StarLight is a general framework, which provides various algorithms for Neural Architecture Search (NAS), Model Pruning, and Model Quantization.

	Frameworks & Libraries	Algorithms
Built-in	Supported Frameworks PyTorch Keras TensorFlow MXNet Caffe2 More...	Neural Architecture Search Classification DARTS P-DARTS PC-DARTS R-DARTS S-DARTS SGAS SNAS GDAS NASP Semantic egmentation Auto-DeepLab FasterSeg Object Detection xxx xxx Model Compression Pruning AGP Pruner Slim Pruner FPGM Pruner NetAdapt Pruner SimulatedAnnealing Pruner ADMM Pruner AutoCompress Pruner Quantization QAT Quantizer DoReFa Quantizer Multi-Task TrackRCNN MTL Prune

• document - 文档信息
• html - 针对各自模块添加对应的html文件
• modelcompress - 模型压缩对应的代码文件夹
• nas - 模型搜索对应的代码文件夹
• qtui - qt对应的ui文件
• tools - 共用的一些小函数

• python >= 3.6

• pyqt5==5.12, simply use pip to for installation:

    pip install pyqt5==5.12
    pip install PyQtWebEngine==5.12	

1. Example for model compression

    cd modelcompress
    python modelcomp.py

2. Use Qt Creator to edit UI file:

• Installation for Qt Creator, here we take Ubuntu as an example:

    sudo apt install qtcreator

• For other systems, you can download the suitable version of Qt Creator from this link: http://download.qt.io/archive/qtcreator/4.4/4.4.1/

• After editing the UI file, you can get its corresponding Python code by running:

    # first arg is PY file to be converted and the second is the UI file
    pyuic5 -o modelcomp_ui.py modelcompress.ui