search

JiahuiYu / wdsr_ntire2018

Code of our winning entry to NTIRE super-resolution challenge, CVPR 2018
http://www.vision.ee.ethz.ch/ntire18/
598 stars 123 forks source link

What is the number of epoch of the experiment? #15

Closed BigChen28 closed 5 years ago

BigChen28 commented 5 years ago

Hi, Jiahui. Could you tell me how many epoches the L1 loss convergence or PSNR reaches the optimal value when you train WSDR_B?

JiahuiYu commented 5 years ago

@BigChen28 In all ablation experiments we use same 250 epochs (Table 1, Figure 3 and 4).

Meanwhile, I find longer training epochs help to WDSR (while to EDSR the performance is similar). I guess it is mainly because we use 10x larger learning rate with weight normalization.

BigChen28 commented 5 years ago

I got it,thanks for your reply.