This repository contains the implementation of the research paper titled "ZiCo: Zero-shot NAS via Inverse Coefficient of Variation on Gradients," presented at ICLR 2023.
Neural Architecture Search (NAS) has become a crucial aspect of designing effective neural networks. This repository focuses on the research presented in the paper "ZiCo: Zero-shot NAS via Inverse Coefficient of Variation on Gradients" by Guihong Li, Yuedong Yang, Kartikeya Bhardwaj, and Radu Marculescu. The work explores novel methods for NAS, and the experiments are conducted on the NAS-Bench-201 dataset.
For more details, please refer to the original paper: ZiCo: Zero-shot NAS via Inverse Coefficient of Variation on Gradients.
If you find this work useful in your research, please consider citing the following papers:
@article{li2023zico,
title={ZiCo: Zero-shot NAS via Inverse Coefficient of Variation on Gradients},
author={Li, Guihong and Yang, Yuedong and Bhardwaj, Kartikeya and Marculescu, Radu},
journal={arXiv preprint arXiv:2301.11300},
year={2023}
}
@article{dong2020bench,
title={Nas-bench-201: Extending the scope of reproducible neural architecture search},
author={Dong, Xuanyi and Yang, Yi},
journal={arXiv preprint arXiv:2001.00326},
year={2020}
}