cs.CL（2026-10-16）

📊 共 38 篇论文 | 🔗 5 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (29 🔗5) 支柱二：RL算法与架构 (RL & Architecture) (8) 支柱一：机器人控制 (Robot Control) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (29 篇)

#题目一句话要点标签🔗⭐
1 Competence-Based Evaluation of Logical Reasoning in Large Language Models 提出基于能力的评估方法以提升大语言模型的逻辑推理能力 large language model chain-of-thought
2 The Reliance Hypothesis: Chain-of-Thought Faithfulness Rises as Language Resource Level Falls 提出依赖假说以解决低资源语言中的推理问题 large language model chain-of-thought
3 On the Position Bias of Large Language Models: A Comprehensive Study 研究位置偏差以提升大语言模型的评估可靠性 large language model chain-of-thought
4 What Shapes Brain Alignment of Large Language Models during Dialogue? Insights from Language Production and Comprehension 研究对话中大语言模型的脑对齐机制提供新见解 large language model
5 Verbalising Model's Computation for Chain-of-Thought Faithfulness Detection 提出VERITAS以解决链式思维的可信性检测问题 chain-of-thought
6 Behavioral Couplings: From Formation to Behavioral Emergence in Language Models 研究行为耦合以揭示语言模型的意外行为 large language model
7 Controlling the Diversity-Quality Trade-Off of LLMs by Merging Models Before and After Alignment 提出深度感知FrankenMerging以优化LLMs的多样性与质量平衡 large language model
8 Truth in Contrast: Calibrated Factual Evidence from Language Model Likelihoods 提出对比生成似然性以解决事实声明置信度估计问题 large language model
9 Prism Attention: Proposal-Refined Index Sharing Mechanism for Efficient LLMs Inference 提出Prism Attention以解决长上下文解码效率问题 large language model
10 Uncertainty-Aware Budget Allocation for Efficient LLM Tree Search 提出不确定性感知预算分配以提升LLM树搜索效率 large language model
11 Steering beyond Language: Improving Multilingual Training-Free Text Embeddings via Paired-Residual Subspace Removal 提出PRSR以改善多语言训练无关文本嵌入 language conditioned
12 DLite: History-free Speculative Decoding for Agentic AI 提出DLite框架以解决长历史上下文带来的推理效率问题 large language model
13 The Cognitive Cost of LLM Writing: A Reading-Time Comparison with Human Text 提出RoBERTa-RT模型以评估人类与LLM文本的阅读时间差异 large language model
14 Perceive, Solve, Reason: Progressive Capability Formation in Time Series-LLMs 提出时间序列-LLMs的渐进能力形成方法以提升理解与推理能力 large language model
15 "You" Is Not Toxic: How LLM Labelers Rewrite the Web's Conversation Laws 提出新方法以解决LLM标签者对网络对话的误判问题 large language model
16 On the Tip of the Tongue: First-Token Cueing as a Behavioral Probe of Recall Failures in LLMs 提出首个标记提示以解决大型语言模型的回忆失败问题 large language model
17 Who Pays for Efficient Inference? Confidence-Based Small-to-Large LLM Routing Across Tasks and Languages 提出基于置信度的小到大语言模型路由以降低推理成本 large language model ✅
18 MAS-Bai: Linguistically Informed Adaptive Multi-Agent Translation for Bowen-Script Bai 提出MAS-Bai框架以解决低资源语言翻译问题 large language model
19 What LLMs Do in the Shadows: Reading Context Reliance from the Unembedding Dark Space 提出DarkLens以高效检测大语言模型的上下文依赖性 large language model
20 Cross-Centaur: Modeling Individual Decision-Making Across Psychological Tasks with Language Models 提出Cross-Centaur以解决个体决策建模问题 large language model ✅
21 Before Retrieval Begins: Safety and Tool-Use Decisions in LLM Agents 探讨大型语言模型工具调用前的安全决策机制 large language model
22 ParetoChat: An LLM-Based Interface for Interactive Multi-Objective Optimization 提出ParetoChat以解决多目标优化中的用户交互问题 large language model
23 ECHO: Replacing the Compliance Proxy with Response Harmfulness in Optimization-Based Jailbreak Attack 提出ECHO以解决语言模型的优化攻击问题 large language model ✅
24 Self-Evolving Agents Can Repair Failures but Struggle to Generalize from Them 提出冷启动失败演化方法以解决自我演化智能体的泛化问题 large language model
25 KiranaBench: Can a Language Model Run a Neighbourhood Shop? 提出KiranaBench以评估多语言环境下的LLM商店运营能力 large language model
26 SAFE-Mem: Secure Agentic Memory with Fine-Grained Authority and Governance-Aware Retrieval 提出SAFE-Mem以解决大语言模型记忆安全问题 large language model ✅
27 Beyond Surface Cues: Robust LLM-Generated Korean Text Detection 提出KoLing以解决韩文LLM生成文本检测问题 large language model
28 NLData2Opt: Benchmarking Parameter Derivation and Prediction for Modeling Optimization Problems from Natural Language and Data 提出NLData2Opt以解决自然语言与数据建模优化问题 large language model ✅
29 Veil-Then-Reveal: Protecting PII Privacy via Reversible Pseudo-Anonymization in Zero-Trust User--LLM Interaction 提出Veil-Then-Reveal框架以保护PII隐私 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (8 篇)

#题目一句话要点标签🔗⭐
30 FederatedBytes: Tokenizer-Free, Byte-Level Small Language Models for Efficient Multimodal Federated Learning 提出FederatedBytes以解决多模态联邦学习中的计算与通信开销问题 linear attention multimodal
31 Language Models Learn to Improve Language Models 提出LM2LM以解决语言模型自我改进的反馈挑战 reinforcement learning world model world models
32 Prune the Leaky Neurons: Mitigating Private Data Leakage in Fine-Tuned LLMs via Joint-Embedding-Guided Pruning 提出JEPA-Prune以解决大语言模型中的隐私数据泄露问题 JEPA Joint-Embedding Predictive Architecture joint-embedding predictive architecture
33 Self-Evolving Coding Agents via Harness-Policy Optimization 提出自我进化编码代理以解决编码任务中的优化问题 reinforcement learning policy learning large language model
34 MultiAgent-OPD: End-to-End Training of Hierarchical Multi-Agent Systems with On-Policy Distillation 提出MultiAgent-OPD以解决层次化多智能体系统训练难题 reinforcement learning distillation
35 HydraLLM: Adaptive Width Slicing for Elastic Text and Vision-Language Models 提出HydraLLM以解决多模态模型在异构硬件上的适应性问题 distillation multimodal
36 Aligning Language Models Towards Counterfactual Invariance 提出一种新方法以解决语言模型的反事实不变性问题 reinforcement learning large language model
37 Reasoning Reranking with Semantic IDs as the Sole Item Representation 提出ReRankID以解决仅用语义ID进行排序的问题 reinforcement learning distillation

🔬 支柱一：机器人控制 (Robot Control) (1 篇)

#题目一句话要点标签🔗⭐
38 MARS: Memory-Aware Routing for Session-Level LLM Assistants 提出MARS框架以解决长历史LLM助手的记忆与路由问题 shared control

⬅️ 返回 cs.CL 首页 · 🏠 返回主页