Why you add Dropout on you model? about fairnas

[cxxgtxy]

I read the paper carefully.
It says that the rank experiment is performed "we don’t employ any other tricks
like dropout [21], cutout [6] or mixup [28]". When compared with ProxylessNAS, they said they used the training tricks like google's MNASNET(google used dropout.).
Thus, it should be two different experiments.

[PistonY]

Ok that, so the final result FairNAS-A got 75.34 top-1 on ImageNet using dropout? And is there any additional tricks?

[cxxgtxy]

I think so. However, I need to reproduce it to answer your question.

[fairnas]

@PistonY As @cxxgtxy says, in order to rank models, the sampled models are not trained with dropout. For the published model, it is thus trained to be on par with MnasNet, ProxylessNet.

[fairnas]

@PistonY Please refer Section 6.3 as we use similar hyperparameters and tricks like in literature MnasNet [24]. So the FairNAS-A is indeed trained with dropout.

[PistonY]

@cxxgtxy @fairnas Ok, thanks! I also want to reproduce the results of the paper myself, so I need to make sure my understanding is right.