Scaling the Scaling Logic: Agentic Meta-Synthesis of Logic Reasoning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Liu, Bowen, Wu, Zhi, Xie, Runquan, Kang, Zhanhui, Li, Jia
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915917131153408
author Liu, Bowen
Wu, Zhi
Xie, Runquan
Kang, Zhanhui
Li, Jia
author_facet Liu, Bowen
Wu, Zhi
Xie, Runquan
Kang, Zhanhui
Li, Jia
contents Reinforcement Learning from Verifiable Rewards (RLVR) is bottlenecked by data: existing synthesis pipelines rely on expert-written code or fixed templates, confining growth to instance-level perturbations. We shift the evolvable unit from problem instances to task-family specifications. SSLogic is an agentic meta-synthesis framework in which LLM agents iteratively author and refine executable Generator-Validator pairs inside a closed Generate-Validate-Refine loop, producing families with new rules and difficulty gradients rather than parameter variations of old ones. A Multi-Gate Validation Protocol -- multi-strategy consensus plus Adversarial Blind Review, where independent agents solve each instance by writing and executing code -- filters ill-posed tasks before they enter training. Starting from 400 seed families, two evolution rounds yield 953 families and 21,389 verifiable instances. Three converging comparisons (step-matched, token-matched, and size-controlled on external Enigmata data) consistently show higher training utility of evolved data, with gains of SynLogic +5.2, AIME25 +3.0, and BBH +5.5 on Enigmata. Fine-grained KORBench evaluation reveals selective improvements in logic (+13.2%) and operation (+9.6%), linking structural evolution to downstream gains. Code: https://github.com/AdAstraAbyssoque/Scaling-the-Scaling-Logic
format Preprint
id arxiv_https___arxiv_org_abs_2602_13218
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Scaling the Scaling Logic: Agentic Meta-Synthesis of Logic Reasoning
Liu, Bowen
Wu, Zhi
Xie, Runquan
Kang, Zhanhui
Li, Jia
Artificial Intelligence
Computation and Language
Machine Learning
Logic in Computer Science
68T50, 68T05, 68N30
I.2.7; I.2.6; I.2.3; D.1.2
Reinforcement Learning from Verifiable Rewards (RLVR) is bottlenecked by data: existing synthesis pipelines rely on expert-written code or fixed templates, confining growth to instance-level perturbations. We shift the evolvable unit from problem instances to task-family specifications. SSLogic is an agentic meta-synthesis framework in which LLM agents iteratively author and refine executable Generator-Validator pairs inside a closed Generate-Validate-Refine loop, producing families with new rules and difficulty gradients rather than parameter variations of old ones. A Multi-Gate Validation Protocol -- multi-strategy consensus plus Adversarial Blind Review, where independent agents solve each instance by writing and executing code -- filters ill-posed tasks before they enter training. Starting from 400 seed families, two evolution rounds yield 953 families and 21,389 verifiable instances. Three converging comparisons (step-matched, token-matched, and size-controlled on external Enigmata data) consistently show higher training utility of evolved data, with gains of SynLogic +5.2, AIME25 +3.0, and BBH +5.5 on Enigmata. Fine-grained KORBench evaluation reveals selective improvements in logic (+13.2%) and operation (+9.6%), linking structural evolution to downstream gains. Code: https://github.com/AdAstraAbyssoque/Scaling-the-Scaling-Logic
title Scaling the Scaling Logic: Agentic Meta-Synthesis of Logic Reasoning
topic Artificial Intelligence
Computation and Language
Machine Learning
Logic in Computer Science
68T50, 68T05, 68N30
I.2.7; I.2.6; I.2.3; D.1.2
url https://arxiv.org/abs/2602.13218