Skip to content

Intel® Neural Compressor (formerly known as Intel® Low Precision Optimization Tool), targeting to provide unified APIs for network compression technologies, such as low precision quantization, sparsity, pruning, knowledge distillation, across different deep learning frameworks to pursue optimal inference performance.

License

Notifications You must be signed in to change notification settings

xin3he/neural-compressor

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Introduction to Intel® Neural Compressor

Intel® Neural Compressor (formerly known as Intel® Low Precision Optimization Tool) is an open-source Python library running on Intel CPUs and GPUs, which delivers unified interfaces across multiple deep learning frameworks for popular network compression technologies, such as quantization, pruning, knowledge distillation. This tool supports automatic accuracy-driven tuning strategies to help user quickly find out the best quantized model. It also implements different weight pruning algorithms to generate pruned model with predefined sparsity goal and supports knowledge distillation to distill the knowledge from the teacher model to the student model.

Note

GPU support is under development.

Visit the Intel® Neural Compressor online document website at: https://intel.github.io/neural-compressor.

Infrastructure

Intel® Neural Compressor features an architecture and workflow that aids in increasing performance and faster deployments across infrastructures.

Architecture

Architecture

Click the image to enlarge it.

Workflow

Workflow

Click the image to enlarge it.

Supported Frameworks

Supported deep learning frameworks are:

Note: Intel Optimized TensorFlow 2.5.0 requires to set environment variable TF_ENABLE_MKL_NATIVE_FORMAT=0 before running Neural Compressor quantization or deploying the quantized model.

Note: From the official TensorFlow 2.6.0, oneDNN support has been upstreamed. Download the official TensorFlow 2.6.0 binary for the CPU device and set the environment variable TF_ENABLE_ONEDNN_OPTS=1 before running the quantization process or deploying the quantized model.

Installation

Select the installation based on your operating system.

Linux Installation

You can install Neural Compressor using one of three options: Install just the library from binary or source, or get the Intel-optimized framework together with the library by installing the Intel® oneAPI AI Analytics Toolkit.

Prerequisites

The following prerequisites and requirements must be satisfied for a successful installation:

  • Python version: 3.7 or 3.8 or 3.9

  • C++ compiler: 7.2.1 or above

  • CMake: 3.12 or above

common build issues

Issue 1: ValueError: numpy.ndarray size changed, may indicate binary incompatibility. Expected 88 from C header, got 80 from PyObject

Solution: reinstall pycocotools by "pip install pycocotools --no-cache-dir"

Issue 2: ImportError: libGL.so.1: cannot open shared object file: No such file or directory

Solution: apt install or yum install opencv

Option 1 Install from binary

# install stable version from pip
pip install neural-compressor

# install nightly version from pip
pip install -i https://test.pypi.org/simple/ neural-compressor

# install stable version from from conda
conda install neural-compressor -c conda-forge -c intel 

Option 2 Install from source

git clone https://github.com/intel/neural-compressor.git
cd neural-compressor
git submodule sync
git submodule update --init --recursive
pip install -r requirements.txt
python setup.py install

Option 3 Install from AI Kit

The Intel® Neural Compressor library is released as part of the Intel® oneAPI AI Analytics Toolkit (AI Kit). The AI Kit provides a consolidated package of Intel's latest deep learning and machine optimizations all in one place for ease of development. Along with Neural Compressor, the AI Kit includes Intel-optimized versions of deep learning frameworks (such as TensorFlow and PyTorch) and high-performing Python libraries to streamline end-to-end data science and AI workflows on Intel architectures.

The AI Kit is distributed through many common channels, including from Intel's website, YUM, APT, Anaconda, and more. Select and download the AI Kit distribution package that's best suited for you and follow the Get Started Guide for post-installation instructions.

Download AI Kit AI Kit Get Started Guide

Windows Installation

Prerequisites

The following prerequisites and requirements must be satisfied for a successful installation:

  • Python version: 3.7 or 3.8 or 3.9

  • Download and install anaconda.

