Closed Gincoeng closed 5 years ago
您好,我想请问一下当训练完CombinedMargin_spherenet36_E_model.prototxt之后,得到的模型特别大,若想要得到deploy.caffemodel,应该怎么样对其进行裁剪才能把innerproduct去掉,使网络变小。
Use 'deploy.prototxt' to load a net in caffe, the extra params in train_result.caffemodel will be ignored. Then save the net's parameters to deploy.caffemodel.
thank you so much
您好,我想请问一下当训练完CombinedMargin_spherenet36_E_model.prototxt之后,得到的模型特别大,若想要得到deploy.caffemodel,应该怎么样对其进行裁剪才能把innerproduct去掉,使网络变小。