The Transformer is a neural network architecture that processes a whole sequence at once using self-attention instead of stepping through it one token at a time. It solves the problem that recurrent networks could not parallelise training and struggled to relate distant positions in a sequence. It exists because attention alone turned out to be sufficient for translation quality, and because removing recurrence made training scale across many accelerators. Almost every large language model in use today is a Transformer variant, which makes it the single most load-bearing node in this graph.
A recurrent neural network processes a sequence one step at a time, carrying a hidden state that summarises what it has seen. It solves the problem that a fixed-size input cannot represent a sequence of arbitrary length. It exists because order matters in language and time series. Training it is slow and unstable over long sequences, which is the limitation the Transformer was designed to remove.
