hwalsuklee

Hi. I'm really sorry for late response. I could not understand what your problem is. Did you mean that final test accuracy is not the same when changing batch-size?

Hi. I'm really sorry for late response. I have been faced with the same problems that you mentioned. Honestly speaking, I could not achieved spirals in "Starry night" from the...

Hi @nlminh All pre-trained models are obtained from the same default parameters.

Those are caused by tensorflow version mismatch. I think you're using tensorflow version over 1.0 Please check https://www.tensorflow.org/install/migration 1. tf.pack must be changed into tf.stack for tensorflow version over 1.0...

Hey! check the code :D

깃허브 링크는 깨져있던데 https://github.com/seungjaeryanlee/random-network-distillation-pytorch 도커 이미지만 활용해서 돌려볼 수 있는건가요?

Thank you for pointing out. As you mentioned, ASTER is a recognition algorithm. So you can see ASTER in 'Text Recognition' part. However, A paper of ASTER also showed end-2-end...

Hi. As it is mentioned on the top in ops.py, this is borrowed from https://github.com/carpedm20/DCGAN-tensorflow It is just an implementation of fully connected layer. Input×weight+bias Hwalsuk Lee

Hi, from my experience, I suggest you to use deeper architectures. Generated images are a bit sensitive to network architecture. DCGAN like architecture would be good start point.

Yes! f-GAN is also under the plan. I expect to implement f-GAN after finishing celeba-generation that I'm working on.