seq2seq-encoder-decoder-2014

IN premise — entries/2026/06/21/wiki-Recurrent_neural_network-chunk-1.md

Created 2026-06-21T09:55:52+00:00

The encoder-decoder (seq2seq) architecture uses two RNNs (typically LSTMs) for sequence transduction, originated circa 2014, and was instrumental in developing attention mechanisms that led to the transformer architecture.