1. Foreword

Looking up a lot of information, I found that the originator of this error report was tf.concat “Expected int32, got a list containing tendors of type”_ The reason and solution of “message ‘instead.”

2. Reasons

The specific analysis of errors caused by TensorFlow version is as follows:

In TensorFlow versions before 1.0, the parameter order of concat() is number first and tensors second. The example is as follows:

tf.concat(3,pooled_outputs_i)

For versions after TensorFlow 1.0, the parameter order of concat() is tensors first and numbers second. For example:

tf.concat(pooled_outputs_i,3)

3. Solutions

Smart as you, you should know how to debug when you see the two explanations in the reason. That is to say, adjust the order of concat parameters and everything will be OK