CODEWITHSUNDEEP
pythonkerasactivation
hyper
hyper

Reputation: 21

Keras give error: ValueError: ('Could not interpret activation function identifier:'

Keras gives the error : ValueError: ('Could not interpret activation function identifier:', [, ])

importing the activations: 

from keras.activations import relu, elu, linear, sigmoid

Defining grid of parameters: 

params = {'lr': [0.001, 0.1],
    'first_neuron':[5,9],
     'hidden_layers':[1,5,10],
     'batch_size': [30,40,50],
     'epochs': [40],
     'dropout': [0,0.2],
     'kernel_initializer': ['normal'],
     'optimizer': [Adam],
     #'loss':[mean_absolute_error],
     'activation':[],
     'last_activation':['linear']

     }

Calling a model :

regression__model = regression_model(X_air_train, y_air_train, X_air_valid, y_air_valid, params)

Upvotes: 2

Views: 5726

Answers (2)

user6688
user6688

Reputation: 29

Your activation parameter is empty.

To solve this...

from keras.activations import *

Note: the (*) sign allows you to use any of the activation functions in Keras without the need to specify their names.

params = {'lr': [0.001, 0.1],
'first_neuron':[5,9],
 'hidden_layers':[1,5,10],
 'batch_size': [30,40,50],
 'epochs': [40],
 'dropout': [0,0.2],
 'kernel_initializer': ['normal'],
 'optimizer': ['Adam'],  ## << notice the added quotation marks
 #'loss':[mean_absolute_error],
 'activation':['relu', 'elu', 'sigmoid'],
 'last_activation':['linear']

 }

Upvotes: 1

user10595337
user10595337

Reputation:

Activation parameter is empty.

Upvotes: 0

Related Questions