search

WeiChengTseng / Pytorch-PCGrad

Pytorch reimplementation for "Gradient Surgery for Multi-Task Learning"
BSD 3-Clause "New" or "Revised" License
302 stars 42 forks source link

Can PCGrad be used in conjunction with DWA？ #3

Closed fuweifu-vtoo closed 2 years ago

fuweifu-vtoo commented 3 years ago

hello,thanks for your work I have a question about PCGrad and DWA It is :Can PCGrad be used in conjunction with DWA？ if the answer is YES, How can i apply it in the code?

WeiChengTseng commented 3 years ago

Hi, I am not quite sure what the DWA means. Is that https://arxiv.org/pdf/2010.14838.pdf ?

fuweifu-vtoo commented 3 years ago

I mean Dynamic Weight Average，mentioned in the "End-to-End Multi-Task Learning with Attention",

lorenmt commented 2 years ago

I mean Dynamic Weight Average，mentioned in the "End-to-End Multi-Task Learning with Attention",

Yes it can.