CODEWITHSUNDEEP
pythontensorflowkeras
Sachin Gupta
Sachin Gupta

Reputation: 39

'tensorflow' has no attribute 'Session'

I am trying to convert a Tensor to a numpy array. The tensor i have has a shape as below

LastDenseLayer.output.shape
TensorShape([None, 128])

When i am running the code as below, 

with tf.Session() as sess:
    LastLayer = LastDenseLayer.output.eval()

getting the below error

I am running Keras model and trying to get the values of a specific layer out of that.

Unable to understand that is wrong here.

Regards Sachin

Upvotes: 0

Views: 971

Answers (2)

SagRU
SagRU

Reputation: 448

It's recommended to update your code to meet Tensorflow 2.0 requirements. As a quick solution you can use tf.compat.v1.Session() instead of tf.Session():

with tf.compat.v1.Session() as sess:
    LastLayer = LastDenseLayer.output.eval()

Upvotes: 0

jkr
jkr

Reputation: 19260

TensorFlow 2.x removed tf.Session because eager execution is now a default. Please refer to the TensorFlow migration guide for more information.

Upvotes: 1

Related Questions