What is the equivalent of weight_filler from Caffe in Tensorflow?

Question

I'm trying to convert a TensorFlow to Caffe model, but in Caffe what about weight_filler? My model in tf is:

model.add(Conv2D(16, kernel_size=(5, 5), padding="same", activation='selu', 
                 input_shape=(64, 64, 1)))

model.add(Conv2D(16, kernel_size=(5, 5), padding="same", activation='selu'))

model.add(Conv2D(16, kernel_size=(5, 5), padding="same", activation='selu'))

model.add(Conv2D(16, kernel_size=(5, 5), padding="same", activation='selu'))

model.add(Conv2D(16, kernel_size=(5, 5), padding="same", activation='selu'))

model.add(Conv2D(16, kernel_size=(5, 5), padding="same", activation='selu'))

model.add(Conv2D(16, kernel_size=(5, 5), padding="same", activation='selu'))

model.add(Conv2D(16, kernel_size=(5, 5), padding="same", activation='selu'))

model.add(MaxPooling2D(pool_size=(2, 2),strides=(2,2)))

Answer 1

weight_filler is the type of generator used to initialize weights and biases. In tensorflow if it's not specified the default initializer is glorot_uniform_initializer which is also called Xavier uniform initializer so the equivalent initializer in Caffe is xavier:

weight_filler {
  type: "xavier"
}