deJQK / FracBits

Neural Network Quantization With Fractional Bit-widths

Network Quantizing with Fractional Bitwidths

This is the codebase of searching bit-widths for neural networks based on fractional bit-widths, as proposed in FracBits arXiv AAAI2021.

Run

0. Requirements:
   • python3, pytorch 1.0, torchvision 0.2.1, pyyaml 3.13.
   • Prepare ImageNet-1k data following pytorch example.
1. Training and Testing:
   • The codebase is a general ImageNet training framework using yaml config under apps dir, based on PyTorch.
   • To test, download pretrained models to logs dir and directly run command.
   • To train, comment test_only and pretrained in config file. You will need to manage visible gpus by yourself.
   • Command: python train.py app:{apps/***.yml}. {apps/***.yml} is config file. Do not miss app: prefix.
2. Still have questions?
   • If you still have questions, please search closed issues first. If the problem is not solved, please open a new.

Technical Details

Implementing network quantizing is straightforward:

• Quantization layers are implemented in models/quantizable_ops.
• Training with quantizing is implemented by setting a reasonable [kappa] in the yml file.
• [q_mobilenetv1_uint8_train_val.yml] is a good start yml example. For ablation test, please run test_ablation with the corresponding test ablation yml file.

License

CC 4.0 Attribution-NonCommercial International

The software is for educaitonal and academic research purpose only.

Citing

@article{yang2020fracbits,
  title={FracBits: Mixed Precision Quantization via Fractional Bit-Widths},
  author={Yang, Linjie and Jin, Qing},
  journal={arXiv preprint arXiv:2007.02017},
  year={2020}
}