Small Encoders Can Rival Large Decoders in Detecting Groundedness

TL;DR

This paper demonstrates that small, task-specific encoder models like RoBERTa and NomicBERT can effectively detect whether responses are grounded in context, rivaling larger LLMs while significantly reducing inference time.

Contribution

The study shows that lightweight encoders fine-tuned for groundedness detection can match large LLMs' accuracy, offering a more efficient solution.

Findings

Encoder models achieve comparable accuracy to large LLMs.

Groundedness detection reduces inference latency significantly.

Fine-tuned encoders are effective for factual consistency tasks.

Abstract

Augmenting large language models (LLMs) with external context significantly improves their performance in natural language processing (NLP) tasks. However, LLMs struggle to answer queries reliably when the provided context lacks information, often resorting to ungrounded speculation or internal knowledge. Groundedness - generating responses strictly supported by the context - is essential for ensuring factual consistency and trustworthiness. This study focuses on detecting whether a given query is grounded in a document provided in context before the costly answer generation by LLMs. Such a detection mechanism can significantly reduce both inference time and resource consumption. We show that lightweight, task specific encoder models such as RoBERTa and NomicBERT, fine-tuned on curated datasets, can achieve accuracy comparable to state-of-the-art LLMs, such as Llama3 8B and GPT4o, in groundedness detection while reducing inference latency by orders of magnitude. The code is available at : https://github.com/chandarlab/Hallucinate-less