AI Paper Atlas

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

StepFun-Formalizer: Unlocking the Autoformalization Potential of LLMs Through Knowledge-Reasoning Fusion

Yutong Wu, Di Huang, Ruosi Wan, Yue Peng, Shijie Shang, Chenrui Cao, Lei Qi, Rui Zhang, Xishan Zhang, Zidong Du, Jie Yan, Xing Hu

AAAI 2026 · Large language models › Agents, code and math · approach: Inductive and abstract reasoning evaluation · 3 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 stage-1 formal knowledge SFT data (ablation)StepFun-Formalizer-7B · ProverBenchBEq@1 (%): 24.5 -> 25.1Improved
add stage-1 formal knowledge SFT data (ablation)StepFun-Formalizer-7B · CombiBenchBEq@1 (%): 3.9 -> 5.2Improved
add informal-to-formal reasoning SFT data (ablation)StepFun-Formalizer-7B · ProverBenchBEq@16 (%): 25.3 -> 37.9Improved
add informal-to-formal reasoning SFT data (ablation)StepFun-Formalizer-7B · ProverBenchBEq@1 (%): 21.8 -> 25.1Improved
add informal-to-formal reasoning SFT data (ablation)StepFun-Formalizer-7B · CombiBenchBEq@1 (%): 5.3 -> 5.2Hurt
use expert template-guided reasoning instead of direct distillation from Claude4-thinking (ablation)StepFun-Formalizer-7B · ProverBenchBEq@1 (%): 21.8 -> 25.1Improved
use expert template-guided reasoning instead of direct distillation from Claude4-thinking (ablation)StepFun-Formalizer-7B · ProverBenchBEq@16 (%): 33.3 -> 37.9Improved
add LLM-based validity selection of knowledge data (ablation)StepFun-Formalizer-7B · ProverBenchBEq@16 (%): 35.6 -> 37.9No clear effect
add LLM-based validity selection of knowledge data (ablation)StepFun-Formalizer-7B · ProverBenchBEq@1 (%): 25.4 -> 25.1No clear effect
start from DeepSeek-R1-Distill-Qwen-7B instead of Kimina-Autoformalizer-7B (ablation)StepFun-Formalizer-7B · ProverBenchBEq@1 (%): 21.4 -> 25.1Improved
majority voting over 16 samplesKimina-Autoformalizer-7B · FormalMATH-LiteBEq (%): 35.1 -> 45.9Improved
add RLVR with BEq reward after SFT (ablation)StepFun-Formalizer-7B · average of benchmarksaverage BEq@1: 0.258 -> 0.303Improved

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