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

Researchers introduce TTSR, a new framework that enables AI models to improve their reasoning abilities during test time by having a single model alternate between student and teacher roles. The system allows models to learn from their mistakes by analyzing failed reasoning attempts and generating targeted practice questions for continuous improvement.

• →TTSR uses a single pretrained language model that switches between student and teacher roles to improve reasoning at test time.
• →The framework addresses unreliable self-generated labels and inefficient learning by focusing on specific reasoning weaknesses.
• →Experimental results show consistent improvements in mathematical reasoning across different model architectures.
• →The teacher component analyzes failed reasoning attempts and creates targeted variant questions for learning.
• →This approach enables continual self-improvement without requiring additional training data or model fine-tuning.