search

zma-c-137 / VarGFaceNet

312 stars 84 forks source link

Do you have a plan to release the training code to reproduce the paper results? #4

Open TatsuyaShirakawa opened 5 years ago

TatsuyaShirakawa commented 5 years ago

Do you have a plan to release the training code to reproduce the paper results? Thanks.

yanmengjia commented 5 years ago

Hi, We will give you details of training if you update your questions in issues. Thanks.

clhne commented 5 years ago

@TatsuyaShirakawa @yanmengjia Hi, all Could you give the details of training, include the hyper-parameter settings, etc. Look forward to your reply. Thanks.

Coderx7 commented 4 years ago

I'm looking for this as well. Could you kindly provide the codes and training scripts so the results can be replicated and we hopefully can better experiment with the code . Its greatly appreciated

nullptr-leo commented 4 years ago

@yanmengjia 请问下 recursive=1 的蒸馏就是普通蒸馏吗？还是说普通蒸馏完再来一次？ 或者换句话说，有没有 recursive=0 这种配置？

anandhakrishnanh commented 4 years ago

@zma-c-137 Looking forward for the training code as well. The accuracy you claim seems very impressive.