Confidence Calibration and Rationalization for LLMs via Multi-Agent Deliberation
This addresses uncertainty estimation for LLMs, which is crucial for reliability in applications, but it is an incremental improvement over existing calibration methods.
The paper tackles the problem of poor calibration and over-confidence in large language models, especially after RLHF, by proposing Collaborative Calibration, a post-hoc training-free strategy that uses multi-agent deliberation to improve both accuracy and calibration on generative QA tasks.
Uncertainty estimation is a significant issue for current large language models (LLMs) that are generally poorly calibrated and over-confident, especially with reinforcement learning from human feedback (RLHF). Unlike humans, whose decisions and confidences not only stem from intrinsic beliefs but can also be adjusted through daily observations, existing calibration methods for LLMs focus on estimating or eliciting individual confidence without taking full advantage of the "Collective Wisdom": the interaction among multiple LLMs that can collectively improve both accuracy and calibration. In this work, we propose Collaborative Calibration, a post-hoc training-free calibration strategy that leverages the collaborative and expressive capabilities of multiple tool-augmented LLM agents in a simulated group deliberation process. We demonstrate the effectiveness of Collaborative Calibration on generative QA tasks across various domains, showing its potential in harnessing the rationalization of collectively calibrated confidence assessments and improving the reliability of model predictions.