Where can I find your training code?
johnnylu305 opened this issue · comments
Hi, nice to meet you. I am reading your paper and in your paper you mentioned that you will release the code here. However, I cannot find the training code. Where can I find your training code?
Do u got the train code？
@yuqian2019 No, I do not. I tried to write it by myself. But, the performance is not good. Perhaps, something wrong in my code. I hope the authors can share it, but I got no response from them.
@yuqian2019 You can find my personal training code in my repo. But, the performance is not good. Again, probably something wrong in my code.
@yuqian2019 You can find my personal training code in my repo. But, the performance is not good. Again, probably something wrong in my code.
ok, thank u very much. I will view ur code, and if i find any error, I will mention u.
@yuqian2019 No, I do not. I tried to write it by myself. But, the performance is not good. Perhaps, something wrong in my code. I hope the authors can share it, but I got no response from them.
hello，i tried to use your training code, but i met some mistakes.
"img, kernels, offsets_h, offsets_v, = ctx.saved_tensors
ValueError: not enough values to unpack (expected 4, got 0)"
So how to solve it.
thank you!
@yuqian2019 No, I do not. I tried to write it by myself. But, the performance is not good. Perhaps, something wrong in my code. I hope the authors can share it, but I got no response from them.
hello，i tried to use your training code, but i met some mistakes.
"img, kernels, offsets_h, offsets_v, = ctx.saved_tensors
ValueError: not enough values to unpack (expected 4, got 0)"
So how to solve it.
thank you!
Oh!, i found it, i have solved it, thank u.
@yuqian2019 No, I do not. I tried to write it by myself. But, the performance is not good. Perhaps, something wrong in my code. I hope the authors can share it, but I got no response from them.
hello，i tried to use your training code, but i met some mistakes.
"img, kernels, offsets_h, offsets_v, = ctx.saved_tensors
ValueError: not enough values to unpack (expected 4, got 0)"
So how to solve it.
thank you!Oh!, i found it, i have solved it, thank u.
ok! Have u recovery the result? @vardenppp
@johnnylu305 May I ask a question, in your train code, where did you define the dataset dir (both train and val). I couldn't figure it out. Would appreciate if you can shed some light on it.
@johnnylu305 I know it should be somewhere in DIV2K_dataset.py But I'm not quite sure what are LQ_root and GT_root. and where is the correct way to put data dir? Could you be more elaborate? Thank you.
@johnnylu305 hi, really thank you for your code, but i have question. You just use L1 loss in train code. However, in a paper there are 3 loss including TV loss, offset constraint loss. So... i wonder if the results using your code are same with the paper.