Hi I want to reproduce the results mentioned in the paper. So after training the SuperNet, how do you get the evaluation results that you described in the paper for different MAC targets because as I see in the run_NAS.py file, it takes constraints based on latency and not MACs.
Also, you have mentioned "supplementary" material in the paper. Can you provide that so that I can take some help for reproducing the results?
Thanks!
Hi I want to reproduce the results mentioned in the paper. So after training the SuperNet, how do you get the evaluation results that you described in the paper for different MAC targets because as I see in the run_NAS.py file, it takes constraints based on latency and not MACs.
Also, you have mentioned "supplementary" material in the paper. Can you provide that so that I can take some help for reproducing the results?
Thanks!