I find a potential mistake, could you please have a check?

MozarTuring commented 5 years ago

Line 136 in mnist_trainer.py, the operation (p_loss > 0).sum() may give unwanted result because p_loss > 0 returns a BytetTensor which can represent value from 0~255. However, the summation may exceed 255.

XiaoHao-Chen commented 5 years ago

Have you managed to run through the mnist_trainer.py program?

MozarTuring commented 5 years ago

yes, I did.

XiaoHao-Chen commented 5 years ago

How did you solve the memory overload problem? Can you tell me how much memory your graphics card has？ Thank you

XiaoHao-Chen commented 5 years ago

是的，我做到了。

I have no good solution to the problem of the running program of pixelcnn, can you tell me how you solved the problem at that time? That is, the issue of out of memory. Thank you very much for your help.

XiaoHao-Chen commented 5 years ago

yes, I did.

Can you give me some advice on the operation of this code? Suggestions for modifications within the hardware or code? Thank you very much!

XiaoHao-Chen commented 5 years ago

是的，我做到了。

I still want to know how you solved the memory overflow problem. I still haven't solved it for so long. Please！

MozarTuring commented 5 years ago

You are using gpu or cpu? What’s the memory of yours?

发自我的 iPhone

在 2019年4月30日，下午5:21，331801070049 notifications@github.com 写道：

是的，我做到了。

I still want to know how you solved the memory overflow problem. I still haven't solved it for so long. Please！

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

XiaoHao-Chen commented 5 years ago

你在使用gpu还是cpu？你的记忆是什么？发自我的iPhone ... 在 2019年4月30日，下午5:21，331801070049 @.***> 写道：是的，我做到了。 I still want to know how you solved the memory overflow problem. I still haven't solved it for so long. Please！ — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or mute the thread.

All the GPU,6G,8G,12G I used have been used. 150G CPU has also been used, but all memory overflows are displayed

MozarTuring commented 5 years ago

Are you sure the reason is out of memory? running minist_trainer.py only took up less than 8g

发自我的 iPhone

在 2019年4月30日，下午7:30，331801070049 notifications@github.com 写道：

你在使用gpu还是cpu？你的记忆是什么？发自我的iPhone ... 在 2019年4月30日，下午5:21，331801070049 @.***> 写道：是的，我做到了。 I still want to know how you solved the memory overflow problem. I still haven't solved it for so long. Please！ — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or mute the thread.

All the GPU,6G,8G,12G I used have been used. 150G CPU has also been used, but all memory overflows are displayed

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

XiaoHao-Chen commented 5 years ago

Are you sure the reason is out of memory? running minist_trainer.py only took up less than 8g 发自我的 iPhone … 在 2019年4月30日，下午7:30，331801070049 @.> 写道：你在使用gpu还是cpu？你的记忆是什么？发自我的iPhone ... 在 2019年4月30日，下午5:21，331801070049 @.> 写道：是的，我做到了。 I still want to know how you solved the memory overflow problem. I still haven't solved it for so long. Please！ — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or mute the thread. All the GPU,6G,8G,12G I used have been used. 150G CPU has also been used, but all memory overflows are displayed — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or mute the thread.

But it really goes out of memory. The problem has been bothering me for a long time. I don't know how to change the program. If you don't mind if you can send me your successful minist_trainer.py, let me refer to it. Thank you very much for your help. My email is cxhclc@163.com

XiaoHao-Chen commented 5 years ago

As I look for the problem, I find that when you run to pixelcnn.py, the amount of GPU increases dramatically and does not release.

MozarTuring commented 5 years ago

Could you briefly introduce yourself?

发自我的 iPhone

在 2019年4月30日，下午10:30，331801070049 notifications@github.com 写道：

As I look for the problem, I find that when you run to pixelcnn.py, the amount of GPU increases dramatically and does not release.

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

XiaoHao-Chen commented 5 years ago

I come from China henan zhengzhou, is a master of reading, the direction is image classification, see the bad before gan, this paper found that code, you want to run the trying to reappear the results of the paper, but in running mnist_trainer. Py the existing problems, during this time I try I can think of a way to solve it, but have not had any effect. So if you still have the source code for the successful run, please send it to me for reference. Thank you very much!

XiaoHao-Chen commented 5 years ago

I really need your help with this code. I really can't find a solution to the current problem. I hope you can give me some help. Please! My email is cxhclc@163.com.

MozarTuring commented 5 years ago

check my github homepage

发自我的 iPhone

在 2019年5月1日，上午12:06，331801070049 notifications@github.com 写道：

I really need your help with this code. I really can't find a solution to the current problem. I hope you can give me some help. Please! My email is cxhclc@163.com.

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

XiaoHao-Chen commented 5 years ago

Thank you very much

kimiyoung / ssl_bad_gan

I find a potential mistake, could you please have a check? #8