search

WeitaoVan / L-GM-loss

Implementation of our accepted CVPR 2018 paper "Rethinking Feature Distribution for Loss Functions in Image Classification"
MIT License
172 stars 34 forks source link

question about evaluation #15

Open yuyijie1995 opened 5 years ago

yuyijie1995 commented 5 years ago

In your tensorflow code,I found you use the dist as the logits_eval and take the argmax as the right label . But the dist is the distance between the example feature and the distribution mean ,why the biggest distance distribution is the right label?