TF版本升级问题集合:成功解决由于不同的TensorFlow版本之间某些函数的用法引起的错误

TF版本升级问题集合:成功解决由于不同的TensorFlow版本之间某些函数的用法引起的错误


遇到的问题以及对应解决方法

Transitioning to TensorFlow 1.0

TensorFlow升级导致的错误,TensorFlow 1.0后与以前代码不兼容,所以要根据API去修改。

参照外文:https://www.tensorflow.org/install/migration

1、解决问题AttributeError: module 'tensorflow' has no attribute 'histogram_summary'

将tf.histogram_summary改为tf.summary.histogram
成功解决AttributeError: module 'tensorflow' has no attribute 'histogram_summary'

2、解决问题AttributeError: module 'tensorflow' has no attribute 'scalar_summary'

将 tf.scalar_summary 改为tf.summary.scalar  

3、解决问题AttributeError: module 'tensorflow' has no attribute 'merge_all_summaries'

将 tf.merge_all_summaries() 改为tf.summary.merge_all()

4、解决问题AttributeError: module 'tensorflow.python.training.training' has no attribute 'SummaryWriter'

将tf.train.SummaryWriter 改为tf.summary.FileWriter

5、解决问题Instructions for updating:  Use `tf.global_variables_initializer` instead.

将tf.initialize_all_variables改为tf.global_variables_initializer

6、解决问题AttributeError: module 'tensorflow.python.ops.nn' has no attribute '_seq2seq'

将tf.nn.seq2seq.sequence_loss_by_example 改为tf.contrib.legacy_seq2seq.sequence_loss_by_example

7、将tf.nn.rnn_cell.改为 tf.contrib.rnn

8、解决问题:AttributeError: module 'tensorflow' has no attribute 'mul'

将tf.mul(input1, input2) 改为 tf.multiply(input1, input2) 
TF版本升级问题:解决AttributeError: module 'tensorflow' has no attribute 'mul'

9、解决问题:AttributeError: module 'tensorflow' has no attribute 'sub'

将tf.sub改为tf.subtract

10、解决问题:
tf.merge_all_summaries() 改为 summary_op = tf.summaries.merge_all()

11、解决问题:AttributeError: 'module' object has no attribute 'SummaryWriter'
tf.train.SummaryWriter 改为 tf.summary.FileWriter

12、解决问题:AttributeError: module 'tensorflow.nn.rnn_cell' has no attribute 'linear'
tf.nn.rnn_cell 改为  tf.contrib.rnn.core_rnn_cell

相关文章
tensorflow.models.rnn.rnn_cell.linear在tensorflow1.0版本之后找不到(附tensorflow1.0 API新变化)

(0)

相关推荐