TTSR: Test-Time Self-Reflection for Continual Reasoning Improvement

TL;DR

TTSR introduces a self-reflective, test-time training framework where a language model iteratively identifies and addresses its reasoning weaknesses through self-generated questions, enhancing reasoning abilities during testing.

Contribution

The paper presents TTSR, a novel test-time self-evolving training method that enables models to self-reflect and improve reasoning by alternating roles of student and teacher within a continual loop.

Findings

Consistently improves mathematical reasoning performance.

Generalizes across different model architectures.

Effective in challenging reasoning benchmarks.

Abstract

Test-time Training enables model adaptation using only test questions and offers a promising paradigm for improving the reasoning ability of large language models (LLMs). However, it faces two major challenges: test questions are often highly difficult, making self-generated pseudo-labels unreliable, and existing methods lack effective mechanisms to adapt to a model's specific reasoning weaknesses, leading to inefficient learning. To address these issues, we propose \textbf{TTSR}, a self-reflective test-time self-evolving training framework. TTSR employs a single pretrained language model that alternates between the roles of a \textit{Student} and a \textit{Teacher} at test time. The Student focuses on solving problems and learning from synthesized variant questions, while the Teacher analyzes the Student's failed reasoning trajectories, summarizes recurring reasoning weaknesses, and synthesizes targeted variant questions accordingly. This process guides the model to improve within a learnable regime through a continual self-evolving loop. Experimental results on multiple challenging mathematical reasoning benchmarks show that TTSR consistently improves reasoning performance and generalizes well across different model backbones and general-domain reasoning tasks. These findings suggest that teacher-mediated self-reflection provides an effective pathway for stable and continual reasoning improvement at test time.