trained SRGAN model : https://github.com/tensorlayer/srgan/releases/tag/1.2.0

[Andreababy]

Can you provide me your trained model? It takes me too much time to train the DIV2K dataset. @zsdonghao

[zsdonghao]

what is your email address?

[Andreababy]

2018171324@qq.com thank you

[zsdonghao]

i am testing a new implmentation of subpixel conv 2d, may send you after testing it.

[Andreababy]

Ok.

[Lakshya-Kejriwal]

Hi @zsdonghao i would also like to request that if you could provide me a copy of the trained model as well? Thanks

[zsdonghao]

@Lakshya-Kejriwal sure, give me your email.

[Lakshya-Kejriwal]

lkejri24@yahoo.com

[zsdonghao]

@Andreababy @Lakshya-Kejriwal i already send the model to you via wetransfer, please note that, this work is only for academic and non-commercial use only. For commercial use, please contact dhsig552@163.com.

[shawnyuen]

A single Titan X for training takes about 4 days... Thanks for publicly providing the codes!!! @zsdonghao some testing results (bicubic, generated; very nice for animal images, but for people, the result is not very good; more tests are to be done):

[tonymonk007]

Could you please also send me the model? My email is jasongeophysics@gmail.com. I am running on CPU. It takes me three days to reach 20/2000. Thanks a lot.

[RichardIERG]

Could you please send me your trained model? My email address is hc85514301@gmail.com. Thanks a lot!

[flyywh]

Could you please send me your trained model? My email address is ywhlist150578209@gmail.com. Thank you for sharing the work!

[zsdonghao]

@flyywh @jierice @RichardIERG please check your email, i send you via wetransfer.

[zsdonghao]

@shawnyuen for people, the result is not very good. I think the reason is that the training set do not have any people ~

[yanchenj]

Hey @zsdonghao , Could you please send me your trained model? I found training on my PC is super time consuming. My email is yangchenj@gmail.com. Thanks so much for sharing your work! I really appreciate!

[luochuwei]

Could you please send me your trained model? My email address is luochuwei@gmail.com. Thank you very much for sharing the work!

[jzrita]

May I add one more? Could you please send me your trained model? Email: jinzhi_126@163.com. Thanks.

[zsdonghao]

@jzrita @listeng @luochuwei @yanchenj I already send the model to you via wetransfer, please note that, this work is only for academic and non-commercial use only. For commercial use, please contact dhsig552@163.com.

[jzrita]

@zsdonghao Hi, Donghao. The file has been well received and thank you very much.

[JustinhoCHN]

@zsdonghao hello, can you send me a copy of your trained model? Email: justinhochn@gmail.com ,thanks a lot!

[ZJiaQi]

May I add one more? Could you please send me your trained model? Email: zhang_jiaqi01@163.com. Thank you.

[ParanoidPY]

@zsdonghao 你好，我运行您的代码，出现了out of memory的错误，我的GPU为1080，8g显存。我想问一下这个是我配置原因么？您在什么配置下运行程序的？

[hufengshuo07]

@ParanoidPY I succeeded in training on one 1080ti with 11g graphics memory. Maybe you can try a smaller batch size.

[JustinhoCHN]

@ParanoidPY 我用的是1080ti，显存占用了8.5G左右，所以你把batch_size改成8吧，默认是16的。

[ParanoidPY]

@hufengshuo07 @JustinhoCHN THANK YOU very much! i succeeded running the code with batch size 4! Is there any other ways to train the model with the batch size 16 in my hardware settings?

[Andreababy]

测试图片占的内存太大了，降低图片大小

[junhyukk]

Could you please send me your trained model? email: ghgh3269@gmail.com Thank you.

[JiahaoYao]

Could you please send me your trained model? email: zxc604036406@foxmail.com Thank you.

[zsdonghao]

Hi all, @JiahaoYao @ghgh3269 @ZJiaQi @JustinhoCHN sorry for the late reply, please check your email, I send it via wetransfer.

[JiahaoYao]

@zsdonghao Hi, Donghao. The file has been well received and thank you very much.

