如何编写具有可训练参数的caffe python层？ [英] how to write caffe python layer with trainable parameters?

问题描述

我想学习如何编写caffe python层。

但是我只找到有关非常简单的层的示例，例如 pyloss 。

I want to learn how to write caffe python layers.
But I only find examples about very simple layers like pyloss.

如何编写具有可训练参数的python caffe？

How to write python caffe with trainable parameters?

例如，如何编写完全连接的python层？

For example, how to write a fully connected python layer?

推荐答案

Caffe将图层的可训练参数存储为 blobs 的向量。默认情况下，此向量为空，由您决定是否在层的设置中添加参数blob。在 test_python_layer.py 。

Caffe stores the layer's trainable parameters as a vector of blobs. By default this vector is empty and it is up to you to add parameters blobs to it in the setup of the layer. There is a simple example for a layer with parameters in test_python_layer.py.

请参见这篇文章，以获取有关caffe中 Python 层的更多信息。

See this post for more information about "Python" layers in caffe.

这篇关于如何编写具有可训练参数的caffe python层？的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！