Next-Token Distribution

Follow this topic Ask about this topic

Momentum

4 papers in the last four weeks, with none the four weeks before. 0.0% of all new papers.

Jul 6Week of Sep 21

Latest papers 19

All topics

Interpolated Policy Distillation: A Controllable Continuum Between Off-Policy and On-Policy Distillation
Sep 29, 2026Youxu Shi, Yifan Sun, Dacheng Yin +5Off-Policy Learning Next-Token Distribution
University of Science and Technology of China · WeChat Vision, Tencent Inc.
CertMark: Distortion-Free Multi-Bit Watermarking with Certified Decoding
Sep 27, 2026Paweł Batorski, Przemysław Spurek, Paul SwobodaWatermarks Decoding
Heinrich Heine University Düsseldorf · Jagiellonian University · IDEAS Research Institute
Your Transformer Can Hold Two Thoughts at Once: Evidence of Linear Superposition in LLMs
Sep 24, 2026Pavel Tikhonov, Anton Korznikov, Matvey Mikhalchuk +6Transformer Architectures Next-Token Distribution
Next-token functional estimation
Sep 17, 2026Milind Nakul, Vidya Muthukumar, Ashwin PananjadyNext-Token Distribution Surprisal
Schools of Industrial and Systems Engineering and Electrical and Computer Engineering Georgia Institute of Technology
Recommendation Retrievers Need Verifiers: Universal Generative Reranking for Sequential Recommendations
Sep 14, 2026Benyu Zhang, Qiang Zhang, Rui Li +3Sequential Recommendation Cross-Encoder Reranking
Meta MRS
Attention-Steered Vision-Language Models for Sign Language Translation
Jul 31, 2026Meibo Hu, Guohao Sun, Annemarie D. Ross +2Gloss-Free Sign Language Translation Sign Language Translation
Rochester Institute of Technology · National Technical Institute for the Deaf / Rochester Institute of Technology · University of Virginia
Geometric Configurations of Perturbed Jailbreak Prompts
Jul 22, 2026Lynn Delcon, Andres Algaba, Vincent GinisLarge Language Model Jailbreaks Jailbreak Attacks
Department of Business Technology and Operations, Data Analytics Laboratory, Vrije Universiteit Brussel, Pleinlaan 2, 1050 Brussels, Belgium · imec-SMIT, Vrije Universiteit Brussel, Pleinlaan 9, 1050 Brussels, Belgium · School of Engineering and Applied Sciences, Harvard University, Cambridge, Massachusetts 02138, USA
Tracing LLM Behavior to the Training Data with Empirical Next-Token Distributions
Jul 15, 2026Zachary IzzoNext-Token Distribution Mechanistic Interpretability
NEC Labs America
K-Forcing: Joint Next-K-Token Decoding via Push-Forward Language Modeling
Jun 9, 2026Zhiwei Tang, Yuanyu He, Yizheng Han +4Autoregressive Language Models Autoregressive Generation
1DAMO Academy, Alibaba Group · 2Hupan Lab · 3Zhejiang University +1
Rethinking the Role of Positional Encoding: Sliding-Window Transformers without PE Remain Turing Complete
Jun 1, 2026Qian Li, Xinyu Mao, Shang-Hua TengPositional Encoding Transformer Architectures
Shenzhen Research Institute of Big Data · University of Southern California
Bastion: Budget-Aware Speculative Decoding with Tree-structured Block Diffusion Drafting
May 28, 2026Soowon Oh, Nam Cao, Yujin Kim +4Block-Diffusion Drafter Speculative Decoding
KAIST AI · Samsung Advanced Institute of Technology
The Bridge-Garden Dilemma in LLM Distillation: Why Mixing Hard and Soft Labels Works
May 25, 2026Guanghui Wang, Kaiwen Lv Kacuila, Zhiyong Yang +5Knowledge Distillation Dataset Distillation
School of Computer Science and Technology, University of Chinese Academy of Sciences, Beijing, China · Alibaba Group, Hangzhou, China · State Key Laboratory of AI Safety, Institute of Computing Technology, Chinese Academy of Sciences, Beijing, China +2
N-vium: Mixture-of-Exits Transformer for Accelerated Exact Generation
May 13, 2026Aleksander Lorenc, Frédéric Berdoz, Joël Mathys +1Autoregressive Transformers Transformer Architectures
ETH Zurich
Future Validity is the Missing Statistic: From Impossibility to $Φ$ -Estimation for Grammar-Faithful Speculative Decoding
May 8, 2026Wenhua Nie, Zijie Meng, Kun Zou +5Speculative Decoding Context-Free Grammars
National Taiwan University
The Convergence Gap: Instruction-Tuned Language Models Stabilize Later in the Forward Pass
May 8, 2026Yifan ZhouInstruction-Tuned Models Training-Inference Mismatch
University of California, Los Angeles
OLLM: Options-based Large Language Models
Apr 21, 2026Shashank Sharma, Janina Hoffmann, Vinay NamboodiriLLM Reasoning Strategies Large Language Models(Llms
Department of Computer Science University of Bath · Department of Psychology University of Bath
Self-Improving Tabular Language Models via Iterative Reward-Guided Post-Training
Apr 21, 2026Yunbo Long, Tejumade Afonja, Guangya Hao +2Post-Training Next-Token Distribution
Department of Engineering, University of Cambridge · CISPA Helmholtz Center for Information Security, Saarbrücken, Germany · The Alan Turing Institute, London
Deriving Neural Scaling Laws from the statistics of natural language
Feb 7, 2026Francesco Cagnetta, Allan Raventós, Surya Ganguli +1Scaling Laws Next-Token Distribution
Theoretical and Scientific Data Science, International School for Advanced Studies (SISSA), Trieste, Italy · Department of Applied Physics, Stanford University, Stanford, California · Department of Physics and Astronomy, Johns Hopkins University, Baltimore, Maryland +1
Extracting Probabilistic Knowledge from Large Language Models for Bayesian Network Parameterization
May 21, 2025Aliakbar Nafar, Kristen Brent Venable, Zijun Cui +1Bayesian Networks Prior Knowledge Integration
Michigan State University · Florida Institute for Human and Machine Cognition (IHMC) University of West Florida

