The Goldilocks of Pragmatic Understanding: Fine-Tuning Strategy Matters for Implicature Resolution by LLMs

TL;DR

This paper investigates how different fine-tuning strategies affect large language models' ability to understand pragmatic inferences, revealing that instruction tuning significantly improves their contextual interpretative skills.

Contribution

It demonstrates that instruction-tuned LLMs outperform other models in pragmatic inference tasks, highlighting the importance of fine-tuning strategies for contextual understanding.

Findings

Instruction-tuned models perform significantly better on implicature tasks.

Most models perform close to random on pragmatic inference without specific tuning.

Fine-tuning strategies greatly influence models' ability to interpret language in context.

Abstract

Despite widespread use of LLMs as conversational agents, evaluations of performance fail to capture a crucial aspect of communication: interpreting language in context -- incorporating its pragmatics. Humans interpret language using beliefs and prior knowledge about the world. For example, we intuitively understand the response "I wore gloves" to the question "Did you leave fingerprints?" as meaning "No". To investigate whether LLMs have the ability to make this type of inference, known as an implicature, we design a simple task and evaluate four categories of widely used state-of-the-art models. We find that, despite only evaluating on utterances that require a binary inference (yes or no), models in three of these categories perform close to random. However, LLMs instruction-tuned at the example-level perform significantly better. These results suggest that certain fine-tuning strategies are far better at inducing pragmatic understanding in models. We present our findings as the starting point for further research into evaluating how LLMs interpret language in context and to drive the development of more pragmatic and useful models of human discourse.