paper-with-me

홈 › Papers

Calibrating LLMs with Information-Theoretic Evidential Deep Learning

2025-02-10 · Yawei Li, David Rügamer, Bernd Bischl, Mina Rezaei

Fine-tuned large language models (LLMs) often exhibit overconfidence, particularly when trained on small datasets, resulting in poor calibration and inaccurate uncertainty estimates. Evidential Deep Learning (EDL), an uncertainty-aware approach, enables uncertainty estimation in a single forward pass, making it a promising method for calibrating fine-tuned LLMs. However, despite its computational efficiency, EDL is prone to overfitting, as its training objective can result in overly concentrated probability distributions. To mitigate this, we propose regularizing EDL by incorporating an information bottleneck (IB). Our approach IB-EDL suppresses spurious information in the evidence generated by the model and encourages truly predictive information to influence both the predictions and uncertainty estimates. Extensive experiments across various fine-tuned LLMs and tasks demonstrate that IB-EDL outperforms both existing EDL and non-EDL approaches. By improving the trustworthiness of LLMs, IB-EDL facilitates their broader adoption in domains requiring high levels of confidence calibration. Code is available at https://github.com/sandylaker/ib-edl.

📄 PDF Abstract BibTeX arXiv:2502.06351

Code (1)

sandylaker/ib-edl 공식 구현 pytorch

Tasks

Computational EfficiencyDeep Learning

Similar Papers 제목 키워드 기반

Improving Group Robustness on Spurious Correlation via Evidential Alignment

2025-06-12 · Wenqian Ye, Guangtao Zheng, Aidong Zhang

Deep neural networks often learn and rely on spurious correlations, i.e., superficial associations between non-causal features and the targets. For instance, an image classifier may identify camels based on the desert ba…

Uncertainty Quantification

Utilizing and Calibrating Hindsight Process Rewards via Reinforcement with Mutual Information Self-Evaluation

2026-04-13 · Jiashu Yao, Heyan Huang, Zeming Liu, Yuhang Guo arxiv

To overcome the sparse reward challenge in reinforcement learning (RL) for agents based on large language models (LLMs), we propose Mutual Information Self-Evaluation (MISE), an RL paradigm that utilizes hindsight genera…

Reinforcement Learning

Benchmarking Source-Sensitive Reasoning in Turkish: Humans and LLMs under Evidential Trust Manipulation

2026-04-27 · Sercan Karakaş, Yusuf Şimşek arxiv

This paper investigates whether source trustworthiness shapes Turkish evidential morphology and whether large language models (LLMs) track this sensitivity. We study the past-domain contrast between -DI and -mIs in contr…

Learn to Accumulate Evidence from All Training Samples: Theory and Practice

2023-06-19 · Deep Pandey, Qi Yu

Evidential deep learning, built upon belief theory and subjective logic, offers a principled and computationally efficient way to turn a deterministic neural network uncertainty-aware. The resultant evidential models can…

All

Why LLMs Hallucinate, and How to Get (Evidential) Closure: Perceptual, Intensional, and Extensional Learning for Faithful Natural Language Generation

2023-10-23 · Adam Bouyamourn

We show that LLMs hallucinate because their output is not constrained to be synonymous with claims for which they have evidence: a condition that we call evidential closure. Information about the truth or falsity of sent…

Language ModelingLanguage ModellingText Generation