anokland / local-loss

PyTorch code for training neural networks without global back-propagation
162 stars 37 forks source link