AI Paper Atlas

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

CGRiC: Compositional Risk Certification for Structured LLM Outputs

Ibne Farabi Shihab, Sanjeda Akter, Anuj Sharma

ICML 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
upgrade to retrieval+NLI verifier (ablation)Atomic EDFL · HotpotQA-Citeabstention rate: .42 -> .39Improved
add per-claim granularity with max-risk composition (ablation)stronger-verifier atomic baseline · HotpotQA-Citeabstention rate: .39 -> .37Improved
add union-bound composition and localized repair (ablation)per-claim retrieval+NLI baseline · HotpotQA-Citeabstention rate: .37 -> .29Improved
replace atomic EDFL with CGRiC (union)EDFL · HotpotQAabstention rate: .42 -> .29Improved
use graph-structured bound instead of union bound (ablation)CGRiC · HotpotQAabstention rate: .29 -> .26Improved
add robustness extension over skeleton family (ablation)CGRiC (union) · HotpotQAabstention rate: .29 -> .33Hurt
use CGRiC instead of EDFL under adversarial falsehoodsEDFL · HotpotQA-Cite adversarial falsehoodscertified-failure rate: .18 -> .083Improved
distribution shift from in-distribution to adversarial falsehoodsCGRiC · HotpotQA-Cite calibrationcertified-failure rate: .050 -> .083Hurt
20% drop noise in claim extraction (ablation)CGRiC · extraction noisecertified-failure rate: 0.05 -> 0.08Hurt
use lazy CGRiC fast path (ablation)CGRiC · computational overheadlatency overhead: +35-47% -> +18-26%Improved
add localized repair instead of full regeneration (ablation)CGRiC · latency overheadlatency overhead: +85% -> +42%Improved

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