  • Create a virtual environment named nc in anaconda:

    # Here we install python 3.7 for instance. You can also choose python 3.8 or 3.9.
    conda create -n nc python=3.7
    conda activate nc

Installation options

Option 1 Install from binary

# install stable version from pip
pip install neural-compressor

# install nightly version from pip
pip install -i https://test.pypi.org/simple/ neural-compressor

# install from conda
conda install pycocotools -c esri   
conda install neural-compressor -c conda-forge -c intel

Option 2 Install from source

git clone https://github.com/intel/neural-compressor.git
cd neural-compressor
git submodule sync
git submodule update --init --recursive
pip install -r requirements.txt
python setup.py install

Documentation

Get Started

  • APIs explains Intel® Neural Compressor's API.
  • GUI provides web-based UI service to make quantization easier.
  • Transform introduces how to utilize Neural Compressor's built-in data processing and how to develop a custom data processing method.
  • Dataset introduces how to utilize Neural Compressor's built-in dataset and how to develop a custom dataset.
  • Metric introduces how to utilize Neural Compressor's built-in metrics and how to develop a custom metric.
  • Tutorial provides comprehensive instructions on how to utilize Neural Compressor's features with examples.
  • Examples are provided to demonstrate the usage of Neural Compressor in different frameworks: TensorFlow, PyTorch, MXNet, and ONNX Runtime.
  • Intel oneAPI AI Analytics Toolkit Get Started Guide explains the AI Kit components, installation and configuration guides, and instructions for building and running sample apps.
  • AI and Analytics Samples includes code samples for Intel oneAPI libraries.

Deep Dive

  • Quantization are processes that enable inference and training by performing computations at low-precision data types, such as fixed-point integers. Neural Compressor supports Post-Training Quantization (PTQ) with different quantization capabilities and Quantization-Aware Training (QAT). Note that (Dynamic Quantization) currently has limited support.
  • Pruning provides a common method for introducing sparsity in weights and activations.
  • Knowledge Distillation provides a common method for distilling knowledge from teacher model to student model.
  • Distributed Training introduces how to leverage Horovod to do multi-node training in Intel® Neural Compressor to speed up the training time.
  • Benchmarking introduces how to utilize the benchmark interface of Neural Compressor.
  • Mixed precision introduces how to enable mixed precision, including BFP16 and int8 and FP32, on Intel platforms during tuning.
  • Graph Optimization introduces how to enable graph optimization for FP32 and auto-mixed precision.
  • Model Conversion introduces how to convert TensorFlow QAT model to quantized model running on Intel platforms.
  • TensorBoard provides tensor histograms and execution graphs for tuning debugging purposes.

Advanced Topics

  • Execution Engine is a bare metal solution domain-specific NLP models as the reference for customers.
  • Adaptor is the interface between components and framework. The method to develop adaptor extension is introduced with ONNX Runtime as example.
  • Strategy can automatically optimized low-precision recipes for deep learning models to achieve optimal product objectives like inference performance and memory usage with expected accuracy criteria. The method to develop a new strategy is introduced.

Publications

Full publication list please refers to here

System Requirements

Intel® Neural Compressor supports systems based on Intel 64 architecture or compatible processors, specially optimized for the following CPUs:

  • Intel Xeon Scalable processor (formerly Skylake, Cascade Lake, Cooper Lake, and Icelake)
  • future Intel Xeon Scalable processor (code name Sapphire Rapids)

Intel® Neural Compressor requires installing the Intel-optimized framework version for the supported DL framework you use: TensorFlow, PyTorch, MXNet, or ONNX runtime.

Note: Intel Neural Compressor supports Intel-optimized and official frameworks for some TensorFlow versions. Refer to Supported Frameworks for specifics.

Validated Hardware/Software Environment

Platform OS Python Framework Version
Cascade Lake

Cooper Lake

Skylake

Ice Lake
CentOS 8.3

Ubuntu 18.04
3.7

3.8

3.9
TensorFlow 2.8.0
2.7.0
2.6.2
1.15.0UP3
PyTorch 1.10.0+cpu
1.9.0+cpu
1.8.0+cpu
IPEX
MXNet 1.8.0
1.7.0
1.6.0
ONNX Runtime 1.10.0
1.9.0
1.8.0

Validated Models

Intel® Neural Compressor provides numerous examples to show promising accuracy loss with the best performance gain. A full quantized model list on various frameworks is available in the Model List.

Validated MLPerf Models

Model Framework Support Example
ResNet50 v1.5 TensorFlow Yes Link
PyTorch Yes Link
DLRM PyTorch Yes Link
BERT-large TensorFlow Yes Link
PyTorch Yes Link
SSD-ResNet34 TensorFlow Yes Link
PyTorch Yes Link
RNN-T PyTorch Yes Link
3D-UNet TensorFlow WIP
PyTorch Yes Link

