Zhuang Liu

Thanks for pointing out. I've just found other people discussing this, and wrote a comment on reddit here https://www.reddit.com/r/MachineLearning/comments/67fds7/d_how_does_densenet_compare_to_resnet_and/?utm_content=title&utm_medium=hot&utm_source=reddit&utm_name=MachineLearning My suggestion is that trying a shallow and wide densenet, by...

Thank you for your interests 1. Yes, choosing 3 or 4 times is ok, and possibly better than 2 times. In our trials, 2 times works better than 1. Probably...

1. In ResNet, the output of a residual block (1x1 -3x3 -1x1) needs to be the same as the input to do the summation, so you need to first reduce...

Thanks. So you found case 2 achieves worse accuracy than case 1?

Ok, I'll try it in Torch, when there is free GPU.

Sorry, for keras I don't know if there are pretrained models. Maybe check the keras implementations listed in our readme page, or search "keras DenseNet" on google or github.

I've added this on our readme page. Thanks

Hi @miraclewkf , could you please make the models public? Some people have asked me about MXNet models, I think it would be really nice of you to give open...

@miraclewkf Yes, I have put your link in the "pretrained model" section in the readme page. But when I tried to download it, access needs to be requested. Are you...

Hello @taki0112 A1. As we mentioned in the paper, we directly followed ResNet's optimization settings (https://github.com/facebook/fb.resnet.torch), except that we train 300 epochs instead of ~160 epochs. We didn't try any...