search

biubug6 / Pytorch_Retinaface

Retinaface get 80.99% in widerface hard val using mobilenet0.25.
MIT License
2.63k stars 774 forks source link

Should I change the learning rate according to different GPU nums? #137

Open xzxedu opened 4 years ago

xzxedu commented 4 years ago

Hi, many thanks for your work! May I ask will the training setting learning rate be changed automatically when I set different gpu number via CUDA_VISIBLE_DEVICES and ngpu in data/config.py? Is there anything else I need to change?

S201961152 commented 3 years ago

Hello，I also want to know this question，Do you solve it？

xzxedu commented 3 years ago

You don't need to change the learning rate if you don't change the batch size.

S201961152 commented 3 years ago

I changed the batchsize，and then how can I change the learning rate？ 

                            1663678924

                                邮箱：1663678924@qq.com

    On 12/21/2020 23:23, Zixiao Xing wrote:

You don't need to change the learning rate if you don't change the batch size.

—You are receiving this because you commented.Reply to this email directly, view it on GitHub, or unsubscribe. [ { "@context": "http://schema.org", "@type": "EmailMessage", "potentialAction": { "@type": "ViewAction", "target": "https://github.com/biubug6/Pytorch_Retinaface/issues/137#issuecomment-749029313", "url": "https://github.com/biubug6/Pytorch_Retinaface/issues/137#issuecomment-749029313", "name": "View Issue" }, "description": "View this Issue on GitHub", "publisher": { "@type": "Organization", "name": "GitHub", "url": "https://github.com" } } ]