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

【Hackathon 5th No.69】 分类大模型--人体视觉任务SOLIDER #2992

Closed
wants to merge 5 commits into from
Closed

【Hackathon 5th No.69】 分类大模型--人体视觉任务SOLIDER #2992

wants to merge 5 commits into from

Conversation

Yang-Changhui
Copy link
Contributor

@Yang-Changhui Yang-Changhui commented Oct 1, 2023

swin_tiny_patch4_window7_224:
image
swin_small_patch4_window7_224:
image
swin_base_patch4_window7_224:
image
swin_small_patch4_window7_224的误差稍大,问题在最后一个nn.LayerNorm,权重一样,但是可能精度不一样,导致误差大于1e-6
权重更新在PaddleClas/docs/zh_CN/models/sodier.md文件中

由于一些模型是新加的,我不知道怎么加入https://github.com/PaddlePaddle/PaddleClas/tree/develop/ppcls/arch/backbone/variant_models,
希望指导
测试图片:
test

@paddle-bot
Copy link

paddle-bot bot commented Oct 1, 2023

Thanks for your contribution!

@cuicheng01
Copy link
Collaborator

对原始的代码侵入比较多,建议在variant_models中创建swin的变体,继承swin后改其中的方法,原始代码尽量不动

@Yang-Changhui Yang-Changhui closed this by deleting the head repository Oct 8, 2023
@Yang-Changhui
Copy link
Contributor Author

Yang-Changhui commented Oct 9, 2023

@cuicheng01 你好,由于develop分支代码更新,所以重新fork,新的pr #2995 ,望查看

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

Successfully merging this pull request may close these issues.

2 participants