cifar10 code，loss remains unchanged

[zhangyu199802]

about the pytorch version for implementing cifar10 tasks ，sorry i don't know where the bug is，loss remains unchanged，low accuracy

[Cristina1213]

hello, did you succeed?

[zhangyu199802]

hello, I gave up this.

…

------------------ 原始邮件 ------------------ 发件人: "yjwu17/STBP-for-training-SpikingNN" ***@***.***>; 发送时间: 2022年7月18日(星期一) 晚上10:03 ***@***.***>; ***@***.******@***.***>; 主题: Re: [yjwu17/STBP-for-training-SpikingNN] cifar10 code，loss remains unchanged (Issue #6) hello, did you succeed? — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: ***@***.***>

[crashkk]

I also found the same problem,but have solved it by setting firing threshold much lower(around 0.25).Probably this issue came from neurons' early death.So I am confused that whether 0.75 the best hyperparameter selection?

[QianpengLi577]

@crashkk Hi, do you still remember the final acc for Cifar10？

[crashkk]

@QianpengLi577 hi, 70%-80% roughly,but with severe overfitting

[QianpengLi577]

@QianpengLi577 hi, 70%-80% roughly,but with severe overfitting

Thanks for your reply