wy1iu / sphereface-plus

SphereFace+ Implementation for <Learning towards Minimum Hyperspherical Energy> in NIPS'18.

wy1iu/sphereface-plus Issues

Effect of mean and scale on the sphereface-plus training
Updated 4 years ago
Base lambda value in the log file
Closed 4 years ago1
Information about "inter_class_dist_output" layer
Updated 4 years ago
do we have to train model for new faces?
Closed 4 years ago4
maximize_inter_class_dist becomes zero when training with just one GPU instead of two
Closed 4 years ago2
Confusion in the display of the loss while training sphereface-plus
Closed 4 years ago2
Hyper-param explained.
Closed 5 years ago2
The loss implemented in Inter_class_layer seems to be different from the MHE loss form on paper
Closed 5 years ago6
how to modify the params of layer "inter_class_dist_output"?
Closed 6 years ago1
About the inter_class_param for finetuning
Closed 6 years ago2