小编Pet*_*tte的帖子

以下类具有let声明为隐式展开变量的' '属性.以前这适用于Xcode 6.2:

class SubView: UIView {

    let pandGestureRecognizer: UIPanGestureRecognizer!

    required init(coder aDecoder: NSCoder) {
        super.init(coder: aDecoder)
        self.pandGestureRecognizer = UIPanGestureRecognizer(target: self, action: "panAction:")
    }

    func panAction(gesture: UIPanGestureRecognizer) {
        // ...
    }
}

更新到Xcode 6.3(使用Swift 1.2)后,会发生以下编译错误:

Property 'self.panGestureRecognizer' not initialized at super.init call
Immutable value 'self.panGestureRecognizer' may only be initialized once

在super.init通话前移动以下行:

self.pandGestureRecognizer = UIPanGestureRecognizer(target: self, action: "panAction:")

给出以下错误:

'self' is used before super.init call

该属性' panGestureRecognizer'不需要变异,因此必须将其声明为常量' let'.由于它是常量,因此必须在声明时具有初始值,或者在init方法中初始化它.要初始化它,需要self在' target'参数中传递' ' …

我正在尝试堆叠2层来tf.nn.conv2d_transpose()对张量进行上采样.它在前馈期间工作正常,但在向后传播时出现错误: ValueError: Incompatible shapes for broadcasting: (8, 256, 256, 24) and (8, 100, 100, 24).

基本上,我只是将第一个的输出设置conv2d_transpose为第二个的输入:

convt_1 = tf.nn.conv2d_transpose(...)
convt_2 = tf.nn.conv2d_transpose(conv_1)

只使用一个conv2d_transpose,一切正常.只有多个conv2d_transpose堆叠在一起时才会出现错误.

我不确定实现多层的正确方法conv2d_transpose.任何关于如何解决这个问题的建议都将非常感激.

这是一个复制错误的小代码:

import numpy as np
import tensorflow as tf

IMAGE_HEIGHT = 256
IMAGE_WIDTH = 256
CHANNELS = 1

batch_size = 8
num_labels = 2

in_data = tf.placeholder(tf.float32, shape=(batch_size, IMAGE_HEIGHT, IMAGE_WIDTH, CHANNELS))
labels = tf.placeholder(tf.int32, shape=(batch_size, IMAGE_HEIGHT, IMAGE_WIDTH, 1))

# Variables
w0 = tf.Variable(tf.truncated_normal([3, 3, CHANNELS, …