You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
In the code, it seems that you use the model stored in ema to do the validation and get the final result.Here i have two questions:
1.Why do you use the ema model to do the validation, but do not use the model_wrapper which is directely preuned?
2.Is it supposed that ema model will get better result than model_wrapper?And why?
Thanks very much for your reply!
The text was updated successfully, but these errors were encountered:
Thanks very much for your reply!
I am quite interested in this phenomenon.
However, It seems that there is no discription about this problem in your paper? Does there have any other papers which mentioned about this problem?
It's a part of the Mnasnet training schedule which is mentioned in the paper. We also give MobileNetV2's performance for reference (73.6 top-1). Single-Path NAS: Designing Hardware-Efficient ConvNets in less than 4 Hours mentioned this phenomenon.
In the code, it seems that you use the model stored in ema to do the validation and get the final result.Here i have two questions:
1.Why do you use the ema model to do the validation, but do not use the model_wrapper which is directely preuned?
2.Is it supposed that ema model will get better result than model_wrapper?And why?
Thanks very much for your reply!
The text was updated successfully, but these errors were encountered: