关于代码完整性

[Ien001]

hi there, im quite interested in your work and i have try your code some days ago, however, i cant get the same result as the that in the paper. These days, i notice that you are rebuilding your code these days, would u plz inform me how soon can you finish it? I`m just cant wait to try it. Thanks a lot!

Ian

[songdejia]

Thanks for your attention to my work. Because my work is very busy, please forgive me, for so long no contact with you. I'll come back to you as soon as possible.

[songdejia]

@Ien001

[Ien001]

thanks a lot! I see it

[Ien001]

@songdejia

[Ien001]

hi, I find that the section 3.3 layer initialization in the paper is not implemented in your code, and I am a little confused about this section, maybe it`s not useful? By the way, I only get 78.6% with your code...

Ian

[Blue-Clean]

@songdejia hi, I noticed that you didn't implement layer initialization in the paper. Can you share it in your leisure time?

[pengshiqi]

@Ien001
Hello, I have the same problem as yours... that my best result with this code is only 77.9%... Have you got any improvements in these days？

Sincerely.

[wangkangnian]

@Ien001 Hello, I have the same problem as yours... that my best result with this code is only 77.9%... Have you got any improvements in these days？

Sincerely.

Hi , How do you get the tests accuracy 77.9%. Which I only get 40% on top-1 Is it the loss1-3 ratio which affects?

[pengshiqi]

@wangkangnian
Adjusting lr from 0.1 to 0.01 / 0.001 helps. The loss ratio also affects the result.

[Ien001]

@pengshiqi Not yet... I am working on it these days... and how does the ratio of loss influence the result? have u ever tried others? thanks

[Ien001]

@pengshiqi Also, using tencrop on the pics makes a difference, but only limited...

[Ien001]

@songdejia hi， would you plz post your loss value when you get 85.14% in test set? thanks,

Ian