Validated Quantized Models

Framework version model Accuracy Performance/ICX8380/1s4c10ins1bs/throughput(samples/sec)
INT8 FP32 Acc Ratio[(INT8-FP32)/FP32] INT8 FP32 Performance Ratio[INT8/FP32]
tensorflow 2.6.0 resnet50v1.0 74.11% 74.27% -0.22% 1287.00 495.29 2.60x
tensorflow 2.6.0 resnet50v1.5 76.82% 76.46% 0.47% 1218.03 420.34 2.90x
tensorflow 2.6.0 resnet101 77.50% 76.45% 1.37% 849.62 345.54 2.46x
tensorflow 2.6.0 inception_v1 70.48% 69.74% 1.06% 2202.64 1058.20 2.08x
tensorflow 2.6.0 inception_v2 74.36% 73.97% 0.53% 1751.31 827.81 2.11x
tensorflow 2.6.0 inception_v3 77.28% 76.75% 0.69% 868.06 384.17 2.26x
tensorflow 2.6.0 inception_v4 80.40% 80.27% 0.16% 569.48 197.28 2.89x
tensorflow 2.6.0 inception_resnet_v2 80.44% 80.40% 0.05% 269.03 137.25 1.96x
tensorflow 2.6.0 mobilenetv1 71.79% 70.96% 1.17% 3831.42 1189.06 3.22x
tensorflow 2.6.0 mobilenetv2 71.79% 71.76% 0.04% 2570.69 1237.62 2.07x
tensorflow 2.6.0 ssd_resnet50_v1 37.86% 38.00% -0.37% 65.52 24.01 2.73x
tensorflow 2.6.0 ssd_mobilenet_v1 22.97% 23.13% -0.69% 842.46 404.04 2.08x
tensorflow 2.6.0 ssd_resnet34 21.69% 22.09% -1.81% 41.23 10.75 3.83x
Framework version model Accuracy Performance/ICX8380/1s4c10ins1bs/throughput(samples/sec)
INT8 FP32 Acc Ratio[(INT8-FP32)/FP32] INT8 FP32 Performance Ratio[INT8/FP32]
pytorch 1.9.0+cpu resnet18 69.59% 69.76% -0.24% 692.04 363.64 1.90x
pytorch 1.9.0+cpu resnet50 76.00% 76.13% -0.17% 453.10 186.67 2.43x
pytorch 1.9.0+cpu resnext101_32x8d 79.02% 79.31% -0.36% 196.27 70.08 2.80x
pytorch 1.9.0+cpu bert_base_mrpc 88.12% 88.73% -0.69% 199.32 107.34 1.86x
pytorch 1.9.0+cpu bert_base_cola 59.06% 58.84% 0.37% 198.53 105.29 1.89x
pytorch 1.9.0+cpu bert_base_sts-b 88.72% 89.27% -0.62% 203.29 107.03 1.90x
pytorch 1.9.0+cpu bert_base_sst-2 91.74% 91.86% -0.13% 197.86 105.31 1.88x
pytorch 1.9.0+cpu bert_base_rte 70.40% 69.68% 1.04% 192.90 107.25 1.80x
pytorch 1.9.0+cpu bert_large_mrpc 87.66% 88.33% -0.75% 94.08 33.84 2.78x
pytorch 1.9.0+cpu bert_large_squad 92.69 93.05 -0.38% 20.93 11.18 1.87x
pytorch 1.9.0+cpu bert_large_qnli 91.12% 91.82% -0.76% 93.75 33.73 2.78x
pytorch 1.9.0+cpu bert_large_rte 72.20% 72.56% -0.50% 52.80 33.62 1.57x
pytorch 1.9.0+cpu bert_large_cola 62.07% 62.57% -0.80% 94.97 33.77 2.81x
pytorch 1.9.0+cpu inception_v3 69.48% 69.54% -0.09% 418.59 207.77 2.01x
pytorch 1.9.0+cpu peleenet 71.61% 72.08% -0.66% 461.47 359.58 1.28x
pytorch 1.9.0+cpu yolo_v3 24.50% 24.54% -0.17% 98.11 37.50 2.62x

