Calibrate to Discriminate:Improve In-Context Learning with Label-Free Comparative Inference

Wei Cheng, Tianlu Wang, Yanmin Ji, Keren Tan, Fan Yang, Yiyu Zheng

Keywords: Calibration. In-context learning. Comparative inference.

Abstract: While in-context learning with large language models (LLMs) has shown impressive performance, we have discovered a unique miscalibration behavior where both correct and incorrect predictions are assigned the same level of confidence. We refer to this phenomenon as \textit{indiscriminate miscalibration}. We found that traditional calibration metrics, such as Expected Calibrated Errors (ECEs), are unable to capture this behavior effectively. To address this issue, we propose new metrics to measure the severity of indiscriminate miscalibration. Additionally, we develop a novel in-context comparative inference method to alleviate miscalibrations and improve classification performance. Through extensive experiments on five datasets, we demonstrate that our proposed method can achieve more accurate and calibrated predictions compared to regular zero-shot and few-shot prompting.

Primary Area: applications to computer vision, audio, language, and other modalities

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Submission Number: 12386