mlsanityData to 7 Oct 2026

APR: Penalizing Structural Redundancy in Large Reasoning Models via Anchor-based Process Rewards

arXiv 2602.00760 ↗First posted 31 Jan 2026cs.CLcited by 0

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

Abstract (arXiv)

Test-Time Scaling (TTS) has significantly enhanced the capabilities of Large Reasoning Models (LRMs) but introduces a critical side-effect known as Overthinking. We conduct a preliminary study to rethink this phenomenon from a fine-grained perspective. We observe that LRMs frequently conduct repetitive self-verification without revision even after obtaining the final answer during the reasoning process. We formally define this specific position where the answer first stabilizes as the Reasoning Anchor. By analyzing pre- and post-anchor reasoning behaviors, we uncover the structural redundancy fixed in LRMs: the meaningless repetitive verification after deriving the first complete answer, which we term the Answer-Stable Tail (AST). Motivated by this observation, we propose Anchor-based Process Reward (APR), a structure-aware reward shaping method that localizes the reasoning anchor and penalizes exclusively the post-anchor AST. Leveraging the policy optimization algorithm suitable for length penalties, our APR models achieved the performance-efficiency Pareto frontier at 1.5B and 7B scales averaged across five mathematical reasoning datasets while requiring substantially fewer computational resources for RL training.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in Test-time compute and efficient thinking, one of 17 directions in LLM reasoning. Builds on 34 of its 41 references that are in the map. First posted on arXiv 31 Jan 2026.

Builds on

34 of 41 references.

Closest ancestors · 19

Same direction, newest first

  1. 2025Leash: Adaptive Length Penalty and Reward Shaping for Efficient Large Reasoning Model25 Dec
  2. 2025DLER: Doing Length pEnalty Right - Incentivizing More Intelligence per Token via Reinforcement Learning16 Oct
  3. 2025Train Long, Think Short: Curriculum Learning for Efficient Reasoning12 Aug
  4. 2025Step-level Verifier-guided Hybrid Test-Time Scaling for Large Language Models21 Jul
  5. 2025Optimizing Length Compression in Large Reasoning Models17 Jun
  6. 2025Just Enough Thinking: Efficient Reasoning with Adaptive Length Penalties Reinforcement Learning5 Jun
  7. 2025Self-Route: Automatic Mode Switching via Capability Estimation for Efficient Reasoning27 May
  8. 2025Thinking Fast and Right: Balancing Accuracy and Reasoning Length with Adaptive Rewards23 May
  9. 2025Learn to Reason Efficiently with Adaptive Length-based Reward Shaping21 May
  10. 2025ThinkSwitcher: When to Think Hard, When to Think Fast20 May
  11. 2025AdaptThink: Reasoning Models Can Learn When to Think19 May
  12. 2025Between Underthinking and Overthinking: An Empirical Study of Reasoning Length and correctness in LLMs30 Apr
7 more
  1. 2025ShorterBetter: Guiding Reasoning Models to Find Optimal Inference Length for Efficient Reasoning30 Apr
  2. 2025ThinkPrune: Pruning Long Chain-of-Thought of LLMs via Reinforcement Learning2 Apr
  3. 2025L1: Controlling How Long A Reasoning Model Thinks With Reinforcement Learning6 Mar
  4. 2025s1: Simple test-time scaling31 Jan
  5. 2025O1-Pruner: Length-Harmonizing Fine-Tuning for O1-Like Reasoning Pruning22 Jan
  6. 2024Do NOT Think That Much for 2+3=? On the Overthinking of o1-Like LLMs30 Dec
  7. 2024Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameters6 Aug

Other references · 15

  1. 2024DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Modelscited by 8,851
  2. 2025Qwen3 Technical Reportcited by 8,195
  3. 2023Efficient Memory Management for Large Language Model Serving with PagedAttentioncited by 5,767
  4. 2021Measuring Mathematical Problem Solving With the MATH Datasetcited by 5,203
  5. 2025DAPO: An Open-Source LLM Reinforcement Learning System at Scalecited by 2,678
  6. 2024HybridFlow: A Flexible and Efficient RLHF Frameworkcited by 1,980
  7. 2022Solving Quantitative Reasoning Problems with Language Modelscited by 1,359
  8. 2025Kimi k1.5: Scaling Reinforcement Learning with LLMscited by 1,193
  9. 2024OlympiadBench: A Challenging Benchmark for Promoting AGI with Olympiad-Level Bilingual Multimodal Scientific Problemscited by 1,099
  10. 2024Tulu 3: Pushing Frontiers in Open Language Model Post-Trainingcited by 988
  11. 2025Does Reinforcement Learning Really Incentivize Reasoning Capacity in LLMs Beyond the Base Model?cited by 948
  12. 2025Training Language Models to Reason Efficientlycited by 242
3 more
  1. 2024Efficient Prompting Methods for Large Language Models: A Surveycited by 46
  2. 2025ConciseRL: Conciseness-Guided Reinforcement Learning for Efficient Reasoning Modelscited by 14
  3. 2025Efficient Reasoning via Reward Modelcited by 2

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2025Scaling over Scaling: Exploring Test-Time Scaling Plateau in Large Reasoning Models
  2. 2025From Long to Short: LLMs Excel at Trimming Own Reasoning Chains
  3. 2026Efficient Test-Time Scaling via Temporal Reasoning Aggregation
  4. 2025When to Continue Thinking: Adaptive Thinking Mode Switching for Efficient Reasoning
  5. 2026ThinkRetrieve: Retrieval-Augmented Reasoning Traces for Test-Time Scaling
  6. 2026Planned Test-Time Scaling with Coordinated Reasoning Paths
  7. 2025Inverse Scaling in Test-Time Compute
  8. 2025ARISE: An Adaptive Resolution-Aware Metric for Test-Time Scaling Evaluation in Large Reasoning Models