STRETCH the Boundaries: A Unified Self-Taught Framework for Progressive LLM Evolution

Researchers proposed STRETCH (Self-Taught Reasoning Evolution via Targeted CHallenge), a unified framework that uses a dynamic Stretch Zone mechanism to align question difficulty with an LLM's evolving solving capability, according to arXiv paper 2609.18642v1. Within a single parameter space, the model alternates between a Scaffolder that generates adaptive, boundary-pushing challenges and a Learner that optimizes solving trajectories through reinforcement learning. Experiments on negotiation and operation research benchmarks showed STRETCH consistently outperformed strong prompting and domain-specific baselines, with scaffolder configuration analysis indicating dynamic difficulty alignment is critical for sustained capability improvement.

arXiv:2609.18642v1 Announce Type: new Abstract: Large language models LLMs often suffer from capability stagnation in self-improvement training because fixed difficulty levels fail to adapt to their evolving proficiency. To address this issue, we propose STRETCH Self-Taught Reasoning Evolution via Targeted CHallenge , a unified framework inspired by cognitive scaffolding theory. STRETCH introduces a dynamic Stretch Zone mechanism that continuously aligns question difficulty with the model's solving capability. Within a single parameter space, the model alternates between a Scaffolder that generates adaptive, boundary-pushing challenges and a Learner that that optimizes its solving trajectories through reinforcement learning. This dual-loop co-evolution effectively stabilizes training, mitigates reward hacking and promote progressive reasoning growth. Experiments on both negotiation and operation research benchmarks demonstrate that STRETCH consistently outperforms strong prompting and domain-specific baselines. Further scaffolder configuration analysis shows that dynamic difficulty alignment is critical for sustained capability improvement and synchronized reasoning evolution.