HalluCana: Fixing LLM Hallucination with A Canary Lookahead
This addresses the critical issue of unreliable outputs in LLMs for applications requiring factual accuracy, though it appears incremental as it builds on existing detection techniques.
The paper tackles the problem of factuality hallucinations in Large Language Models during long-form generation by introducing HalluCana, a canary lookahead method that detects and corrects hallucinations as they emerge. The method improves generation quality by up to 2.5x on biography generation while reducing compute usage by over 6 times.
In this paper, we present HalluCana, a canary lookahead to detect and correct factuality hallucinations of Large Language Models (LLMs) in long-form generation. HalluCana detects and intervenes as soon as traces of hallucination emerge, during and even before generation. To support timely detection, we exploit the internal factuality representation in the LLM hidden space, where we investigate various proxies to the LLMs' factuality self-assessment, and discuss its relation to the models' context familiarity from their pre-training. On biography generation, our method improves generation quality by up to 2.5x, while consuming over 6 times less compute.