Project - Predict the hourly rain gauge total

Hello Team,
I am struggling mid way of the Project specially with respect to creating LSTM Model. First referring to the Kaggle site from where we downloaded the dataset, I was trying to build models as mentioned using the code below:

On execution CuDNNLSTM do not seem to work. Also If I try to use:
lstm_cells = [tf.contrib.rnn.BasicLSTMCell(num_units=n_neurons)
for layer in range(n_layers)]

Then here I would need inputs such as:

n_steps = ?
n_inputs = ?
n_neurons = ?
n_outputs = ?
n_layers = ?


def get_model_simple(shape=(19,22)):
inp = Input(shape)
x = CuDNNLSTM(64, return_sequences=False)(inp)
x = Dense(1)(x)

model = Model(inp, x)
return model

model_0 = get_model_simple((19,22))
model_0.compile(optimizer=‘adadelta’, loss=‘mae’)
model_0.summary()


If anyone has done this project can you please help guide?

Appreciate your help!
Thanks & regards