AI Paper Atlas

Research map / Large language models / Language, safety and interpretability

From Sampling to Cognition: Modeling Internal Cognitive Confidence in Language Models for Robust Uncertainty Calibration

Hao Li, Tao He, Jiafeng Liang, Zheng Chu, Ming Liu

AAAI 2026 · Large language models › Language, safety and interpretability · approach: Uncertainty quantification for LLMs · 1 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
supervise with CogConf instead of sampling accuracy (ablation)CogAlign Llama-3-8B · TriviaQA+CommonsenseQAECE: 10.2 -> 7.5Improved
supervise with CogConf instead of sampling accuracy (ablation)CogAlign Mistral-7B · TriviaQA+CommonsenseQAAUROC: 70.2 -> 72.7Improved
supervise with CogConf instead of semantic entropy (ablation)CogAlign Llama-3-8B · TriviaQA+CommonsenseQAECE: 11.5 -> 7.5Improved
supervise with CogConf instead of semantic entropy (ablation)CogAlign Mistral-7B · TriviaQA+CommonsenseQAECE: 15.4 -> 7.8Improved
supervise with CogConf instead of sampling accuracy (ablation)CogAlign Llama-3-8B · SportQAECE: 21.3 -> 14.0Improved
supervise with CogConf instead of semantic entropy (ablation)CogAlign Mistral-7B · ScienceQAECE: 15.6 -> 6.2Improved
supervise with CogConf instead of sampling accuracy (ablation)CogAlign · IK vs IDK confidence gapconfidence gap: 1.17 -> 3.70Improved
supervise with CogConf instead of semantic entropy (ablation)CogAlign · IK vs IDK confidence gapconfidence gap: 0.72 -> 3.70Improved
CogAlign instead of RewardingDoubtcalibration RL · Llama-3-8B in-domainECE: 9.8 -> 7.5Improved
CogAlign instead of RewardingDoubtcalibration RL · Mistral-7B in-domainAUROC: 62.7 -> 72.7Improved
CogAlign instead of RewardingDoubtcalibration RL · Mistral-7B ARC-ChallengeECE: 3.2 -> 6.2Hurt
CogAlign instead of RewardingDoubtcalibration RL · Llama-3-8B SportQAECE: 7.5 -> 14.0Hurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy