TheoremForge: Scaling up Formal Data Synthesis with Low-Budget Agentic Workflow

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Tao, Yicheng, Xu, Hongteng
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912845304692736
author Tao, Yicheng
Xu, Hongteng
author_facet Tao, Yicheng
Xu, Hongteng
contents The high cost of agentic workflows in formal mathematics hinders large-scale data synthesis, exacerbating the scarcity of open-source corpora. To address this, we introduce \textbf{TheoremForge}, a cost-effective formal data synthesis pipeline that decomposes the formalization process into five sub-tasks, which are \textit{statement formalization}, \textit{proof generation}, \textit{premise selection}, \textit{proof correction} and \textit{proof sketching}. By implementing a \textit{Decoupled Extraction Strategy}, the workflow recovers valid training signals from globally failed trajectories, effectively utilizing wasted computation. Experiments on a 2,000-problem benchmark demonstrate that TheoremForge achieves a Verified Rate of 12.6\%, surpassing the 8.6\% baseline, at an average cost of only \textbf{\$0.481} per successful trajectory using Gemini-3-Flash. Crucially, our strategy increases data yield by \textbf{1.6$\times$} for proof generation compared to standard filtering. These results establish TheoremForge as a scalable framework for constructing a data flywheel to train future expert models. Our code is available \href{https://github.com/timechess/TheoremForge}{here}.
format Preprint
id arxiv_https___arxiv_org_abs_2601_17332
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle TheoremForge: Scaling up Formal Data Synthesis with Low-Budget Agentic Workflow
Tao, Yicheng
Xu, Hongteng
Artificial Intelligence
The high cost of agentic workflows in formal mathematics hinders large-scale data synthesis, exacerbating the scarcity of open-source corpora. To address this, we introduce \textbf{TheoremForge}, a cost-effective formal data synthesis pipeline that decomposes the formalization process into five sub-tasks, which are \textit{statement formalization}, \textit{proof generation}, \textit{premise selection}, \textit{proof correction} and \textit{proof sketching}. By implementing a \textit{Decoupled Extraction Strategy}, the workflow recovers valid training signals from globally failed trajectories, effectively utilizing wasted computation. Experiments on a 2,000-problem benchmark demonstrate that TheoremForge achieves a Verified Rate of 12.6\%, surpassing the 8.6\% baseline, at an average cost of only \textbf{\$0.481} per successful trajectory using Gemini-3-Flash. Crucially, our strategy increases data yield by \textbf{1.6$\times$} for proof generation compared to standard filtering. These results establish TheoremForge as a scalable framework for constructing a data flywheel to train future expert models. Our code is available \href{https://github.com/timechess/TheoremForge}{here}.
title TheoremForge: Scaling up Formal Data Synthesis with Low-Budget Agentic Workflow
topic Artificial Intelligence
url https://arxiv.org/abs/2601.17332