Extract number from tf.Tensor model

Question

I am using softmax to get the probability vector from a CNN-MFCC based model. I've started to use the command

model(prediction_feature)

instead of

 model.predict(prediction_feature)

as I kept getting warning messages while passing the model.predict command over a loop. However, now when I record the probability, the value gets returned in the form:

tf.Tensor(0.00087791355, shape=(), dtype=float32)

instead for just the number 0.00087791355 when I use the model.predict command.

Is there a way to extract just the number from the model(prediction_feature) approach?