Self-ensemble: Mitigating Confidence Mis-calibration for Large Language Models

TL;DR

This paper introduces Self-ensemble, a method to improve confidence calibration in large language models for multiple-choice questions, especially with many options, by aggregating predictions across groups without extra training.

Contribution

The paper proposes a plug-and-play Self-ensemble approach that mitigates confidence distortion in LLMs during multi-choice tasks without needing labeled data for tuning.

Findings

Self-ensemble improves LLM accuracy on multi-choice questions.

It reduces over-confidence and under-confidence issues.

Outperforms standard inference and baseline methods.

Abstract

Although Large Language Models (LLMs) perform well in general fields, they exhibit a confidence distortion problem on multi-choice question-answering (MCQA), particularly as the number of answer choices increases. Specifically, on MCQA with many choices, LLMs suffer from under-confidence in correct predictions and over-confidence in incorrect ones, leading to a substantially degraded performance. To solve this problem, we propose Self-ensemble in this work. Our method splits the choices into several groups and ensembles LLM predictions across these groups to reach a final decision. The advantage of Self-ensemble is its plug-and-play nature, where it can be integrated into existing LLM architecture based on a designed attention mask and positional encoding, without requiring labeled datasets for parameter tuning. Experimental results on three LLMs and datasets demonstrate that Self-ensemble comprehensively addresses the confidence distortion problem of LLMs, outperforming standard inference as well as baseline methods.