Revisiting Test-Time Scaling: A Survey and a Diversity-Aware Method for Efficient Reasoning
This work addresses the efficiency of reasoning in LLMs for AI applications, offering an incremental improvement by enhancing diversity in existing Test-Time Scaling methods.
The paper tackles the problem of limited output diversity in reasoning-optimized Large Language Models during Test-Time Scaling, which reduces efficiency, and proposes ADAPT, a lightweight diversity-aware prefix fine-tuning method that achieves 80% accuracy on mathematical reasoning tasks with eight times less compute than baselines.
Test-Time Scaling (TTS) improves the reasoning performance of Large Language Models (LLMs) by allocating additional compute during inference. We conduct a structured survey of TTS methods and categorize them into sampling-based, search-based, and trajectory optimization strategies. We observe that reasoning-optimized models often produce less diverse outputs, which limits TTS effectiveness. To address this, we propose ADAPT (A Diversity Aware Prefix fine-Tuning), a lightweight method that applies prefix tuning with a diversity-focused data strategy. Experiments on mathematical reasoning tasks show that ADAPT reaches 80% accuracy using eight times less compute than strong baselines. Our findings highlight the essential role of generative diversity in maximizing TTS effectiveness.