Language Models Can be Efficiently Steered via Minimal Embedding Layer Transformations.

Diogo Tavares, David Semedo, Alexander Rudnicky, Joo Magalhes

Browse the full EMNLP paper archive.