[JiahaoYao]

However, I have a silly question: when I set the BatchSize to be 16, it stuck into OOM problem; when I set the BatchSize to be 8, I have the following problem: Traceback (most recent call last): File "main.py", line 316, in train() File "main.py", line 144, in train tl.vis.save_images(sample_imgs_96, [ni, ni], save_dir_ginit+'/_train_sample_96.png') File "/home/deeplearning/SRGAN/SRGAN/tensorlayer/visualize.py", line 53, in save_images assert len(images) <= size[0] size[1], "number of images should be equal or less than size[0] size[1] {}".format(len(images)) AssertionError: number of images should be equal or less than size[0] * size[1] 8

I really want to run the code, what can I do?

[ZJiaQi]

Hi, Jiahao, where should I to find the file? I haven't found out model in my emails. Thanks.

发自网易邮箱大师 On 08/26/2017 22:36, Jiahao Yao wrote:

@zsdonghao Hi, Donghao. The file has been well received and thank you very much.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or mute the thread.

[zsdonghao]

@JiahaoYao check how to use tl.vis.save_images man

[ZJiaQi]

@zsdonghao I recevied the model, and I use the model you send to reconstruct the test images, but I got picture like this. What's wrong with me? Thank you.

[zsdonghao]

@ZJiaQi Make sure the image format the same with the test code, I guess your image scale didn't rescale to -1 to 1.

[JustinhoCHN]

may I ask why we need to rescale the images to [-1, 1] ? Is this something like zero-centred data or something?

[zsdonghao]

@JustinhoCHN it is for zero-centreing, just like other CNN training.

[agingdata]

@zsdonghao Could you please send me your trained model? I want to study it. Emai:jk747586650@163.com

[limwenyao]

@zsdonghao Hello, I would appreciate a copy of your trained model as well, thanks! Email: limwenyao@hotmail.com

[53756e4c69]

@zsdonghao Could you please send me your trained model? Emai:53756e4c69@gmail.com

[JiahaoYao]

I have a question, why there is always a memory error. I have used 4 GPUs, K80. Did it mean that I would have to use even more GPUs.

[] save images [] checkpoint/g_srgan.npz saved Traceback (most recent call last): File "main.py", line 320, in train() File "main.py", line 246, in train tl.files.save_npz(net_d.all_params, name=checkpointdir+'/d{}.npz'.format(tl.global_flag['mode']), sess=sess) File "/home/jhyao/deeplearning/SRGAN/SRGAN/tensorlayer/files.py", line 552, in save_npz np.savez(name, params=save_list_var) File "/home/jhyao/anaconda3/envs/deeplearning/lib/python3.6/site-packages/numpy/lib/npyio.py", line 593, in savez _savez(file, args, kwds, False) File "/home/jhyao/anaconda3/envs/deeplearning/lib/python3.6/site-packages/numpy/lib/npyio.py", line 703, in _savez pickle_kwargs=pickle_kwargs) File "/home/jhyao/anaconda3/envs/deeplearning/lib/python3.6/site-packages/numpy/lib/format.py", line 576, in write_array pickle.dump(array, fp, protocol=2, **pickle_kwargs) MemoryError

[john820911]

@zsdonghao Can I also have your trained model? It would be much appreciated to have one :) My email: r05943135@ntu.edu.tw

[naseeresmael]

@zsdonghao Could you please send me your trained model? My email address is omar44khatab@gmail.com. Thank you for sharing the work!

[zsdonghao]

Check your email for the model. Note that this work can't be used in commerical project.

[naseeresmael]

great I recevied the model, thank you @zsdonghao .

[yan7109]

@zsdonghao please share your trained model to derek.zh.yan@stanford.edu, thanks

[53756e4c69]

i got you mail,but the Transfer has deleted Could you please send me your trained model again? My Email : 53756e4c69@gmail.com thank you

[Aixile]

@zsdonghao Could you please send me your trained model? jinyh13@fudan.edu.cn, Thanks

[kankur]

@zsdonghao Can you please send me the trained model on kadam.ankur04@gmail.com.

[zsdonghao]

wiqi@alumni.ubc.ca kadam.ankur04@gmail.com jinyh13@fudan.edu.cn 53756e4c69@gmail.com derek.zh.yan@stanford.edu

Please check your email.