cs.CL（2025-10-19）

📊 共 21 篇论文

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (16) 支柱二：RL算法与架构 (RL & Architecture) (4) 支柱一：机器人控制 (Robot Control) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (16 篇)

#题目一句话要点标签🔗⭐
1 Does Visual Grounding Enhance the Understanding of Embodied Knowledge in Large Language Models? 提出具身知识理解基准，评估视觉 grounding 是否提升大语言模型感知能力。 large language model multimodal visual grounding
2 Real-Time World Crafting: Generating Structured Game Behaviors from Natural Language with Large Language Models 提出一种基于LLM的实时游戏世界构建框架，通过自然语言生成结构化游戏行为。 large language model chain-of-thought
3 Knowing the Facts but Choosing the Shortcut: Understanding How Large Language Models Compare Entities 揭示大语言模型实体比较中的启发式偏见，并探究模型规模的影响 large language model chain-of-thought
4 ChiKhaPo: A Large-Scale Multilingual Benchmark for Evaluating Lexical Comprehension and Generation in Large Language Models ChiKhaPo：大规模多语言基准，评估大语言模型的词汇理解与生成能力 large language model
5 Beacon: Single-Turn Diagnosis and Mitigation of Latent Sycophancy in Large Language Models Beacon：单轮诊断与缓解大语言模型中潜在的谄媚现象 large language model
6 Structured yet Bounded Temporal Understanding in Large Language Models 研究大型语言模型在不同时间参照框架下的时间理解能力 large language model
7 Natural Language Processing for Cardiology: A Narrative Review 综述性研究：利用自然语言处理技术解决心血管疾病领域的挑战 large language model multimodal
8 Mapping from Meaning: Addressing the Miscalibration of Prompt-Sensitive Language Models 提出基于语义概念空间采样的Prompt校准方法，提升语言模型不确定性估计。 large language model
9 Vocab Diet: Reshaping the Vocabulary of LLMs with Vector Arithmetic 提出Vocab Diet，利用向量运算重塑LLM词汇表，提升词汇多样性。 large language model
10 Bits Leaked per Query: Information-Theoretic Bounds on Adversarial Attacks against LLMs 提出信息论框架，量化LLM对抗攻击中的信息泄露，指导透明度与安全性的平衡。 large language model
11 Neuronal Group Communication for Efficient Neural representation 提出神经元组通信(NGC)框架，提升神经网络效率、模块化和可解释性。 large language model
12 Verifiable Fine-Tuning for LLMs: Zero-Knowledge Training Proofs Bound to Data Provenance and Policy 提出可验证微调方法，为大语言模型提供数据溯源和策略约束的零知识训练证明。 large language model
13 U-Codec: Ultra Low Frame-rate Neural Speech Codec for Fast High-fidelity Speech Generation 提出U-Codec，一种超低帧率神经语音编解码器，用于快速高保真语音生成。 large language model
14 so much depends / upon / a whitespace: Why Whitespace Matters for Poets and LLMs 关注诗歌空白符：研究空白符对诗歌创作和LLM生成诗歌的影响 large language model
15 Investigating the Impact of Rationales for LLMs on Natural Language Understanding 研究推理链对LLM在自然语言理解任务中的影响，并提出相应优化方法 chain-of-thought
16 All You Need is One: Capsule Prompt Tuning with a Single Vector 提出CaPT：利用单向量胶囊提示调整，提升大语言模型在下游任务的性能。 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (4 篇)

#题目一句话要点标签🔗⭐
17 SafeSearch: Do Not Trade Safety for Utility in LLM Search Agents SafeSearch：提出兼顾安全与效用的LLM搜索代理多目标强化学习方法 reinforcement learning large language model
18 Online Learning Defense against Iterative Jailbreak Attacks via Prompt Optimization 提出基于在线学习的防御框架，对抗迭代式越狱攻击 reinforcement learning large language model
19 Prompt-MII: Meta-Learning Instruction Induction for LLMs Prompt-MII：通过元学习指令归纳，提升大语言模型在上下文学习中的效率。 reinforcement learning large language model
20 MOSAIC: Masked Objective with Selective Adaptation for In-domain Contrastive Learning 提出MOSAIC框架，通过选择性自适应的掩码目标，实现文本嵌入模型在特定领域的对比学习。 contrastive learning

🔬 支柱一：机器人控制 (Robot Control) (1 篇)

#题目一句话要点标签🔗⭐
21 Atomic Literary Styling: Mechanistic Manipulation of Prose Generation in Neural Language Models 原子文学风格化：神经语言模型中散文生成的机制性操控 manipulation

⬅️ 返回 cs.CL 首页 · 🏠 返回主页