Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

实验复现精度和论文中的精度有所偏差 #9

Open
wangyan123454321 opened this issue Jun 20, 2023 · 3 comments
Open

实验复现精度和论文中的精度有所偏差 #9

wangyan123454321 opened this issue Jun 20, 2023 · 3 comments

Comments

@wangyan123454321
Copy link

您好呀,很荣幸能读到您的这篇文章。我在复现过程中,论文中提到的spiking MNIST精度98.84%,我复现结果是97.30%,NMNIST论文精度是98.88,我复现精度是96.65%,语音数据集TI46论文精度是90.50%,我复现精度只达到了82.91%。其中前两个实验结果我是完全按照您在代码种的最佳结果的log.txt中参数设置。请问一下我复现精度有点偏差,是因为您使用了论文中未提到的trick吗?还有TI46语音数据集的训练,方便提一下它的参数设置的txt文件吗,我也尝试了很多参数设置,但是始终达不到精度?谢谢您百忙之中的回复。

@jinyyy666
Copy link
Owner

jinyyy666 commented Jun 20, 2023 via email

@wangyan123454321
Copy link
Author

您好,我用cuda8.0训练spikingMNIST和NMNSIT数据集,结果还是差一两个点。因为我想在您论文基础上改进呢,但是一直连您的精度也复现不出来,这两个数据集是有什么别的trick吗

@wangyan123454321
Copy link
Author

我注意到您在网络中的initRandom函数中设置了srand函数,但是后来您注释掉了;我在猜想精度的不同可能与不同编译器中生成未指定种子的随机数有关,请问一下您是因为什么注释掉的这一段代码呢?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants