Awesome Deep Model Compression

A useful list of Deep Model Compression related research papers, articles, tutorials, libraries, tools and more.
Currently the Repos are additional given tags either [Pytorch/TF]. To quickly find hands-on Repos in your commonly used framework, please Ctrl+F to get start 😃

Contents

• Papers
  • General
  • Architecture
  • Quantization
  • Binarization
  • Pruning
  • Distillation
  • Low Rank Approximation
• Articles
  • Blogs
• Tools
  • Libraries
  • Cross Platform
  • Hard-ware Integration

---

Papers

General

Architecture

Quantization

Binarization

Pruning

• The Lottery Ticket Hypothesis | ICLR, 2019, Google | Paper | Code

Distillation

Low Rank Approximation

---

Articles

Blogs

• Pruning deep neural networks to make them fast and small [Pytorch] By using pruning a VGG-16 based Dogs-vs-Cats classifier is made x3 faster and x4 smaller.
• All The Ways You Can Compress BERT - An overview of different compression methods for large NLP models (BERT) based on different characteristics and compares their results.
• Deep Learning Model Compression methods.
• Do We Really Need Model Compression in the future?

---

Tools

Libraries

• torch.nn.utils.prune [Pytorch]
  Pytorch official supported sparsify neural networks and custom pruning technique.
• Neural Network Intelligence[Pytorch/TF]
  There are some popular model compression algorithms built-in in NNI. Users could further use NNI’s auto tuning power to find the best compressed model, which is detailed in Auto Model Compression.
• Condensa [Pytorch]
  A Programming System for Neural Network Compression. | paper
• IntelLabs distiller [Pytorch]
  Neural Network Distiller by Intel AI Lab: a Python package for neural network compression research. | Documentation
• Torch-Pruning[Pytorch]
  A pytorch toolkit for structured neural network pruning and layer dependency.
• CompressAI [Pytorch]
  A PyTorch library and evaluation platform for end-to-end compression research.
• Model Compression[Pytorch]
  A onestop pytorch model compression repo. | Reposhub
• TensorFlow Model Optimization Toolkit [TF]
  Accompanied blog post, TensorFlow Model Optimization Toolkit — Pruning API
• XNNPACK
  XNNPACK is a highly optimized library of floating-point neural network inference operators for ARM, WebAssembly, and x86 (SSE2 level) platforms. It's a based on QNNPACK library. However, unlike QNNPACK, XNNPACK focuses entirely on floating-point operators.

Cross Platform

• Loading a TorchScript Model in C++ [Pytorch x C++]
  From an existing Python model to a serialized representation that can be loaded and executed purely from C++, with no dependency on Python.
• Open Neural Network Exchange (ONNX)[Pytorch, TF, Keras...etc]
  An open standard format for representing machine learning models.

Hard-ware Integration

• TensorRT (NVIDIA) [Pytorch, TF,Keras ...etc]
  
  
  

  • torch2trt [Pytorch]
    An easy to use PyTorch to TensorRT converter
  • How to Convert a Model from PyTorch to TensorRT and Speed Up Inference [Pytorch]

• TVM (Apache) [Pytorch, TF]
  Open deep learning compiler stack for cpu, gpu and specialized accelerators

• Pytorch Glow
  Glow is a machine learning compiler and execution engine for hardware accelerators. It is designed to be used as a backend for high-level machine learning frameworks. The compiler is designed to allow state of the art compiler optimizations and code generation of neural network graphs.

• CoreML (Apple) [Pytorch,TF,Keras,SKLearn ...etc]
  Core ML provides a unified representation for all models. Your app uses Core ML APIs and user data to make predictions, and to train or fine-tune models, all on the user’s device.
  Introduction

• Tensorflow Lite (Google) [TF]
  An open source deep learning framework for on-device inference.

• Intel Nervana Neon [Deprecated]