Access to internal shutter forget value in lstm node

Question

Currently, I have created an LSTM network using Keras and should get the internal delay forgetting values ​​for each node of the network. I have to get the shutter / forget values ​​every time the calculate () function is called. can this be done?

If not, is there any other library that provides access to these internal shutter values ​​in a convenient way at every step?

Looking forward to some help on this. Thank.

Answer 1

How to get internal weights and parameters?

Keras Tensorflow ( ), (, ) tf.global_variables(). , tf.Variable:

variable_dict = dict([])
for variable in tf.global_variable():
    variable_dict[variable.name] = variable

. LSTM Keras (, model.summary() , LSTM 'lstm_1') : 'lstm_1/kernel:0', 'lstm_1/recurrent_kernel:0' 'lstm_1/bias:0'. , Session Tensor Tensorflow.

: Keras ? - , , , . LSTMCell, build (, ) call (, ). , , *_i, *_f, *_c, *_o (, self.kernel_o).

---

?

, . , : Tensorflow - .

---

, , ?

, Keras Tensorflow . , .