search

thuml / Transfer-Learning-Library

Transfer Learning Library for Domain Adaptation, Task Adaptation, and Domain Generalization
http://transfer.thuml.ai
MIT License
3.39k stars 553 forks source link

Selection of super parameters such as learning rate and weight decay #125

Closed JonasZero closed 2 years ago

JonasZero commented 2 years ago

Hi, I found that the settings of learning rate, weight decay and lr scheduler of domain adaptation method are inconsistent. Is it necessary to set the same training parameters for different methods? And does the benchmark of different task algorithms provided in the document follow the same training parameters?

JunguangJiang commented 2 years ago

It's not necessary to set the same training parameters for different methods.

JonasZero commented 2 years ago

It's not necessary to set the same training parameters for different methods.

Okay, thank you.