ValueError: No gradients provided for any variable error when using gru unit and data with ndim = 3 with diffrent batch size

Question

Code :

from pandas import DataFrame
from pandas import concat
from keras.models import Sequential
from keras.layers import Dense
from keras.layers import GRU

# y.shape : (2325, 2, 13)
# X.shape : (2325, 10, 13)
X = X.reshape(len(X),len(x[0]),len(x[0][0]))
# configure network
n_batch = 16
n_epoch = 100
n_neurons = 16
# design network
model = Sequential()
model.add(GRU(n_neurons, batch_input_shape=(len(X),len(x[0]),len(x[0][0])), 
stateful=True))
model.add(Dense(13))
model.compile(loss=root_mean_squared_error, optimizer='adam')

for i in range(n_epoch): 
    model.fit(X, y, epochs=1, batch_size=n_batch, verbose=1, shuffle=False)
    model.reset_states()

yhat = model.predict(X, batch_size=n_batch)
for i in range(len(y)):
    print('>Expected=%.1f, Predicted=%.1f' % (y[i], yhat[i]))

Hello, I have a problem with this code. I want to use batch size in GRU unit but I get this error :

ValueError: No gradients provided for any variable: ['gru_20/gru_cell_20/kernel:0', 'gru_20/gru_cell_20/recurrent_kernel:0', 'gru_20/gru_cell_20/bias:0', 'dense_20/kernel:0', 'dense_20/bias:0'].