Validated Pruning Models

Tasks FWK Model fp32 baseline gradient sensitivity with 20% sparsity +onnx dynamic quantization on pruned model
accuracy% drop% perf gain (sample/s) accuracy% drop% perf gain (sample/s)
SST-2 pytorch bert-base accuracy = 92.32 accuracy = 91.97 -0.38 1.30x accuracy = 92.20 -0.13 1.86x
QQP pytorch bert-base [accuracy, f1] = [91.10, 88.05] [accuracy, f1] = [89.97, 86.54] [-1.24, -1.71] 1.32x [accuracy, f1] = [89.75, 86.60] [-1.48, -1.65] 1.81x
Tasks FWK Model fp32 baseline Pattern Lock on 70% Unstructured Sparsity Pattern Lock on 50% 1:2 Structured Sparsity
accuracy% drop% accuracy% drop%
MNLI pytorch bert-base [m, mm] = [84.57, 84.79] [m, mm] = [82.45, 83.27] [-2.51, -1.80] [m, mm] = [83.20, 84.11] [-1.62, -0.80]
SST-2 pytorch bert-base accuracy = 92.32 accuracy = 91.51 -0.88 accuracy = 92.20 -0.13
QQP pytorch bert-base [accuracy, f1] = [91.10, 88.05] [accuracy, f1] = [90.48, 87.06] [-0.68, -1.12] [accuracy, f1] = [90.92, 87.78] [-0.20, -0.31]
QNLI pytorch bert-base accuracy = 91.54 accuracy = 90.39 -1.26 accuracy = 90.87 -0.73
QnA pytorch bert-base [em, f1] = [79.34, 87.10] [em, f1] = [77.27, 85.75] [-2.61, -1.54] [em, f1] = [78.03, 86.50] [-1.65, -0.69]
Framework Model fp32 baseline Compression dataset acc(drop)%
Pytorch resnet18 69.76 30% sparsity on magnitude ImageNet 69.47(-0.42)
Pytorch resnet18 69.76 30% sparsity on gradient sensitivity ImageNet 68.85(-1.30)
Pytorch resnet50 76.13 30% sparsity on magnitude ImageNet 76.11(-0.03)
Pytorch resnet50 76.13 30% sparsity on magnitude and post training quantization ImageNet 76.01(-0.16)
Pytorch resnet50 76.13 30% sparsity on magnitude and quantization aware training ImageNet 75.90(-0.30)

Validated Knowledge Distillation Examples

Example Name Dataset Student
(Accuracy)
Teacher
(Accuracy)
Student With Distillation
(Accuracy Improvement)
ResNet example ImageNet ResNet18
(0.6739)
ResNet50
(0.7399)
0.6845
(0.0106)
BlendCnn example MRPC BlendCnn
(0.7034)
BERT-Base
(0.8382)
0.7034
(0)
BiLSTM example SST-2 BiLSTM
(0.7913)
RoBERTa-Base
(0.9404)
0.8085
(0.0172)

Validated Engine Examples

model Accuracy Performance/ICX8380/1s4c10ins1bs/seq_len128/throughput(samples/sec) Performance/ICX8380/2s4c20ins64bs/seq_len128/throughput(samples/sec)
INT8 FP32 Acc   Ratio[(INT8-FP32)/FP32] INT8 FP32 Preformance   Ratio[INT8/FP32] INT8 FP32 Preformance   Ratio[INT8/FP32]
bert_large_squad 90.74 90.87 -0.14% 44.9 12.33 3.64x 362.21 88.38 4.10x
distilbert_base_uncased_sst2 90.14% 90.25% -0.12% 1003.01 283.69 3.54x 2104.26 606.58 3.47x
minilm_l6_h384_uncased_sst2 89.33% 90.14% -0.90% 2739.73 999 2.74x 5389.98 2333.14 2.31x
roberta_base_mrpc 89.46% 88.97% 0.55% 506.07 142.13 3.56x 1167.09 311.5 3.75x
bert_base_nli_mean_tokens_stsb 89.27% 89.55% -0.31% 503.52 140.98 3.57x 1096.46 332.54 3.30x
bert_base_sparse_mrpc 70.34% 70.59% -0.35% 506.59 142.33 3.56x 1133.04 339.96 3.33x
distilroberta_base_wnli 56.34% 56.34% 0.00% 1026.69 290.7 3.53x 2309.9 620.81 3.72x
paraphrase_xlm_r_multilingual_v1_stsb 86.72% 87.23% -0.58% 509.68 142.73 3.57x 1169.45 311.59 3.75x
distilbert_base_uncased_mrpc 84.07% 84.07% 0.00% 1002 280.27 3.58x 2107.96 606.95 3.47x
finbert_financial_phrasebank 82.74% 82.80% -0.07% 919.12 272.48 3.37x 1101.13 331.88 3.32x
distilbert_base_uncased_emotion 93.85% 94.20% -0.37% 1003.01 283.53 3.54x 2103.22 607.08 3.46x

Additional Content

Hiring

We are hiring. Please send your resume to [email protected] if you have interests in model compression techniques.

About

Intel® Neural Compressor (formerly known as Intel® Low Precision Optimization Tool), targeting to provide unified APIs for network compression technologies, such as low precision quantization, sparsity, pruning, knowledge distillation, across different deep learning frameworks to pursue optimal inference performance.

Resources

License

Code of conduct

Stars

Watchers

Forks

Packages

No packages published

Languages

  • Python 82.4%
  • C++ 10.4%
  • TypeScript 3.7%
  • HTML 2.6%
  • SCSS 0.7%
  • CMake 0.2%