Pytorch implementation of Robust Bi-Tempered Logistic Loss Based on Bregman Divergences
This repository contains the direct translation from Tensorflow to Pytorch of the paper "Robust Bi-Tempered Logistic Loss Based on Bregman Divergences" (https://arxiv.org/abs/1906.03361). The repo translates the official Tensorflow repo which can be found here: https://github.com/google/bi-tempered-loss
To test the loss function with similar values to the test in the original repo, run the following file:
python test_loss.py