Skip to content
AI News HubLIVE
Original source2 min read

Depth-Entropy Guided Sampling for Training-Free LLM Reasoning

Summary

This paper introduces Depth-Entropy Guided Sampling (DEGS), a training-free test-time method that exploits layer-wise entropy collapse depth as an intrinsic quality signal. Combined with sequence likelihood in an MCMC power-sampling framework, DEGS achieves state-of-the-art training-free accuracy across multiple reasoning benchmarks with single-digit-percent wall-clock overhead.

SourcearXiv Machine LearningAuthor: Zibin Meng, Peng Xie, Kani Chen
Depth-Entropy Guided Sampling for Training-Free LLM Reasoning
Report an error

The correction channel is not available yet. You can copy the article reference below for later.

Correction instructions
Read article

[Submitted on 19 Jun 2026]

Title:Depth-Entropy Guided Sampling for Training-Free LLM Reasoning

View a PDF of the paper titled Depth-Entropy Guided Sampling for Training-Free LLM Reasoning, by Zibin Meng and 2 other authors

View PDF HTML (experimental)

Abstract:Reinforcement learning (RL) has become the dominant paradigm for improving the reasoning capabilities of large language models, but it requires expensive training, curated data, and reward signals. Recent work shows that sampling from sharpened base-model distributions at test time recovers much of the RL gain, yet existing methods rely solely on output-layer likelihoods and ignore the transformer's internal forward-pass dynamics. We introduce Depth-Entropy Guided Sampling (DEGS), a training-free, test-time method that exploits layer-wise entropy collapse as an intrinsic quality signal. We observe that stronger reasoners -- including RL-posttrained variants -- exhibit a distinctive "late collapse": logit-lens decoded entropy stays elevated until deeper layers before converging. We define a per-sequence collapse depth $D(\mathbf{x})$ and a joint objective $\pi(\mathbf{x}) \propto p(\mathbf{x})^\alpha \exp(\beta D(\mathbf{x}))$ that combines sequence likelihood with this depth-entropy structure, instantiated inside an MCMC power-sampling framework (DEGS-MCMC). Across three open-weight models and four reasoning benchmarks, this near-chance per-candidate signal compounds over the sampling trajectory into state-of-the-art training-free accuracy, with gains largest out of domain and on the harder splits -- exactly where likelihood alone falls short -- at single-digit-percent wall-clock overhead. DEGS narrowly trails an in-house GRPO reference on the math splits GRPO was trained for, yet surpasses it out of domain on GPQA for all three models, without any training, reward model, or labeled data.

Subjects:

Machine Learning (cs.LG); Artificial Intelligence (cs.AI)

Cite as: arXiv:2607.09693 [cs.LG]

(or arXiv:2607.09693v1 [cs.LG] for this version)

https://doi.org/10.48550/arXiv.2607.09693

arXiv-issued DOI via DataCite

Submission history

From: Zibin Meng [view email] [v1] Fri, 19 Jun 2026 12:36:23 UTC (1,210 KB)

Full-text links:

Access Paper:

View a PDF of the paper titled Depth-Entropy Guided Sampling for Training-Free LLM Reasoning, by Zibin Meng and 2 other authors

View PDF

HTML (experimental)

TeX Source

view license

Current browse context:

cs.LG

new | recent | 2026-07

Change to browse by:

cs cs.AI

References & Citations

NASA ADS

Google Scholar

Semantic Scholar

Loading...

Data provided by:

Bibliographic Tools

Bibliographic and Citation Tools

Bibliographic Explorer Toggle

Bibliographic Explorer (What is the Explorer?)

Connected Papers Toggle

Connected Papers (What is Connected Papers?)

Litmaps Toggle

Litmaps (What is Litmaps?)

scite.ai Toggle

scite Smart Citations (What are Smart Citations?)

Code, Data, Media

Code, Data and Media Associated with this Article

alphaXiv Toggle

alphaXiv (What is alphaXiv?)

Links to Code Toggle

CatalyzeX Code Finder for Papers (What is CatalyzeX?)

DagsHub Toggle

DagsHub (What is DagsHub?)

GotitPub Toggle

Gotit.pub (What is GotitPub?)

Huggingface Toggle

Hugging Face (What is Huggingface?)

ScienceCast Toggle

ScienceCast (What is ScienceCast?)

Demos

Demos

Replicate Toggle

Replicate (What is Replicate?)

Spaces Toggle

Hugging Face Spaces (What is Spaces?)

Spaces Toggle

TXYZ.AI (What is TXYZ.AI?)

Related Papers

Recommenders and Search Tools

Link to Influence Flower

Influence Flower (What are Influence Flowers?)

Core recommender toggle

CORE Recommender (What is CORE?)

IArxiv recommender toggle

IArxiv Recommender (What is IArxiv?)

Author

Venue

Institution

Topic

About arXivLabs

arXivLabs: experimental projects with community collaborators

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)

Key points and analysis

Article intelligence

EngineersAdvanced

Key points

  • DEGS exploits a 'late collapse' of logit-lens decoded entropy across layers to distinguish stronger reasoning models, including RL-posttrained variants.
  • The method requires no training, reward model, or labeled data and surpasses a GRPO reference out of domain on GPQA for all three models.
  • Across three open-weight models and four reasoning benchmarks, DEGS achieves state-of-the-art training-free accuracy at single-digit-percent wall-clock overhead.

Highlights and analysis are generated automatically and may contain errors. Check the original source.