leaderj1001 / MobileNetV3-Pytorch

Implementing Searching for MobileNetV3 paper using Pytorch
MIT License
292 stars 70 forks source link

dropout #8

Closed harristeague closed 5 years ago

harristeague commented 5 years ago

It doesn't look like the dropout parameter is implemented. Have you trained with dropout (as mentioned in the paper).

jlhou commented 5 years ago

I have the same idea with you!May be it did't.

It doesn't look like the dropout parameter is implemented. Have you trained with dropout (as mentioned in the paper).

leaderj1001 commented 5 years ago

I'm so sorry. I could not confirm it because I did not have an Issue notification. I really sorry. I will reflect it as soon as possible. It was a part I forgot. Thank you !!

leaderj1001 commented 5 years ago

Again, I am so sorry for the delay.

leaderj1001 commented 5 years ago

Google's MobileNet-V2 code shows dropouts applied only to classifiers. So I have applied it, and I will try adding a dropout to the MobileBlock. Thank you !!