Sep 29, 2026, cs.LGJ/K move · Enter open · S save

Interpolated Policy Distillation: A Controllable Continuum Between Off-Policy and On-Policy Distillation

Youxu Shi, Yifan Sun, Dacheng Yin, Haomiao Tang+4

University of Science and Technology of China · WeChat Vision, Tencent Inc.

Off-Policy Learning Next-Token Distribution Token-Level Uncertainty

Open paper AskPDF arXiv

Off-policy and on-policy distillation have traditionally been formulated as separate paradigms, each favoring a different property of distillation trajectories. Teacher-generated (off-policy) traces are typically high-quality but lie far from the student's distribution, whereas student-generated (on-policy) rollouts are more learnable but often contain erroneous reasoning. We view these paradigms as the endpoints of a policy continuum and posit that a more effective rollout policy may lie in between. We introduce \textbf{Interpolated Policy Distillation (IPD)}, which defines the next-token distribution at every decoding step as an explicit linear interpolation between the student and teacher distributions. The interpolation operates at the distribution level, token by token, and its coefficient provides direct control over the balance between trajectory quality and student learnability. Naively sampling from this policy would require sequentially querying the teacher at every token and is thus expensive. To make IPD practical, we accelerate it with a new speculative-decoding rule while exactly preserving the interpolated next-token distribution.At the trajectory level, the resulting rollouts naturally interleave student- and teacher-generated segments. Unlike recent heuristic segment-interleaving methods, however, this interleaving is induced by an exactly realized token-level interpolated policy rather than by hand-designed switching rules. Across text-only and multimodal reasoning benchmarks, IPD consistently outperforms both endpoint policies (SFT and OPD), their conventional two-stage combination (SFT-then-OPD), and recent heuristic segment-interleaving methods, demonstrating that token-level policy interpolation better balances trajectory quality and student learnability.

Momentum

Related topics

Latest papers 19