- TORCH_HOME: used to custom model dir when downloading pretrained models from model zoos;
- Split2TnV: used to create train and validation set in the form of txt file;
- ItemList: taken the txt files create from Split2TnV and used to create pytorch datasets which then can be load with pytorch dataloader;
- SVSoftmax: Support Vector Guided Softmax Loss for Face Recognition: https://arxiv.org/pdf/1812.11317.pdf;
- Arcface: ArcFace: Additive Angular Margin Loss for Deep Face Recognition: https://arxiv.org/pdf/1801.07698.pdf;
- CyclicalLR: Cyclical Learning Rates for Training Neural Networks: https://arxiv.org/pdf/1506.01186.pdf;
- CosineLR: SGDR: STOCHASTIC GRADIENT DESCENT WITH WARM RESTARTS: https://arxiv.org/pdf/1608.03983.pdf;
- T_0: cycle length;
- T_mult: cycle length multiplier;
- max_lr: maximum learning rate value;
- total_steps: cycle length, can also be set by using epochs & steps_per_epoch;
