AI Paper Atlas

Research map / Large language models / Agents, code and math

Decompose, Structure, and Repair: A Neuro-Symbolic Framework for Autoformalization via Operator Trees

Xiaoyang Liu, Zineng Dong, Yifan Bai, Yantao Li, Yuntian Liu, Tao Luo

ICML 2026 · Large language models › Agents, code and math · approach: Formal verification and proof data · 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
add operator tree supervision to the component formalizer (ablation)DSR Formalizer (Pass@1) · ProverBenchCC pass rate: 21.54 -> 24.92Improved
add operator tree supervision without curriculum (ablation)DSR Formalizer (Pass@1) · PRIMESC pass rate: 22.44 -> 19.87Hurt
add curriculum learning on top of operator trees (ablation)DSR Formalizer (Pass@1) · PRIMESC pass rate: 19.87 -> 23.08Improved
add curriculum learning on top of operator trees (ablation)DSR Formalizer (Pass@1) · ProofNetCC pass rate: 11.86 -> 16.44Improved
add operator tree supervision without curriculum (ablation)DSR Formalizer (Pass@4) · PRIMECC pass rate: 25.00 -> 21.15Hurt
add tree-guided repair instead of statement-level global repair (ablation)DSR · ProofNetCC pass rate: 76.01 -> 79.51Improved
add tree-guided repair instead of statement-level global repair (ablation)DSR · ProverBenchCC pass rate: 82.77 -> 84.00Improved
add tree-guided repair instead of statement-level global repair (ablation)DSR · PRIMESC pass rate: 83.97 -> 80.13Hurt
add tree-guided repair instead of statement-level global repair (ablation)DSR · ProofNetSC pass rate: 89.76 -> 87.33Hurt
add operator tree supervision to end-to-end statement model (ablation)End-to-end model (Pass@1) · ProofNetSC pass rate: 65.77 -> 70.89Improved
add operator tree supervision to end-to-end statement model (ablation)End-to-end model (Pass@1) · PRIMECC pass rate: 44.23 -> 40.38Hurt
add curriculum learning to end-to-end statement model (ablation)End-to-end model (Pass@1) · ProofNetCC pass rate: 41.51 -> 39.08No clear effect

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