Revisiting Test-Time Scaling: A Survey and a Diversity-Aware Method for Efficient Reasoning

TL;DR

This paper surveys Test-Time Scaling methods for Large Language Models and introduces ADAPT, a diversity-aware prefix tuning technique that significantly enhances reasoning performance with less compute by promoting output diversity.

Contribution

It provides a structured survey of TTS methods and proposes ADAPT, a novel diversity-focused fine-tuning approach that improves reasoning accuracy efficiently.

Findings

ADAPT achieves 80% accuracy on reasoning tasks.

ADAPT uses eight times less compute than strong baselines.

Diversity is crucial for maximizing TTS effectiveness.

Abstract

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.