Tensorflow 2.1 报错整合

Tensorflow 2.1 报错整合

RuntimeError: loss passed to Optimizer.compute_gradients should be a function when eager execution is enabled.

RuntimeError: Attempting to capture an EagerTensor without building a function.

RuntimeError: When eager execution is enabled, var_list must specify a list or dict of variables to save

• 当eager execution开启的时候，loss应该是一个Python函数。
• 在Tensorflow 2.0 中，eager execution 是默认开启的。
• 所以，需要先关闭eager execution
• tf.compat.v1.disable_eager_execution()

tf.compat.v1.placeholder(tf.float32, shape=(1024, 1024))

RuntimeError: tf.placeholder() is not compatible with eager execution.

• 如果要将代码从tensorflow v1转换为tensorflow v2，则必须实现tf.compat.v1并且占位符存在于tf.compat.v1.placeholder;
• 当eager execution开启的时候，loss应该是一个Python函数。
• 在Tensorflow 2.0 中，eager execution 是默认开启的。
• 所以，需要先关闭关闭紧急执行,eager execution。
• tf.compat.v1.disable_eager_execution()

AttributeError: module ‘tensorflow’ has no attribute ‘random_normal’

• tf.random_normal
• tf.random.normal

AttributeError: module ‘tensorflow’ has no attribute ‘global_variables_initializer’

• tf.global_variables_initializer()
• tf.compat.v1.global_variables_initializer()

AttributeError: module ‘tensorflow’ has no attribute ‘Session’

• tf.Session()
• tf.compat.v1.Session()

AttributeError: module ‘tensorflow’ has no attribute ‘assign’

• tf.assign()
• tf.compat.v1.assign()

AttributeError: module ‘tensorflow_core._api.v2.train’ has no attribute ‘Saver’

• tf.train.Saver()
• tf.compat.v1.train.Saver()

AttributeError: module ‘tensorflow’ has no attribute ‘placeholder’

• tf.placeholder(tf.float32)
• tf.compat.v1.placeholder(tf.float32)

AttributeError: module ‘tensorflow’ has no attribute ‘mul’

• tf.mul(input1, input2)
• tf.multiply(input1, input2)

AttributeError: module ‘tensorflow_core._api.v2.train’ has no attribute ‘SummaryWriter’

• tf.train.SummaryWriter("./tmp", sess.graph)
• tf.compat.v1.summary.FileWriter("./tmp", sess.graph)

ModuleNotFoundError: No module named ‘tensorflow.examples.tutorials’

解决方案1–针对没有tutorials文件报错的解决

1. 先检查tensorflow中是否含有tutorials
   • 检测路径：…\Lib\site-packages\tensorflow_core\examples；
   • 文件夹下只有saved_model这个文件夹，没有tutorials文件夹；
2. 进入github的tensorflow主页下载缺失的文件
   下载地址：tensorflow tutorials
3. 拷贝下载的tutorials整个文件夹
   • 解压下载的包，找到tutorials整个文件夹；
   • 拷贝至…\Lib\site-packages\tensorflow_core\examples下；
4. 再次加载包
   from tensorflow.examples.tutorials.mnist import input_data

解决办法2

tensorflow2.0的数据集集成到keras高级接口之中，使用如下代码一般都能下载

mint=tf.keras.datasets.mnist
(x_,y_),(x_1,y_1)=mint.load_data()

input_data.read_data_sets() 下载数据集失败

from tensorflow.examples.tutorials.mnist import input_data
print ("Download and Extract MNIST dataset")
mnist = input_data.read_data_sets('data/', one_hot=True)

1. mnist数据集获取:可从Yann LeCun教授管网获取;
2. 导入下载到本地的mnist数据集; "data/"为数据集存放的位置;

AttributeError: module ‘tensorflow’ has no attribute ‘log’

• tf.log()
• tf.math.log

TypeError: reduce_sum() got an unexpected keyword argument ‘reduction_indices’

• tf.reduce_sum(actv, reduction_indices=1)
• reduction_indices: axis的旧名称(已弃用)。
• keep_dims: keepdims的弃用别名。
• tf.reduce_sum(actv)

AttributeError: module ‘tensorflow_core._api.v2.train’ has no attribute ‘GradientDescentOptimizer’

• optm = tf.train.GradientDescentOptimizer(learning_rate).minimize(cost)
• optm = tf.compat.v1.train.GradientDescentOptimizer(learning_rate).minimize(cost)

AttributeError: module ‘tensorflow’ has no attribute ‘InteractiveSession’

• tf.InteractiveSession()
• tf.compat.v1.InteractiveSession()

ValueError: Cannot evaluate tensor using eval(): No default session is registered. Use with sess.as_default() or pass an explicit session to eval(session=sess)

• tf.compat.v1.disable_eager_execution() 与这个有关；
• 方案1

sess = tf.compat.v1.InteractiveSession()

• 方案2

# SESSION
with tf.compat.v1.Session() as sess:
    sess.run(init)

AttributeError: module ‘tensorflow’ has no attribute ‘Print’

