In keras based LSTM, Is it ok feature size of input shape is smaller than the data size?

Question

If the shape of my training set is [Sample=1, timestep=6, features=6], if we put (Sample=1, timestep=6, features=1) in batch input shape on keras, will