How does it work if we have different "time-steps" and "LSTM-units"?

Asked Mar 01 '19 at 10:27

Active Mar 01 '19 at 12:36

Viewed 82 times

As I know we can create a LSTM layer like: model.add(LSTM(units), input_shape = (time-steps, feature_number)) And the units number and time-steps number can be different numbers! So how does it work? For ex, if I specify my LSTM like this: model.add(LSTM(10), input_shape = (1, feature_number))

How will it feed the network? Shouldn't each unit of LSTM be 1 time-step? So how is it possible when I have 10 units but 1 time-step?

edited Mar 01 '19 at 12:36

asked Mar 01 '19 at 10:27

user3486308

1,270
5
18
28

units refers do one recurrent neuron or node. If you have 10 timesteps, each of the neurons would run the 10 timesteps. – Shubham Panchal Mar 01 '19 at 12:21
@ShubhamPanchal: Please look at the picture that I have added. It seems each time-step relates to only 1 unit! – user3486308 Mar 01 '19 at 12:37

How does it work if we have different "time-steps" and "LSTM-units"?

0 Answers0