8 projects
surMSE
surMSE is a A loss function @PyTorch, that allows the target to be exceeded in the target direction.
FWU
FWU is a series of PyTorch-based neural network units. It has higher information utilization, prevents neuronal necrosis, and eliminates the need to add additionall activation functions.
orientLoss
orientLoss is a classification loss function @pytorch. It less prone to overfitting compared to crossEntropy. Compatible with random vector/embeddings with direction-indicating categories、probabilistic、oneHot as target.
glassInit
glassInit is a way to initialize the parameters of a neural network layer @pytorch. It is suitable for shortcut connections where tensor sizes need to be adjusted. The flow of information between forward and backward propagation is maximized while adjusting the tensor size.
Atiny
Atiny is a gradient based PyTorch optimizer. Atiny uses half of Adam's memory to achieve results that are not inferior to Adam.
fftLoss
The fftLoss @PyTorch is a frequency domain loss function that prevents the problem of weak frequency components being suppressed by strong frequency components when using a regular loss function.
RRPad
ReplicationReflectionPad combines ReplicationPad and ReflectionPad to prevent errors from being reported due to small input sizes when using non-constant padding types.
Afine
Afine is a gradient based PyTorch optimizer.
