Skip to content

MC-DARTS : Model Size Constrained Differentiable Architecture Search

License

Notifications You must be signed in to change notification settings

itigo-11111/MC-DARTS

Repository files navigation

MC-DARTS

MC-DARTS (Model Size Constrained Differentiable Architecture Search) adding constraints to DARTS to search for a network architecture considering the accuracy and the model size.

MC-DARTS has been accepted at NeurIPS 2022 Has it Trained Yet? A Workshop for Algorithmic Efficiency in Practical Neural Network Training! MC-DARTS : Model Size Constrained Differentiable Architecture Search

This code was created based on the DARTS. New branches for loss and update formulas have been added (Detailed modifications ).

Requirement ( my setting version)

Python (3.7)

Numpy (1.21.5)

Pytorch (1.5.0)

tqdm (4.64.1)

torchsummary (1.5.1)

python-graphviz (0.20)

Usage

Architecture search (CIFAR10)

python train_search.py --limit_param 2500000

The paper prepares it in advance, but this code downloads the dataset so that it can be run immediately.

Architecture evaluation ( arch => use architecture)

python train.py --arch "mc_darts2900000"

Network Architecture (Searched)

See genotypes.py

Licence

apache license 2.0

Citation

If you use our code, please cite our paper.

@inproceedings{
hemmi2022mcdarts,
title={{MC}-{DARTS} : Model Size Constrained Differentiable Architecture Search},
author={Kazuki Hemmi and Yuki Tanigaki and Masaki Onishi},
booktitle={Has it Trained Yet? NeurIPS 2022 Workshop},
year={2022},
url={https://openreview.net/forum?id=jKJ6OcvqdQ}
}

Releases

No releases published

Packages

No packages published

Languages