Search papers, labs, and topics across Lattice.
This paper introduces Implicit Reward Model (IRM), a zero-shot method for detecting LLM-generated text by leveraging implicit reward signals derived from instruction-tuned and base models. IRM eliminates the need for preference collection or task-specific fine-tuning, unlike prior reward-based detection methods. Experiments on the DetectRL benchmark show that IRM outperforms existing zero-shot and supervised methods in detecting LLM-generated text.
Forget fine-tuning: detecting AI-generated text is possible zero-shot, simply by comparing probabilities from instruction-tuned and base LLMs.
Large language models (LLMs) have demonstrated remarkable capabilities across various tasks. However, their ability to generate human-like text has raised concerns about potential misuse. This underscores the need for reliable and effective methods to detect LLM-generated text. In this paper, we propose IRM, a novel zero-shot approach that leverages Implicit Reward Models for LLM-generated text detection. Such implicit reward models can be derived from publicly available instruction-tuned and base models. Previous reward-based method relies on preference construction and task-specific fine-tuning. In comparison, IRM requires neither preference collection nor additional training. We evaluate IRM on the DetectRL benchmark and demonstrate that IRM can achieve superior detection performance, outperforms existing zero-shot and supervised methods in LLM-generated text detection.
