Getting the tensor values of a layer after calling model.predict once

Question

How would i get the tensor of the output activations of one of the layers in a neural network after running the following code.

  x= model.predict(frame)

Answer 1

Here's a short example of how you could do it using Keras' functional API:

inputs = Input(shape=(10,))
x = Dense(5)(inputs)
latent_outputs = Dense(2)(x)
x = Dense(5)(latent_outputs)
outputs = Dense(10)(x)

encoder = Model(inputs, latent_outputs)
autoencoder = Model(inputs, outputs)

autoencoder.compile(optimizer='adam', loss='mse')
autoencoder.fit(np.zeros((10, 10)), np.zeros((10, 10)))

latent_output = encoder.predict(np.zeros((1, 10)))
print(latent_output)