Question

使用tf.layers.conv2d时，设置初始化程序很简单，可以通过其参数完成。但是，如果我使用tf.nn.conv2d怎么办？我用这个代码。这相当于在kernel_initializer中设置tf.layers.conv2d参数吗？虽然程序运行没有错误，但我不知道如何验证它是否能达到预期的效果。

 with tf.name_scope('conv1_2') as scope:
        kernel = tf.get_variable(initializer=tf.contrib.layers.xavier_initializer(), 
                                 shape=[3, 3, 32, 32], name='weights')
        conv = tf.nn.conv2d(conv1_1, kernel, [1, 1, 1, 1], padding='SAME')
        biases = tf.Variable(tf.constant(0.0, shape=[32], dtype=tf.float32),
                             trainable=True, name='biases')
        out = tf.nn.bias_add(conv, biases)
        self.conv1_2 = tf.nn.relu(out, name=scope)
        self.parameters += [kernel, biases]

Answer 1

下面的操作是相同的（见here）。

至于内核及其初始化，我瞥了一眼代码，看起来一样...... layers.conv2d在结尾处tf.get_variable调用tf.layers.conv2d那一天。

但是我想通过经验来看，所以这里有一个测试代码，它使用每个方法（tf.nn.conv2d和import tensorflow as tf import numpy as np # the way you described in your question def _nn(input_tensor, initializer, filters, size): kernel = tf.get_variable( initializer=initializer, shape=[size, size, 32, filters], name='kernel') conv = tf.nn.conv2d( input=input_tensor, filter=kernel, strides=[1, 1, 1, 1], padding='SAME') return kernel # the other way def _layer(input_tensor, initializer, filters, size): tf.layers.conv2d( inputs=input_tensor, filters=filters, kernel_size=size, kernel_initializer=initializer) # 'conv2d/kernel:0' is the name of the generated kernel return tf.get_default_graph().get_tensor_by_name('conv2d/kernel:0') def _get_kernel(method): # an isolated context for each conv2d graph = tf.Graph() sess = tf.Session(graph=graph) with graph.as_default(), sess.as_default(): # important so that same randomness doesnt play a role tf.set_random_seed(42) # arbitrary input tensor with compatible shape input_tensor = tf.constant(1.0, shape=[1, 64, 64, 32]) initializer = tf.contrib.layers.xavier_initializer() kernel = method( input_tensor=input_tensor, initializer=initializer, filters=32, size=3) sess.run(tf.global_variables_initializer()) return sess.run(kernel) if __name__ == '__main__': kernel_nn = _get_kernel(_nn) kernel_layer = _get_kernel(_layer) print('kernels are ', end='') # compares shape and values if np.array_equal(kernel_layer, kernel_nn): print('exactly the same') else: print('not the same!')）声明一个conv2d，评估初始化的内核并进行比较。

我随意设置了不应该干扰比较的东西，比如输入张量和步幅。

ScoreStats

输出是...... 内核完全相同。

文档，顺便说一下：tf.nn.conv2d和tf.layers.conv2d。

使用权重初始化程序与tf.nn.conv2d

1 个答案: