Model ordered data with recurrent networks, gated units, temporal convolutions, and sequence-to-sequence architectures. Handle variable-length inputs, masking, teacher forcing, and long-range dependencies.
