Beam Search

Follow this topic Ask about this topic

Momentum

3 papers in the last four weeks, level with the four weeks before. 0.0% of all new papers.

Jul 13Week of Sep 28

Latest papers 26

All topics

GrammarRL: Effective Grammar-Constrained Decoding via Reinforcement Learning
Sep 30, 2026Gabriele Tuccio, Antonino Furnari, Aldo Gangemi +1Language Modeling Constrained Decoding
University of Catania · ISTC - National Research Council, Italy · University of Bologna
Provable Test-Time Scaling for Beam Search in LLM Reasoning
Sep 29, 2026Qijia He, Yu Huang, Yuan Cheng +2Beam Search LLM Reasoning Strategies
The Ohio State University · University of Pennsylvania · National University of Singapore
Beam Search as Test-Time Self-Distillation via Counterfactual Contexts
Sep 29, 2026Su Ee Tan, Xiaotong Ji, Rasul Tutunov +2Self-Distillation Framework Inference-Time
Huawei Noah’s Ark Lab · UCL Centre for AI
Predictive Zonotope Reduction: Precise Runtime Monitoring under Uncertainty
Sep 3, 2026Vladimir Krsmanovic, Florian Kohn, Bernd Finkbeiner +1Model Predictive Control Uncertainty
CISPA Helmholtz Center for Information Security · Technical University of Munich · beanTech
Non-Parametric Spatiotemporal Trajectory Prediction via State-Conditioned Transition Sampling
Aug 14, 2026Michael Fore, Akshay Jain, Justin Downes +2Spatiotemporal Prediction Energy-Based Transformer
Amazon Web Services
Adaptive Semantic Capacity Allocation for Parallel Generative Recommendation
Aug 10, 2026Chenxi Li, Yuchen Lu, Xu YangSequential Recommendation Real-World Content Recommendation Problem
University of the Chinese Academy of Sciences · Institute of Automation, Chinese Academy of Sciences
Thought-Level Beam Search for Reasoning
Aug 8, 2026Lijie Yang, Hongyin Luo, Jiawei Zhao +2Large Reasoning Models Beam Search
Princeton University · MIT CSAIL · Meta AI
Neuro-Evolved Heuristics for Variable Gapped Common Subsequence Identification
Aug 1, 2026Marko Djukanović, Christian Blum, Aleksandar Kartelj +2Heuristics Sequence Modeling
University of Nova Gorica, Nova Gorica, Slovenia · Artificial Intelligence Research Institute (IIIA-CSIC), Barcelona, Spain · Faculty of Mathematics, University of Belgrade, Belgrade, Serbia +2
Rank-Conditioned Sample Reuse for the Plackett--Luce Best-of- $K$ Objective
Jul 13, 2026Melveena Jolly, Midhun XavierRejection Sampling Best-Of-N
Independent Researcher
FlashTrie: A GPU-Accelerated Constrained Beam Search for Generative Retrieval
Jul 10, 2026Dakshitha Anandakumar, Anurag Mukkara, Wenxiang Hu +5Beam Search Constrained Decoding
Microsoft, Redmond, WA, USA · Nvidia, Santa Clara, CA, USA
SeqGPT: A Constrained Transformer Agent for the Inverse Designof Multi-Panel Composite Structures
Jul 3, 2026Driss Chraibi, Alejandro García Pis, Stéphane Grihon +1Generative Inverse Design Beam Search
Toulouse INP · Airbus Operations SAS, Toulouse, France · IRIT +1
Cost-Optimal Decision Diagrams for Stochastic Boolean Function Evaluation
Jun 23, 2026Xia Zong, Tuomo Lehtonen, Jussi RintanenHeuristics Beam Search
Department of Computer Science, Aalto University
STORM: Stepwise Token Optimization with Reward-Guided Beam Search
Jun 9, 2026Arthur Satouf, Giulio D'Erasmo, Yuxuan Zong +3Retrieval Layer Retrievers
MILA – Quebec AI Institute & ILLS, Canada · 5Sapienza, University of Rome, Italy · 4Sorbonne Université & ISIR & CNRS, France +2
Structural Grid Descriptors Predict Within-Task Solver Success on ARC-AGI
Jun 8, 2026Ayan PendharkarFunctional Code Solvers Solvers
When Does Memory Help Multi-Trajectory Inference for Tool-Use LLM Agents?
May 27, 2026Xinzhe Li, Yaguang TaoLanguage-Model Agents Beam Search
RMIT University
Why are language models less surprised than humans? Testing the Parse Multiplicity Mismatch Hypothesis
May 14, 2026William Timkey, Brian Dillon, Tal LinzenSurprisal Natural Language
Department of Linguistics, New York University, 10 Washington Pl, New York, NY 10003, USA · Department of Linguistics, University of Massachusetts Amherst, 650 N Pleasant St, Amherst, MA 01002, USA · Center for Data Science, New York University, 60 5th Ave, New York, NY 10011, USA
Efficient Generative Retrieval for E-commerce Search with Semantic Cluster IDs and Expert-Guided RL
May 14, 2026Jianbo Zhu, Xing Fang, Jing Wang +5Beam Search Factual Recall
Taobao & Tmall Group of Alibaba Hangzhou, China
Enabling AI-Native Mobility in 6G: A Real-World Dataset for Handover, Beam Management, and Timing Advance
May 12, 2026Mannam Veera Narayana, Rohit Singh, Deepa M. R +1Mobility Naturalistic Driving Data
Department of Electrical Engineering Indian Institute of Technology Madras Chennai, India 600036
V-ABS: Action-Observer Driven Beam Search for Dynamic Visual Reasoning
May 11, 2026Zhiwei Ning, Xuanang Gao, Jiaxi Cao +6Multimodal Large Language Models Beam Search
School of Automation and Intelligent Sensing, Shanghai Jiao Tong University · Institute of Image Processing and Pattern Recognition, Shanghai Jiao Tong University · SenseTime Research +1
Entropy-informed Decoding: Adaptive Information-Driven Branching
May 10, 2026Benjamin Patrick Evans, Sumitra Ganesh, Leo ArdonConstrained Decoding Token-Level Entropy
JP Morgan AI Research, London, UK · JP Morgan AI Research, New York, USA
Lost in Decoding? Reproducing and Stress-Testing the Look-Ahead Prior in Generative Retrieval
Apr 25, 2026Kidist Amde Mekonnen, Yongkang Li, Yubao Tang +2Beam Search Constrained Decoding
University of Amsterdam Amsterdam, The Netherlands
An Efficient Beam Search Algorithm for Active Perception in Mobile Robotics
Apr 25, 2026Kaixian Qu, Han Wang, Victor Klemm +2Viewpoint-Dependent Active Perception Geometric Planning
Robotic Systems Lab, ETH Zurich
On Solving the Multiple Variable Gapped Longest Common Subsequence Problem
Apr 19, 2026Marko Djukanović, Nikola Balaban, Christian Blum +3Beam Search
University of Nova Gorica, Nova Gorica, Slovenia · Faculty of Natural Sciences and Mathematics, University of Banja Luka, Banja Luka, Bosnia and Herzegovina · Artificial Intelligence Research Institute (IIIA-CSIC), Barcelona, Spain +3
Layer-wise MoE Routing Locality under Shared-Prefix Code Generation: Token-Identity Decomposition and Compile-Equivalent Fork Redundancy
Apr 19, 2026Shun-ichiro Hayashi, Daichi Mukunoki, Tetsuya Hoshino +1Mixture-Of-Experts Large Language Models Code Generation
Graduate School of Informatics, Nagoya University, Japan · Information Technology Center, Nagoya University, Japan
xGR: Efficient Generative Recommendation Serving at Scale
Dec 12, 2025Qingxiao Sun, Tongxuan Liu, Shen Zhang +13Sequential Recommendation Beam Search
Beihang University · JD Company · University of Science and Technology Beijing +1
MASPRM: Multi-Agent System Process Reward Model
Oct 28, 2025Milad Yazdani, Mahdi Mostajabdaveh, Zirui Zhou +1Process Reward Model Beam Search
Department of Electrical and Computer Engineering, University of British Columbia, Vancouver, BC V6T1Z4, Canada · Huawei Technologies Canada, Burnaby, BC V5C6S7, Canada

