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

add onecycle #5252

Merged
merged 1 commit into from
Jan 16, 2022
Merged

add onecycle #5252

merged 1 commit into from
Jan 16, 2022

Conversation

bupt906
Copy link
Contributor

@bupt906 bupt906 commented Jan 13, 2022

论文链接](https://arxiv.org/pdf/2012.12645.pdf)
参考代码
我的工作

  • 将pytorch代码转换成paddle代码,按照PaddleOCR的代码风格集成到项目中。

  • 调整代码,没有使用源码中有关momentum的参数。

  • 在自己的中文数据集(200W+)上做实验验证结构的可行性:

    • 不加载预训练模型进行训练(30轮),backbone模型用的是MobileNetV1Enhance,学习率策略为:name: Piecewise decay_epochs : [20, 30]. values : [0.01, 0.002]. warmup_epoch: 1. 训练acc75%,验证集acc86%。
    • 不加载预训练模型进行训练(30轮),backbone模型用的是MobileNetV1Enhance,学习率策略为:name: OneCycle max_lr: 0.01。训练acc79%,验证集87%。
  • 实验环境及参数
    单机4卡(V100,32G),num_workers=8,batch_size=256

onecycle.zip
注:由于#5171 cla认证问题,重新提交新的pr。

@paddle-bot-old
Copy link

Thanks for your contribution!

@littletomatodonkey littletomatodonkey merged commit 107a316 into PaddlePaddle:dygraph Jan 16, 2022
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

Successfully merging this pull request may close these issues.

2 participants