mlsanityData to 14 Oct 2026

Sample Where You Struggle: Sharpening Base Model Reasoning via Entropy-Guided Power Sampling

arXiv 2606.09926 ↗First posted 7 Jun 2026cs.LG cs.AIcited by 1

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

Abstract (arXiv)

Sampling from the sequence-level power distribution pα elicits RL-level reasoning from base language models without any parameter updates, but the standard Metropolis--Hastings (MH), a Markov Chain Monte Carlo (MCMC) sampler, is both expensive and slow-mixing. We trace both to a structural mismatch: pα mainly departs from p at a sparse, spatially clustered set of high-entropy decision points, yet MH proposes resampling positions uniformly along the prefix -- wasting compute on near-degenerate conditionals while under-mixing precisely where modes diverge. We propose Entropy-Guided Power Sampling (EGPS), a training-free and verifier-free sampler that re-derives its proposal from token-level entropy already in the forward pass. EGPS skips deterministic blocks, localizes each MCMC move to a high-entropy neighborhood, and applies Multiple-Try Metropolis at decision points -- making sampling cost scale with entropy mass rather than sequence length. On Qwen2.5-Math-7B, EGPS reaches best or tied-best accuracy on all three benchmarks (MATH500 75.8%, HumanEval 62.2%, GPQA 42.4%) at up to a 12.6× wall-clock speedup over the MH baseline.

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 15 of its 19 references that are in the map. First posted on arXiv 7 Jun 2026.

Builds on

15 of 19 references.

Closest ancestors · 4

Same direction, newest first

  1. 2026Power-SMC: Low-Latency Sequence-Level Power Sampling for Training-Free LLM Reasoning10 Feb
  2. 2025Less is More: Improving LLM Reasoning with Minimal Test-Time Intervention15 Oct
  3. 2025Deep Think with Confidence21 Aug
  4. 2024Critical Tokens Matter: Token-Level Contrastive Estimation Enhances LLM's Reasoning Capability29 Nov

Other references · 11

  1. 2024DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Modelscited by 9,379
  2. 2021Evaluating Large Language Models Trained on Codecited by 8,320
  3. 2023GPQA: A Graduate-Level Google-Proof Q&A Benchmarkcited by 2,879
  4. 2024Qwen2 Technical Reportcited by 2,120
  5. 2025Does Reinforcement Learning Really Incentivize Reasoning Capacity in LLMs Beyond the Base Model?cited by 982
  6. 2023Stay on topic with Classifier-Free Guidancecited by 91
  7. 2025Reasoning with Sampling: Your Base Model is Smarter Than You Thinkcited by 89
  8. 2025Outcome-based Exploration for LLM Reasoningcited by 83
  9. 2026Scalable Power Sampling: Unlocking Efficient, Training-Free Reasoning for LLMs via Distribution Sharpeningcited by 31
  10. 2026The Model Knows, the Decoder Finds: Future Value Guided Particle Power Samplingcited by 6
  11. 2026Sampling for Quality: Training-Free Reward-Guided LLM Decoding via Sequential Monte Carlocited by 5

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026Adaptive Power Sampling for LLM Reasoningcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026Reasoning with Sampling: Cutting at Decision Points
  2. 2026Depth-Entropy Guided Sampling for Training-Free LLM Reasoning
  3. 2026More Correct Mass, Worse Answers: Why Power Sampling Can Fail and How to Fix It
  4. 2026Power Distribution Bridges Sampling, Self-Reward RL, and Self-Distillation
  5. 2026Sharpen Without Search: On-Policy Distillation of Sequence-Level Power Distribution
  6. 2026Reject, Resample, Repeat: Understanding Parallel Reasoning in Language Model Inference
  7. 2026Entropy-Gated Latent Recursion
  8. 2026Explore Broadly, Reason Sharply: Push Small Models toward the Frontier via Sampling