About ‘differentiable=False’ in weight initializer of GBHybridLayer

researchmm / DBTNet

Code for our NeurIPS'19 paper "Learning Deep Bilinear Transformation for Fine-grained Image Representation"

105 stars 18 forks source link

Open amswly opened 4 years ago

amswly commented 4 years ago

why set differentiable=False, will it block the gradient of weight and bias.