AI Paper Atlas

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

Conditional Factuality Controlled LLMs with Generalization Certificates via Conformal Sampling

Kai Ye, Qingtao Pan, Shuo Li

CVPR 2026 · Large language models › Language, safety and interpretability · approach: Statistical frameworks for LLM evaluation · 0 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
replace global ICP threshold with feature-conditional CFC thresholdconformal factuality · Synthetic data (alpha=0.10)GSC (%): 57.4 -> 88.7Improved
add exact conformal correction on top of a learned conditional threshold (CFC vs Learnt CP) (ablation)conditional threshold · Synthetic data (alpha=0.10)GSC (%): 84.3 -> 88.7Improved
replace ICP with CFCLlama-3-8B-Instruct · GSM8K (alpha=0.05)APSS: 4.73 -> 2.35Improved
replace ICP with CFCLlama-3-8B-Instruct · TriviaQA (alpha=0.30)GSC (%): 49.3 -> 62.8Improved
add PAC stability adjustment (CFC-P-F instead of CFC) (ablation)CFC · TriviaQA (alpha=0.30)ECR toward target 70%: 68.4 -> 70.0Improved
add PAC stability adjustment (CFC-P-F instead of CFC) (ablation)CFC · GSM8K (alpha=0.05)APSS: 2.35 -> 4.59Hurt
increase sample budget from 5 to 20 candidates (ablation)CFC on GSM8K · GSM8K (alpha=0.10)GSC (%): 86.36 -> 87.73No clear effect
increase sample budget from 5 to 20 candidates (ablation)CFC on GSM8K · GSM8K (alpha=0.10)APSS: 1.49 -> 3.79Hurt
use chosen entropy-and-loss feature map instead of entropy-linear (ablation)CFC on TriviaQA · TriviaQA (alpha=0.30)GSC (%): 45.1 -> 62.8Improved
use chosen feature map instead of max-loss-linear (ablation)CFC on TriviaQA · TriviaQA (alpha=0.30)GSC (%): 56.5 -> 62.8Improved
apply CFC-P-F to a vision-language model (Qwen2-VL) instead of ICPFlickr8k captioning · Flickr8k VLM (alpha=0.03)GSC (%): 85.21 -> 95.21Improved
use plain CFC instead of CFC-P-F on easy Flickr8k (ablation)CFC · Flickr8k VLM (alpha=0.03)ECR toward target 97%: 97.27 -> 95.81Hurt

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