Copertina

Attention Illuminates LLM Reasoning: The Preplan-and-Anchor Rhythm Enables Fine-Grained Policy Optimization

Fuente: arXiv

Salvato in:

Dettagli Bibliografici
Autori principali:	Li, Yang, Dong, Zhichen, Sun, Yuhan, Wang, Weixun, Xiong, Shaopan, Luo, Yijia, Liu, Jiashun, Lu, Han, Wang, Jiamang, Su, Wenbo, Zheng, Bo, Yan, Junchi
Natura:	Preprint
Pubblicazione:	2025
Soggetti:	Computation and Language Machine Learning
Accesso online:	Acceder al recurso
Tags:	Aggiungi Tag Nessun Tag, puoi essere il primo ad aggiungerne!!

Documenti analoghi

Complementary Reinforcement Learning
di: Muhtar, Dilxat, et al.
Pubblicazione: (2026)

Part I: Tricks or Traps? A Deep Dive into RL for LLM Reasoning
di: Liu, Zihe, et al.
Pubblicazione: (2025)

Part II: ROLL Flash -- Accelerating RLVR and Agentic Training with Asynchrony
di: Lu, Han, et al.
Pubblicazione: (2025)

RollPacker: Mitigating Long-Tail Rollouts for Fast, Synchronous RL Post-Training
di: Gao, Wei, et al.
Pubblicazione: (2025)

Asymmetric Proximal Policy Optimization: mini-critics boost LLM reasoning
di: Liu, Jiashun, et al.
Pubblicazione: (2025)

Deconstructing Long Chain-of-Thought: A Structured Reasoning Optimization Framework for Long CoT Distillation
di: Luo, Yijia, et al.
Pubblicazione: (2025)

One Sample to Rule Them All: Extreme Data Efficiency in Multidiscipline Reasoning with Reinforcement Learning
di: Li, Yiyuan, et al.
Pubblicazione: (2026)

Reasoning Palette: Modulating Reasoning via Latent Contextualization for Controllable Exploration for (V)LMs
di: Long, Rujiao, et al.
Pubblicazione: (2025)

Adaptive Segment-level Reward: Bridging the Gap Between Action and Reward Space in Alignment
di: Li, Yanshi, et al.
Pubblicazione: (2024)

LiveThinking: Enabling Real-Time Efficient Reasoning for AI-Powered Livestreaming via Reinforcement Learning
di: Sun, Yuhan, et al.
Pubblicazione: (2025)

RollArt: Scaling Agentic RL Training via Disaggregated Infrastructure
di: Gao, Wei, et al.
Pubblicazione: (2025)

CE-RM: A Pointwise Generative Reward Model Optimized via Two-Stage Rollout and Unified Criteria
di: Hu, Xinyu, et al.
Pubblicazione: (2026)

ShopSimulator: Evaluating and Exploring RL-Driven LLM Agent for Shopping Assistants
di: Wang, Pei, et al.
Pubblicazione: (2026)

ROSE: Rollout On Serving GPUs via Cooperative Elasticity for Agentic RL
di: Gao, Wei, et al.
Pubblicazione: (2026)

Practical Preplanning Campaign Steps
di: Kim Pawlak
Pubblicazione: (2024)

ProgCo: Program Helps Self-Correction of Large Language Models
di: Song, Xiaoshuai, et al.
Pubblicazione: (2025)

Echoes as Anchors: Probabilistic Costs and Attention Refocusing in LLM Reasoning
di: Hao, Zhuoyuan, et al.
Pubblicazione: (2026)

LLM-Seg: Bridging Image Segmentation and Large Language Model Reasoning
di: Wang, Junchi, et al.
Pubblicazione: (2024)

DESIGNER: Design-Logic-Guided Multidisciplinary Data Synthesis for LLM Reasoning
di: Liu, Weize, et al.
Pubblicazione: (2025)

Attention Projection Mixing with Exogenous Anchors
di: Su, Jonathan
Pubblicazione: (2026)

Think-J: Learning to Think for Generative LLM-as-a-Judge
di: Huang, Hui, et al.
Pubblicazione: (2025)

AnchorDrive: LLM Scenario Rollout with Anchor-Guided Diffusion Regeneration for Safety-Critical Scenario Generation
di: Jiang, Zhulin, et al.
Pubblicazione: (2026)

3DAlign-DAER: Dynamic Attention Policy and Efficient Retrieval Strategy for Fine-grained 3D-Text Alignment at Scale
di: Fan, Yijia, et al.
Pubblicazione: (2025)

FineRMoE: Dimension Expansion for Finer-Grained Expert with Its Upcycling Approach
di: Liao, Ning, et al.
Pubblicazione: (2026)

Learning Grouped Lattice Vector Quantizers for Low-Bit LLM Compression
di: Zhang, Xi, et al.
Pubblicazione: (2025)

Attention with Dependency Parsing Augmentation for Fine-Grained Attribution
di: Ding, Qiang, et al.
Pubblicazione: (2024)

Learning Intractable Multimodal Policies with Reparameterization and Diversity Regularization
di: Wang, Ziqi, et al.
Pubblicazione: (2025)

Can Large Language Models Detect Errors in Long Chain-of-Thought Reasoning?
di: He, Yancheng, et al.
Pubblicazione: (2025)

DICE: Enabling Efficient General-Purpose SIMT Execution with Statically Scheduled Coarse-Grained Reconfigurable Arrays
di: Wang, Jiayi, et al.
Pubblicazione: (2026)

Fine-Grained Self-Endorsement Improves Factuality and Reasoning
di: Wang, Ante, et al.
Pubblicazione: (2024)

AnchorFormer: Differentiable Anchor Attention for Efficient Vision Transformer
di: Shan, Jiquan, et al.
Pubblicazione: (2025)

SimO Loss: Anchor-Free Contrastive Loss for Fine-Grained Supervised Contrastive Learning
di: Bouhsine, Taha, et al.
Pubblicazione: (2024)

Trading-R1: Financial Trading with LLM Reasoning via Reinforcement Learning
di: Xiao, Yijia, et al.
Pubblicazione: (2025)

Beyond Safe Answers: A Benchmark for Evaluating True Risk Awareness in Large Reasoning Models
di: Zheng, Baihui, et al.
Pubblicazione: (2025)

Beyond Scores: Diagnostic LLM Evaluation via Fine-Grained Abilities
di: Zhang, Xu, et al.
Pubblicazione: (2026)

LogicVista: Multimodal LLM Logical Reasoning Benchmark in Visual Contexts
di: Xiao, Yijia, et al.
Pubblicazione: (2024)

Fine-Grained VLM Fine-tuning via Latent Hierarchical Adapter Learning
di: Zhao, Yumiao, et al.
Pubblicazione: (2025)

Tactile-Guided Robotic Ultrasound: Mapping Preplanned Scan Paths for Intercostal Imaging
di: Zhang, Yifan, et al.
Pubblicazione: (2025)

STA-FEM: Exact Streaming Assembly for Preplanned Dynamic Tetrahedral Topology Edits
di: Acharya, Manish, et al.
Pubblicazione: (2026)

Theta Rhythm‐Based Attention Switch Training Effectively Modified Negative Attentional Bias
di: Guo Li, et al.
Pubblicazione: (2024)