cs.CL（2026-10-05）

📊 共 25 篇论文 | 🔗 4 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (15 🔗1) 支柱二：RL算法与架构 (RL & Architecture) (10 🔗3)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (15 篇)

#题目一句话要点标签🔗⭐
1 SAFE-MR: Evidence Sufficiency Learning for Selective Multimodal Rumor Detection 提出SAFE-MR框架以解决多模态谣言检测中的证据充分性问题 multimodal
2 Breaking the Tie: A Cluster-Aware Routing Framework for Large Language Models 提出Cluster-Aware Soft-Labeling Routing框架以解决LLM路由噪声问题 large language model
3 Spend Bytes on Breadth: Precision-Count Trade-offs for Decode-Time KV Compression in Long Chain-of-Thought Reasoning 提出BreadthKV以优化长链推理中的KV缓存压缩 chain-of-thought
4 PlotGround: Grounding Plot Digitization in Real Scientific Figures and Their Source Data 提出PlotGround以解决科学图表数字化问题 multimodal
5 Agentic schema-guided extraction of materials process knowledge from scientific literature 提出SciKGExtract框架以解决材料文献知识提取问题 large language model
6 Cross-Lingual Transferability of Training Data Extraction Attacks to Recover Memorized PII 研究跨语言训练数据提取攻击以恢复个人身份信息 large language model
7 LightMTP: Lightweight Latent Multi-Token Prediction 提出LightMTP以解决多标记预测中的参数效率问题 large language model
8 Differentiable Bit-Widths: Co-optimizing Pruning and Quantization via SVD for Ultra-Efficient LLM Compression 提出可微分位宽以统一优化剪枝与量化，提升LLM压缩效率 large language model ✅
9 JEV versus LLMs: Accuracy, Cost and Calibration on Seven Political Science Replications 比较JEV与LLMs在政治科学任务中的准确性与成本 large language model
10 Before Agent Tells The Lie: Has Deception Already Been Represented? 提出内部表示分析以预测代理的欺骗行为 large language model
11 Test-Time Adaptation of Reasoning Strategies with Bayesian Nonparametric Memory 提出贝叶斯非参数记忆以解决推理策略适应问题 large language model
12 Behavior-Preserving KV Cache Compression 提出行为保留的KV缓存压缩以解决长上下文推理瓶颈问题 large language model
13 DeferKV: Rethinking Eviction Timing for One-Shot KV Cache Compression 提出DeferKV以优化KV缓存压缩的驱逐时机 large language model
14 Probabilistic Race and Ethnicity Prediction Using Group-Specific Name Lists 提出$ ext{lBISG}$方法以解决种族和民族预测问题 large language model
15 TrustMI: Causally controlling how assistants trust their users 提出TrustMI以因果控制助手对用户的信任 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (10 篇)

#题目一句话要点标签🔗⭐
16 MedPrune: Topology-Efficient Multimodal Multi-Agent Communication Evolution for Medical VQA Tasks 提出MedPrune以解决医疗VQA任务中的通信效率问题 reinforcement learning large language model multimodal
17 Reward Stealing Attack on Large Language Models 提出奖励窃取攻击以解决大语言模型的安全性问题 reinforcement learning inverse reinforcement learning large language model ✅
18 MemPilot: Orchestrating On-Demand Multimodal Memory Curation for LLM Agents 提出MemPilot以解决多模态记忆管理的灵活性问题 reinforcement learning multimodal
19 Ontology Concept Overlap as a Training Signal: Knowledge-Grounded Reinforcement Learning for Clinical Question Answering 提出基于本体概念重叠的训练信号以解决临床问答问题 reinforcement learning PPO RLHF
20 The Assistance Dilemma: Learning to Teach via Multi-Turn Reinforcement Learning 提出多轮强化学习方法以提升大语言模型的教学能力 reinforcement learning reward design large language model
21 Long-Horizon Textual World Modeling through Structured Reasoning 通过结构化推理提出长时间文本世界建模方法 world model world models
22 From Traces to Agentic Worlds: Agentic Language World Models for Interactive Environment Simulation 提出Trace2Env以解决环境重建困难问题 world model world models
23 Improving Diversity in LLM Short Story Generation 提出DivLM框架以解决LLM短篇故事生成的多样性问题 reinforcement learning large language model instruction following
24 LoGRA: Scaling LLM Reinforcement Learning with Low-Rank Gradient Sketches 提出LoGRA以解决大规模语言模型强化学习的内存瓶颈问题 reinforcement learning large language model ✅
25 HLA: Expressive Hybrid Linear Attention via Chunk-Wise Dynamic Mixing 提出混合线性注意力机制以解决长上下文建模问题 linear attention ✅

⬅️ 返回 cs.CL 首页 · 🏠 返回主页