1 paper
Qinqin Zhou, Fuhai Chen, Jipeng Wu +3
Training-free neural architecture search promises efficient discovery of high-performance networks without costly training. However, existing zero-cost proxies rely on fragmented h…