search

RaphaelMeudec / deblur-gan

Keras implementation of "DeblurGAN: Blind Motion Deblurring Using Conditional Adversarial Networks"
582 stars 244 forks source link

Question about learning rate #80

Open wwjwy opened 3 years ago

wwjwy commented 3 years ago

In the paper, you said "The learning rate is set initially to 10−4 for both generator and critic. After the first 150 epochs we linearly decay the rate to zero over the next 150 epochs". But i can not find relevant code for adjusting learning rate. Can you help me?

gevindharmalim commented 3 years ago

In the paper, you said "The learning rate is set initially to 10−4 for both generator and critic. After the first 150 epochs we linearly decay the rate to zero over the next 150 epochs". But i can not find relevant code for adjusting learning rate. Can you help me?

Hello @wwjwy , i'm currently also trying this code to run but i have some problems when running it and your post is the most recent i can find, i want to ask some issues with it, if you do not mind, can you give me a way to reach you ? any social media or platform would be okay. Thankyou.

p.s.: sorry i could not provide any solution to your issues.

wwjwy commented 3 years ago

In the paper, you said "The learning rate is set initially to 10−4 for both generator and critic. After the first 150 epochs we linearly decay the rate to zero over the next 150 epochs". But i can not find relevant code for adjusting learning rate. Can you help me?

Hello @wwjwy , i'm currently also trying this code to run but i have some problems when running it and your post is the most recent i can find, i want to ask some issues with it, if you do not mind, can you give me a way to reach you ? any social media or platform would be okay. Thankyou.

p.s.: sorry i could not provide any solution to your issues.

my email is wwjhhu@gmail.com

Sylv-Lej commented 3 years ago

LR decay is all done in keras https://keras.io/api/optimizers/learning_rate_schedules/