmark. There are implementations for each of the 7 benchmarks in the MLPerf suite.

• image_classification - Resnet-50 v1 applied to Imagenet.
• object_detection - Mask R-CNN applied to COCO.
• single_stage_detector - SSD applied to COCO 2017.
• speech_recognition - DeepSpeech2 applied to Librispeech.
• translation - Transformer applied to WMT English-German.
• recommendation - Neural Collaborative Filtering applied to MovieLens 20 Million (ml-20m).
• sentiment_analysis - Seq-CNN applied to IMDB dataset.
• reinforcement - Mini-go applied to predicting pro game moves.
• Each reference implementation provides the following:

1. image_classification

See the Readme for description and execution.

2. object_detection

See the Readme for description and execution.

3. single_stage_detector

See the Readme for description and execution.

4. speech_recognition

See the Readme for description and execution.

5. translation

See the Readme for description and execution.