Autoregressive generation

The decoder's mode of operation: generate one token at a time, feed the generated token back as input, generate the next. The causal mask ensures that at training time each position can only see previous positions (no cheating by looking at future tokens).

Producing text one token at a time, feeding each generated token back in to predict the next. How LLMs write. (M01)

Producing output one token at a time, each new token depending on all previous ones. The root cause of most serving challenges.