search

Justin-Tan / generative-compression

TensorFlow Implementation of Generative Adversarial Networks for Extreme Learned Image Compression
MIT License
517 stars 105 forks source link

How is the effect on the test set? #22

Open zbaoye opened 6 years ago

zbaoye commented 6 years ago

Hi, author: When I re-train this model, it works pretty well on training dataset after 250 epoch. but it doesn't appear to work very well on the testing dataset. Is your result the same?

Thank you!

zhouqingyi616 commented 5 years ago

Same. There's no validation evaluating step. Need help.

ChenWei007HB commented 4 years ago

I met the same problem, when retraining the model, the test result of Kodak datasets can not surpass BPG method, even if tested on the model provided by the author.

imsleepy711 commented 4 years ago

Hi, I met the same problem. Do you have some solution now? Thank you!