mlsanityData to 14 Oct 2026

LATTS: Locally Adaptive Test-Time Scaling

arXiv 2509.20368 ↗First posted 16 Sep 2025cs.AIcited by 4

On the map: LLM reasoning → Test-time compute and efficient thinking

Abstract (arXiv)

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.

Key facts. Cited by 4 papers in the map (data through 14 Oct 2026). Placed in Test-time compute and efficient thinking, one of 17 directions in LLM reasoning. Builds on 33 of its 37 references that are in the map. First posted on arXiv 16 Sep 2025.

Builds on

33 of 37 references.

Closest ancestors · 8

Same direction, newest first

  1. 2025Bag of Tricks for Inference-time Computation of LLM Reasoning11 Feb
  2. 2025Can 1B LLM Surpass 405B LLM? Rethinking Compute-Optimal Test-Time Scaling10 Feb
  3. 2025s1: Simple test-time scaling31 Jan
  4. 2025Reward-Guided Speculative Decoding for Efficient LLM Reasoning31 Jan
  5. 2024Learning How Hard to Think: Input-Adaptive Allocation of LM Computation7 Oct
  6. 2024Adaptive Inference-Time Compute: LLMs Can Predict if They Can Do Better, Even Mid-Generation3 Oct
  7. 2024Make Every Penny Count: Difficulty-Adaptive Self-Consistency for Cost-Efficient Reasoning24 Aug
  8. 2024Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameters6 Aug

Other references · 25

  1. 2024The Llama 3 Herd of Modelscited by 14,715
  2. 2022Chain-of-Thought Prompting Elicits Reasoning in Large Language Modelscited by 12,459
  3. 2023Judging LLM-as-a-Judge with MT-Bench and Chatbot Arenacited by 8,626
  4. 2023Direct Preference Optimization: Your Language Model is Secretly a Reward Modelcited by 8,065
  5. 2025DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learningcited by 6,330
  6. 2023Efficient Memory Management for Large Language Model Serving with PagedAttentioncited by 5,944
  7. 2023Let's Verify Step by Stepcited by 3,789
  8. 2019The Curious Case of Neural Text Degenerationcited by 2,916
  9. 2022Fast Inference from Transformers via Speculative Decodingcited by 1,653
  10. 2022Least-to-Most Prompting Enables Complex Reasoning in Large Language Modelscited by 1,407
  11. 2025Kimi k1.5: Scaling Reinforcement Learning with LLMscited by 1,210
  12. 2023Accelerating Large Language Model Decoding with Speculative Samplingcited by 1,006
13 more
  1. 2024Large Language Monkeys: Scaling Inference Compute with Repeated Samplingcited by 991
  2. 2021Show Your Work: Scratchpads for Intermediate Computation with Language Modelscited by 886
  3. 2023RAFT: Reward rAnked FineTuning for Generative Foundation Model Alignmentcited by 686
  4. 2025The Lessons of Developing Process Reward Models in Mathematical Reasoningcited by 415
  5. 2024ProcessBench: Identifying Process Errors in Mathematical Reasoningcited by 292
  6. 2024Free Process Rewards without Process Labelscited by 165
  7. 2024AlphaMath Almost Zero: Process Supervision without Processcited by 162
  8. 2024Mutual Reasoning Makes Smaller LLMs Stronger Problem-Solverscited by 138
  9. 2023Aligning Language Models with Preferences through f-divergence Minimizationcited by 120
  10. 2024Critique-out-Loud Reward Modelscited by 106
  11. 2022On Reinforcement Learning and Distribution Matching for Fine-Tuning Language Models with no Catastrophic Forgettingcited by 82
  12. 2025Can Large Language Models Detect Errors in Long Chain-of-Thought Reasoning?cited by 44
  13. 2024MAgICoRe: Multi-Agent, Iterative, Coarse-to-Fine Refinement for Reasoningcited by 41

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 2; 2026: 2.

4 papers.

  1. 2025e1: Learning Adaptive Control of Reasoning Effortcited by 6
  2. 2025Re-FORC: Adaptive Reward Prediction for Efficient Chain-of-Thought Reasoningcited by 4
  3. 2026Think Again or Think Longer? Selective Verification for Budget-Aware Reasoningcited by 1
  4. 2026Can Language Models Learn to Reject Their Own Bad Reasoning Steps?cited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025Test-time Scaling of LLMs: A Survey from A Subproblem Structure Perspective
  2. 2025AgentTTS: Large Language Model Agent for Test-time Compute-optimal Scaling Strategy in Complex Tasks
  3. 2025A Survey on Test-Time Scaling in Large Language Models: What, How, Where, and How Well?
  4. 2025Scaling Test-time Compute for LLM Agents
  5. 2025The Art of Scaling Test-Time Compute for Large Language Models
  6. 2026Ladders in Chaos: When, How, (and Perhaps Why) Does Test-Time Scaling Improve LLM Machine Translation
  7. 2026Scaling Down the Scaling Laws: Parameter Efficiency and Compute-Optimal Training in Resource-Constrained Large Language Models
  8. 2026Sample Count Is Not Enough: Candidate-Generation Strategy Shapes the Energy and Performance of LLM Test-Time Scaling