Functions — github.com/cvqluu/Angular-Penalty-Softmax-Losses-Pytorch

CodeHub MCPcopy Create free account

hub / github.com/cvqluu/Angular-Penalty-Softmax-Losses-Pytorch / functions

Functions14 in github.com/cvqluu/Angular-Penalty-Softmax-Losses-Pytorch

⨍ Functions 14 ◇ Types & classes 4

↓ 2 callersMethod__init__

↓ 2 callersFunctionget_embeds

(model, loader)

train_fMNIST.py:93

↓ 2 callersFunctionplot

(embeds, labels, fig_path='./example.pdf')

↓ 1 callersFunctionmain

train_fMNIST.py:15

↓ 1 callersFunctionparse_args

train_fMNIST.py:106

↓ 1 callersFunctiontrain_am

(train_loader, loss_type)

train_fMNIST.py:71

↓ 1 callersFunctiontrain_baseline

train_fMNIST.py:49

(self, num_classes=10)

(self, num_classes=10, loss_type='arcface')

Angular Penalty Softmax Loss Three 'loss_types' available: ['arcface', 'sphereface', 'cosface'] These losses are described i

loss_functions.py:8

(self, x, embed=False)

(self, x, labels=None, embed=False)

(self, x, embed=False)

input shape (N, in_features)

loss_functions.py:38