Post-training for Efficient Communication via Convention Formation
read the original abstract
Humans communicate with increasing efficiency in multi-turn interactions, by adapting their language and forming ad-hoc conventions. In contrast, prior work shows that LLMs do not naturally show this behavior. We develop a post-training process to develop this ability through targeted fine-tuning on heuristically identified demonstrations of convention formation. We evaluate with two new benchmarks focused on this capability. First, we design a focused, cognitively-motivated interaction benchmark that consistently elicits strong convention formation trends in humans. Second, we create a new document-grounded reference completion task that reflects in-the-wild convention formation behavior. Our studies show significantly improved convention formation abilities in post-trained LLMs across the two evaluation methods.
This paper has not been read by Pith yet.
Forward citations
Cited by 1 Pith paper
-
LVLMs and Humans Ground Differently in Referential Communication
LVLMs and humans build common ground differently in referential tasks, with LVLMs failing to enable smooth interactive communication in a new corpus of 356 dialogues.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.