mlsanityData to 14 Oct 2026

Dynamic Thinking-Token Selection for Efficient Reasoning in Large Reasoning Models

arXiv 2601.18383 ↗First posted 26 Jan 2026cs.AI cs.CL cs.LGcited by 1

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

Abstract (arXiv)

Large Reasoning Models (LRMs) excel at solving complex problems by explicitly generating a reasoning trace before deriving the final answer. However, these extended generations incur substantial memory footprint and computational overhead, bottlenecking LRMs' efficiency. This work uses attention maps to analyze the influence of reasoning traces and uncover an interesting phenomenon: only some decision-critical tokens in a reasoning trace steer the model toward the final answer, while the remaining tokens contribute negligibly. Building on this observation, we propose Dynamic Thinking-Token Selection (DynTS). This method identifies decision-critical tokens and retains only their associated Key-Value (KV) cache states during inference, evicting the remaining redundant entries to optimize efficiency.

Key facts. Cited by 1 paper 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 24 of its 41 references that are in the map. First posted on arXiv 26 Jan 2026.

Builds on

24 of 41 references.

Closest ancestors · 6

Same direction, newest first

  1. 2025Hold Onto That Thought: Assessing KV Cache Compression On Reasoning12 Dec
  2. 2025Which Heads Matter for Reasoning? RL-Guided KV Cache Compression9 Oct
  3. 2025Think Clearly: Improving Reasoning via Redundant Token Pruning17 Jun
  4. 2025Efficient Reasoning Models: A Survey15 Apr
  5. 2025A Survey on Test-Time Scaling in Large Language Models: What, How, Where, and How Well?31 Mar
  6. 2025Stop Overthinking: A Survey on Efficient Reasoning for Large Language Models20 Mar

Other references · 18

  1. 2025Qwen3 Technical Reportcited by 8,820
  2. 2025DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learningcited by 6,330
  3. 2021Measuring Mathematical Problem Solving With the MATH Datasetcited by 5,371
  4. 2023GPQA: A Graduate-Level Google-Proof Q&A Benchmarkcited by 2,879
  5. 2023GQA: Training Generalized Multi-Query Transformer Models from Multi-Head Checkpointscited by 1,346
  6. 2024PyramidKV: Dynamic KV Cache Compression based on Pyramidal Information Funnelingcited by 432
  7. 2025Towards Reasoning Era: A Survey of Long Chain-of-Thought for Reasoning Large Language Modelscited by 384
  8. 2025Towards Large Reasoning Models: A Survey of Reinforced Reasoning with Large Language Modelscited by 293
  9. 2024Ada-KV: Optimizing KV Cache Eviction by Adaptive Budget Allocation for Efficient LLM Inferencecited by 235
  10. 2025Thought Anchors: Which LLM Reasoning Steps Matter?cited by 129
  11. 2024Rethinking Interpretability in the Era of Large Language Modelscited by 111
  12. 2024Keep the Cost Down: A Review on Methods to Optimize LLM' s KV-Cache Consumptioncited by 97
6 more
  1. 2024A Simple and Effective L2 Norm-Based Strategy for KV Cache Compressioncited by 86
  2. 2024Dynamic-LLaVA: Efficient Multimodal Large Language Models via Dynamic Vision-language Context Sparsificationcited by 59
  3. 2025CAKE: Cascading and Adaptive KV Cache Eviction with Layer Preferencescited by 58
  4. 2024SepLLM: Accelerate Large Language Models by Compressing One Segment into One Separatorcited by 41
  5. 2025Topology of Reasoning: Understanding Large Reasoning Models through Reasoning Graph Propertiescited by 29
  6. 2024BUZZ: Beehive-structured Sparse KV Cache with Segmented Heavy Hitters for Efficient LLM Inferencecited by 9

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026Value-Aware Stochastic KV Cache Eviction for Reasoning Modelscited by 8

Similar, not cited

Close in wording, no citation either way

  1. 2025Thought calibration: Efficient and confident test-time scaling
  2. 2025Tracing the Traces: Latent Temporal Signals for Efficient and Accurate Reasoning
  3. 2025Think or Not? Exploring Thinking Efficiency in Large Reasoning Models via an Information-Theoretic Lens
  4. 2026ThinkRetrieve: Retrieval-Augmented Reasoning Traces for Test-Time Scaling
  5. 2025THINK-Bench: Evaluating Thinking Efficiency and Chain-of-Thought Quality of Large Reasoning Models
  6. 2025The Illusion of Thinking: Understanding the Strengths and Limitations of Reasoning Models via the Lens of Problem Complexity
  7. 2025ThinkSwitcher: When to Think Hard, When to Think Fast
  8. 2025The Evolution of Thought: Tracking LLM Overthinking via Reasoning Dynamics Analysis