LATTS: Locally Adaptive Test-Time Scaling

2025/09/16 by Theo Uscidda, Matthew Trager, Uscidda, Theo +8 · 1 citation
Computer Science · #Artificial Intelligence (cs.AI) #Educational Technology and Assessment #FOS: Computer and information sciences

paper · pdf · doi:10.48550/arxiv.2509.20368

openalex publication_date 2025/09/16 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

One common strategy for improving the performance of Large Language Models (LLMs) on downstream tasks involves using a verifier model to either select the best answer from a pool of candidates or to steer the auto-regressive generation process towards better outputs. This class of methods typically results in improved accuracy at the cost of increased computation at test-time, a paradigm known as test-time scaling. However, most existing approaches increase computation uniformly across all samples and generation steps, without considering the complexity of individual instances, leading to inefficient resource use. We address this limitation by proposing an approach, called Locally Adaptive Test-Time Scaling (LATTS), that allocates variable compute across generation steps. Specifically, at each generation step, LATTS employs a verifier-based acceptance criterion to decide whether to resample, backtrack, restart, or stop the generation process. This criterion effectively adjusts the per-step computational effort based on a precise notion of local difficulty derived from the verifier model. Empirical results show that LATTS achieves significantly superior accuracy--compute tradeoffs compared to standard verifier-based methods.

Citations

• Can Large Language Models Detect Errors in Long Chain-of-Thought Reasoning?
• Bag of Tricks for Inference-time Computation of LLM Reasoning
• Can 1B LLM Surpass 405B LLM? Rethinking Compute-Optimal Test-Time Scaling
• Reward-Guided Speculative Decoding for Efficient LLM Reasoning
• s1: Simple test-time scaling
• DeepSeek-R1 incentivizes reasoning in LLMs through reinforcement learning
• The Lessons of Developing Process Reward Models in Mathematical Reasoning
• Free Process Rewards without Process Labels
• Learning How Hard to Think: Input-Adaptive Allocation of LM Computation
• Adaptive Inference-Time Compute: LLMs Can Predict if They Can Do Better, Even Mid-Generation
• MAgICoRe: Multi-Agent, Iterative, Coarse-to-Fine Refinement for Reasoning
• Make Every Penny Count: Difficulty-Adaptive Self-Consistency for Cost-Efficient Reasoning
• Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameters
• Large Language Monkeys: Scaling Inference Compute with Repeated Sampling
• AlphaMath Almost Zero: Process Supervision without Process
• Efficient Memory Management for Large Language Model Serving with PagedAttention
• Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena
• Direct Preference Optimization: Your Language Model is Secretly a Reward Model
• RAFT: Reward rAnked FineTuning for Generative Foundation Model Alignment
• Aligning Language Models with Preferences through f-divergence Minimization
• Accelerating Large Language Model Decoding with Speculative Sampling
• Fast Inference from Transformers via Speculative Decoding
• On Reinforcement Learning and Distribution Matching for Fine-Tuning Language Models with no Catastrophic Forgetting
• Least-to-Most Prompting Enables Complex Reasoning in Large Language Models
• BNAI, NO-TOKEN, and MIND-UNITY: Pillars of a Systemic Revolution in Artificial Intelligence
• Show Your Work: Scratchpads for Intermediate Computation with Language Models
• The Curious Case of Neural Text Degeneration

Cited by

• e1: Learning Adaptive Control of Reasoning Effort

Related

• Scaling Test-time Compute for LLM Agents
• Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameters
• SETS: Leveraging Self-Verification and Self-Correction for Improved Test-Time Scaling
• Test-time Scaling of LLMs: A Survey from A Subproblem Structure Perspective
• Trust but Verify! A Survey on Verification Design for Test-time Scaling