• tf.Print(a, [a], "a: ")
• tf.compat.v1.Print(a, [a], "a: ")

AttributeError: module ‘tensorflow_core._api.v2.train’ has no attribute ‘AdamOptimizer’

• tf.train.AdamOptimizer(learning_rate=0.001).minimize(cost)
• tf.compat.v1.train.AdamOptimizer(learning_rate=0.001).minimize(cost)

NotFoundError: Restoring from checkpoint failed. This is most likely due to a Variable name or other graph key that is missing from the checkpoint. Please ensure that you have not altered the graph expected based on the checkpoint. Original error:

NotFoundError: Key v1_1 not found in checkpoint [[node save_1/RestoreV2 (defined at :7) ]]

• 不要重复运行
• 保存与读取分开运行

AttributeError: module ‘scipy.misc’ has no attribute ‘imread’

• scipy.misc.imread(path);
• SciPy1.0.0不赞成使用imread，在1.2中已经弃用;
• 可以使用import imageio content_image=imageio.imread()来代替;

Depth of input (32) is not a multiple of input depth of filter (3) for ‘conv1_1/Conv2D’ (op: ‘Conv2D’) with input shapes: [1,3,24,32], [3,3,3,64].

• tf.nn.conv2d(_input_r, _w[‘wc1’], strides=[1, 1, 1, 1], padding=‘SAME’)
• IMG_PATH = “/vgg_data/cat.png”
• IMG_PATH = “/vgg_data/cat.jpg”
• 注意图片格式：jpg

TypeError: Input ‘split_dim’ of ‘Split’ Op has type float32 that does not match expected type of int32.

• x = tf.split(0, n_steps, x) # tf.split(axis, num_or_size_splits, value)
• x = tf.split(x, n_steps, 0) # tf.split(value, num_or_size_splits, axis)

AttributeError: module ‘tensorflow’ has no attribute ‘variable_scope’

• tf.variable_scope
• tf.compat.v1.variable_scope

AttributeError: module ‘tensorflow_core._api.v2.nn’ has no attribute ‘rnn_cell’

• tf.nn.rnn_cell.BasicLSTMCell()
• tf.compat.v1.nn.rnn_cell.BasicLSTMCell
• from tensorflow.python.ops import rnn, rnn_cell

AttributeError: module ‘tensorflow_core._api.v2.nn’ has no attribute ‘rnn’

• tf.nn.rnn
• from tensorflow.python.ops import rnn, rnn_cell

AttributeError: module ‘tensorflow.python.ops.rnn’ has no attribute ‘rnn’

• _LSTM_O, _LSTM_S = rnn.rnn(lstm_cell, _Hsplit, dtype=tf.float32)
• _LSTM_O, _LSTM_S = rnn.static_rnn(lstm_cell, _Hsplit,dtype=tf.float32)

AttributeError: module ‘tensorflow’ has no attribute ‘reset_default_graph’

• tf.reset_default_graph()
• tf.compat.v1.reset_default_graph()

Variable basic/rnn/basic_lstm_cell/kernel does not exist, or was not created with tf.get_variable(). Did you mean to set reuse=tf.AUTO_REUSE in VarScope

调用了两次RNN网络，在第二次调用的时候报了上面这个错误。主要是因为第二次的变量名和第一次的变量名一样，导致了变量命名相同的冲突。在Tensorflow中有两种方法生成变量variable，tf.get_variable()和tf.Variable()。在tf.name_scope()的框架下使用这两种方法，使用tf.Variable()，尽管name一样，但为了不重复变量名，Tensorflow输出的变量名并不一样，所以本质上是不一样的变量；使用tf.get_variable()定义的变量虽然不会被tf.name_scope()中的名字影响，但在未指定共享变量时，如果重名了会报错。要实现变量共享，可以使用tf.variable_scope(reuse=tf.AUTO_REUSE)创建具有相同名称的作用域。

# 出错代码
with tf.name_scope("fw_side"):

# 改成
with tf.name_scope("fw_side"), tf.variable_scope("fw_side", reuse=tf.AUTO_REUSE):

Variable rnn/basic_lstm_cell/kernel already exists, disallowed

• 通过 with tf.variable_scope(‘scope’,reuse=True)，给不同的RNN模型定义不同的作用域
• 在代码中写上 tf.reset_default_graph()

ValueError: Variable basic11111/rnn/basic_lstm_cell/kernel does not exist, or was not created with tf.get_variable(). Did you mean to set reuse=tf.AUTO_REUSE in VarScope?

• tf.compat.v1.variable_scope(_name)；
• _name 重复运行，变量已经存在；

方案1

with tf.compat.v1.variable_scope(_name) as scope:
    # 共享变量
    scope.reuse_variables()

方案2

with tf.compat.v1.variable_scope(_name, reuse=tf.compat.v1.AUTO_REUSE) as scope:

AttributeError: module ‘tensorflow_core._api.v2.train’ has no attribute ‘GradientDescentOptimizer’

• tf.train.GradientDescentOptimizer(learning_rate).minimize(cost)
• tf.compat.v1.train.GradientDescentOptimizer(learning_rate).minimize(cost)