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

请问一下,我利用vig训练mini-imagenet,acc总是100,请问一下这是什么原因 #189

Open
2450961986 opened this issue Mar 26, 2023 · 13 comments

Comments

@2450961986
Copy link

image

@iamhankai
Copy link
Member

太强了,可以发nature了;
大佬带带我

@2450961986
Copy link
Author

抱歉,我确实不太懂,但感觉总是100,是不是我的操作哪有问题,无意冒犯,请教一下您

@iamhankai
Copy link
Member

是不是你的数据就1类啊

@goodmorningsi
Copy link

请问前几天出的这个模型pvig_m_im21k_90e.pth,和之前的pvig_m.pth的模型有哪些改进呢

@iamhankai
Copy link
Member

wow 你这都留意到了,这是我们在imagenet-21k预训练的模型

@goodmorningsi
Copy link

wow 你这都留意到了,这是我们在imagenet-21k预训练的模型
哈哈哈哈,好嘞,明白了,时刻关注着

@goodmorningsi
Copy link

wow 你这都留意到了,这是我们在imagenet-21k预训练的模型

那我也可以利用这个模型作为mask_rcnn的backbone进行训练吧

@iamhankai
Copy link
Member

可以,精度应该会更高

@goodmorningsi
Copy link

可以,精度应该会更高

好嘞🆗

@Konmua
Copy link

Konmua commented Jul 28, 2023

你好,我用的Tiny Imagenet复现也遇到了同样的问题,想请问一下是什么原因呢,该怎样解决呢?
image

@lyzla
Copy link

lyzla commented Jun 2, 2024

你好,请问你用的是单张卡来训练吗?lr是多少?batchsize是多少呀?我也在imagenet-mini上训练,设置的参数是这样的--lr 2e-4、-b 16,但是loss很大
1f797e558789677bdd6162a147181308

@iamhankai
Copy link
Member

你好,请问你用的是单张卡来训练吗?lr是多少?batchsize是多少呀?我也在imagenet-mini上训练,设置的参数是这样的--lr 2e-4、-b 16,但是loss很大 1f797e558789677bdd6162a147181308

学习率太大了,调小点

@lyzla
Copy link

lyzla commented Jun 3, 2024 via email

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

5 participants