AI Paper Atlas

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

SAFER: Risk-Constrained Sample-then-Filter in Large Language Models

Qingni Wang, Yue Fan, Xin Eric Wang

ICLR 2026 · Large language models › Language, safety and interpretability · approach: Uncertainty quantification for LLMs

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
add abstention-aware sampling budget calibration (Clopper-Pearson) instead of TRON's calibrationSAFER sampling stage · TriviaQA and CoQA, five LLMstest-time EER: TRON EER exceeding 0.06 at alpha=0.03 vs SAFER below alphaImproved
add conformalized filtering with risk level beta (ablation)SAFER prediction sets · TriviaQA, LLaMA-3.1-8B-Instructaverage prediction set size: 7.9 -> 5.5Improved
increase beta in the filtering stage (ablation)SAFER · TriviaQA and CoQAprediction set size and EER: EER stays below alpha+beta-alpha*beta boundMixed
use Rouge-L instead of sentence similarity as the correctness criterion (ablation)SAFER · TriviaQA and CoQAtest-time EERNo clear effect
reduce calibration-to-test split ratio from 0.5 to 0.1 (ablation)SAFER · TriviaQA, sentence similaritytest-time EER: 0.5 -> 0.1No clear effect
filter unreliable answers with entropy-based thresholdSAFER · TriviaQA and CoQAempirical coverageMixed

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