Keras Model Gradient Descent

def get_optimizer(name=None):
    if name is not None:
        if name.lower() == 'sgd':
            return keras.optimizers.SGD(lr=0.01, decay=0.1, momentum=0.1, nesterov=False)
    # Use RMSProp by default
    return keras.optimizers.RMSprop(lr=0.01, rho=0.9, epsilon=1e-08)

try:
    n, d, x, y, q_elist_map, f_names = extract_input_output(feature_file)
    # Create the model
    num_epochs = 1000
    weights_init = 'uniform'
    activation = 'sigmoid'
    optimizer_name = 'RMSProp'
    #optimizer_name = 'SGD'

    np.random.seed(1)    # For reproducibility of results
    model = Sequential()
    model.add(Dense(input_dim=d, output_dim=1, init=weights_init, activation=activation))
    model.compile(loss='mse', optimizer=get_optimizer(optimizer_name))

    # Train the model
    history = model.fit(x, y, nb_epoch=num_epochs, batch_size=20, verbose=0, validation_split=0.1)
    # Evaluate the model (print Evaluation score)
    score = model.evaluate(x, y)

except Exception as e:
    print 'Exception caught!'
    traceback.print_exc()