Sep 30, 2026, cs.AIJ/K move · Enter open · S save

GrammarRL: Effective Grammar-Constrained Decoding via Reinforcement Learning

Gabriele Tuccio, Antonino Furnari, Aldo Gangemi, Misael Mongiovì

University of Catania · ISTC - National Research Council, Italy · University of Bologna

Language Modeling Constrained Decoding Large Language Model Reinforcement Learning Grammaticality Beam Search Decoding

Open paper AskPDF arXiv

Grammar-constrained generation guarantees syntactic validity, but can substantially degrade semantic quality when the model's preferred outputs are poorly aligned with the imposed grammar. This trade-off is particularly severe when the prompt is underspecified or the model has limited instruction-following ability. Beam search can partially mitigate these failures by exploring multiple valid sequences, but its computational cost grows with beam width, while sequence-level probability is only an imperfect proxy for semantic quality. We introduce GrammarRL, a label-free reinforcement learning method that adapts language models to grammar constraints without requiring annotated data. GrammarRL optimizes the model using two complementary self-supervised rewards derived from its own likelihoods: a direct reward, measuring how likely the constrained output is given the input, and a reverse reward, measuring how well the input can be reconstructed from the generated output. We optimize these rewards with a Reinforce Leave-One-Out (RLOO) objective over groups of grammar-constrained rollouts, augmented with the top-1 beam-search hypothesis and regularized towards a frozen base model. We evaluate GrammarRL on sign language gloss translation, hierarchical text classification, and named entity recognition using Llama models ranging from 1B to 8B parameters. GrammarRL consistently outperforms constrained greedy decoding, with an average improvement of 9.8 points and gains of up to 22.8 BLEU. It matches or outperforms beam search on two of the three tasks while preserving greedy-decoding inference cost. Ablations further show that the two rewards are complementary: either reward alone can underperform the untrained baseline, whereas their combination consistently improves upon it.

Momentum

Related topics

Latest papers 26