search

martinarjovsky / WassersteinGAN

BSD 3-Clause "New" or "Revised" License
3.2k stars 725 forks source link

Results cannot be reproduced. #67

Closed zeakey closed 5 years ago

zeakey commented 5 years ago

Same as mentioned in #62 and #64, the results on CIFAR10 dataset cannot be reproduced!

I trained with the default settings for 100 epochs (which I think is enough), but the results still don't make sense.

Here is the example fake images generated via the Generator: screenshot from 2018-12-01 15-16-04

Any body reproduced the results on CIFAR10 (or LSUN bedroom) dataset, can you please tell the training details?

FANG-Xiaolin commented 5 years ago

Hi Zeakey,

Do you have any solution to this problem? Is there any necessary changes that should be made, e.g. to the parameters? Thanks

martinarjovsky commented 5 years ago

I don't know what 'results cannot be reproduced'. We don't have any CIFAR-10 results in the paper. Maybe you want to take a look at https://github.com/igul222/improved_wgan_training and https://arxiv.org/abs/1704.00028 .

Cheers :) Martin