MixPath: A Unified Approach for One-shot Neural Architecture Search

This repo provides the supernet of S₁ and our confirmatory experiments on NAS-Bench-101.

Requirements

Python >= 3.6, Pytorch >= 1.0.0, torchvision >= 0.2.0

pip install -r requirement.txt

Datasets

CIFAR-10 can be automatically downloaded by torchvision. It has 50,000 images for training and 10,000 images for validation.

Usage

python S1/train_search.py \
    --exp_name experiment_name \
    --m number_of_paths[1,2,3,4]
    --data_dir /path/to/dataset \
    --seed 2020 \

python NasBench101/nas_train_search.py \
    --exp_name experiment_name \
    --m number_of_paths[1,2,3,4]
    --data_dir /path/to/dataset \
    --seed 2020 \

python S1/train_search.py --exp_name test --data_dir ./data/ --m 4 --seed 2020

Citation

@article{chu2020mixpath,
  title={MixPath: A Unified Approach for One-shot Neural Architecture Search},
  author={Chu, Xiangxiang and Li, Xudong and Lu, Yi and Zhang, Bo and Li, Jixiang},
  journal={arXiv preprint arXiv:2001.05887},
  url={https://arxiv.org/abs/2001.05887},
  year={2020}
}