Knowing the weights biases of each hidden and output layer of a DNN

by: cyberabc, 7 years ago

Last edited: 7 years ago

Hi,

I've seen the tutorials related to deep neural network and modelling that with tensorflow and python. Since I'm a beginner at tensorflow, I don't get how I could access the weights and biases of each hidden layer and output layer.

Thanks in advance for your kind help



You must be logged in to post. Please login or register an account.



This is possible, but I am not sure why you'd want the raw values. What are you hoping to do with them?

You can begin to peak into all of the variables with tf.all_variables()

You can also iterate around by layer, since you referenced that specifically, like:

with tf.variable_scope('layer1', reuse=True):
  tf.get_variable('weights')


Stuff like that. If you google around for getting weights directly, you certainly can find lots of examples.




-Harrison 7 years ago

You must be logged in to post. Please login or register an account.


Thanks Harrison. That was helpful. Actually, I want the weights for a little bit research :-)

-cyberabc 7 years ago

You must be logged in to post. Please login or register an account.