search

YU1ut / MixMatch-pytorch

Code for "MixMatch - A Holistic Approach to Semi-Supervised Learning"
MIT License
636 stars 129 forks source link

No weight decay? #22

Closed vijosam closed 5 years ago

vijosam commented 5 years ago

I notice that weight_decay is not set for Adam optimizer, see https://github.com/YU1ut/MixMatch-pytorch/blob/2df6c279a2106a9b87963c9350c70b7efbbb0e79/train.py#L115, but in the paper, weight_decay is set to ~ 0.02.

guixianjin commented 5 years ago

Is this the weight decay? https://github.com/YU1ut/MixMatch-pytorch/blob/master/train.py#L388

vijosam commented 5 years ago

Is this the weight decay? https://github.com/YU1ut/MixMatch-pytorch/blob/master/train.py#L388

thanks