Skip to main content
arXiv is now an independent nonprofit! Learn more
archive
Search Submit Donate Log in
Press Enter to search · Advanced search

Computer Science > Machine Learning

arXiv:2603.07079 (cs)
[Submitted on 7 Mar 2026 (v1), last revised 12 Jun 2026 (this version, v3)]

Title:Entropy-Aware On-Policy Distillation of Language Models

Authors:Woogyeol Jin, Taywon Min, Yongjin Yang, Dennis Wei, Yi Zhou, Swanand Ravindra Kadhe, Nathalie Baracaldo, Kimin Lee
View a PDF of the paper titled Entropy-Aware On-Policy Distillation of Language Models, by Woogyeol Jin and 7 other authors
View PDF HTML (experimental)
Abstract:On-policy distillation is a promising approach for transferring knowledge between language models, where a student learns from dense token-level signals along its own trajectories. This framework typically uses reverse KL divergence, encouraging the student to match the teacher's high-confidence predictions. However, we show that the mode-seeking property of reverse KL reduces generation diversity and yields unstable learning signals when the teacher distribution has high entropy. To address this, we introduce Entropy-Aware On-Policy Distillation. Our key idea is augmenting the standard reverse KL objective with forward KL when teacher entropy is high, capturing the full range of plausible outputs while retaining precise imitation elsewhere. It balances mode-seeking precision with mode-covering robustness without sacrificing on-policy training efficiency. Experiments show that our method maintains generation diversity (sustained token-level entropy) and improves student-teacher alignment (lower forward KL on high-entropy tokens). Across six math reasoning benchmarks, this yields Pass@8 accuracy gains of +1.37 for Qwen3-0.6B-Base, +2.39 for Qwen3-1.7B-Base, and +5.05 for Qwen3-4B-Base compared to baseline on-policy distillation methods. These results demonstrate that accounting for teacher uncertainty is essential for maintaining diversity and achieving effective knowledge transfer.
Comments: 18 pages, 11 figures, ICML 2026
Subjects: Machine Learning (cs.LG); Computation and Language (cs.CL)
Cite as: arXiv:2603.07079 [cs.LG]
  (or arXiv:2603.07079v3 [cs.LG] for this version)
  https://doi.org/10.48550/arXiv.2603.07079
arXiv-issued DOI via DataCite

Submission history

From: Woogyeol Jin [view email]
[v1] Sat, 7 Mar 2026 07:26:18 UTC (744 KB)
[v2] Fri, 22 May 2026 17:34:18 UTC (201 KB)
[v3] Fri, 12 Jun 2026 19:40:20 UTC (201 KB)
Full-text links:

Access Paper:

    View a PDF of the paper titled Entropy-Aware On-Policy Distillation of Language Models, by Woogyeol Jin and 7 other authors
  • View PDF
  • HTML (experimental)
  • TeX Source
view license

Current browse context:

cs.LG
< prev   |   next >
new | recent | 2026-03
Change to browse by:
cs
cs.CL

References & Citations

  • NASA ADS
  • Google Scholar
  • Semantic Scholar
Loading...

BibTeX formatted citation

Data provided by:

Bookmark

BibSonomy Reddit

Bibliographic and Citation Tools

Bibliographic Explorer (What is the Explorer?)
Connected Papers (What is Connected Papers?)
Litmaps (What is Litmaps?)
scite Smart Citations (What are Smart Citations?)

Code, Data and Media Associated with this Article

alphaXiv (What is alphaXiv?)
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub (What is DagsHub?)
Gotit.pub (What is GotitPub?)
Hugging Face (What is Huggingface?)
ScienceCast (What is ScienceCast?)

Demos

Replicate (What is Replicate?)
Hugging Face Spaces (What is Spaces?)
TXYZ.AI (What is TXYZ.AI?)

Recommenders and Search Tools

Influence Flower (What are Influence Flowers?)
CORE Recommender (What is CORE?)
IArxiv Recommender (What is IArxiv?)
  • Author
  • Venue
  • Institution
  • Topic

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?)
We gratefully acknowledge support from our major funders, member institutions, , and all contributors.
About · Help · Contact · Subscribe · Copyright · Privacy · Accessibility · Operational Status (opens in new tab)
Major funding support from
Simons Foundation Simons Foundation International Schmidt Sciences