cs.CL（2025-12-02）

📊 共 27 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) (22 🔗5) 支柱二：RL算法与架构 (RL & Architecture) (5)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (22 篇)

#题目一句话要点标签🔗⭐
1 The Moral Consistency Pipeline: Continuous Ethical Evaluation for Large Language Models 提出道德一致性管道(MoCoP)，用于持续评估大型语言模型的伦理道德 large language model
2 Fine-Tuned Large Language Models for Logical Translation: Reducing Hallucinations with Lang2Logic 提出Lang2Logic框架，利用微调大语言模型减少逻辑翻译中的幻觉问题 large language model
3 BOOM: Beyond Only One Modality KIT's Multimodal Multilingual Lecture Companion 提出BOOM以解决多模态多语言讲座内容本地化问题 multimodal ✅
4 A benchmark dataset for evaluating Syndrome Differentiation and Treatment in large language models 构建中医领域大型语言模型评测基准TCM-BEST4SDT，用于评估辨证论治能力。 large language model
5 Towards Unification of Hallucination Detection and Fact Verification for Large Language Models 提出UniFact统一框架，弥合LLM幻觉检测与事实验证的研究鸿沟 large language model ✅
6 PEFT-Factory: Unified Parameter-Efficient Fine-Tuning of Autoregressive Large Language Models PEFT-Factory：统一自回归大语言模型的高效参数微调框架 large language model ✅
7 Spoken Conversational Agents with Large Language Models 语音对话Agent正向语音原生LLM演进，本教程提供系统级路线图。 large language model
8 TaleFrame: An Interactive Story Generation System with Fine-Grained Control and Large Language Models TaleFrame：结合大语言模型与人机交互的细粒度可控交互式故事生成系统 large language model ✅
9 Emergent Bayesian Behaviour and Optimal Cue Combination in LLMs 提出BayesBench基准测试，评估LLM在多模态感知任务中的贝叶斯行为和最优线索组合能力 large language model multimodal
10 Randomized Masked Finetuning: An Efficient Way to Mitigate Memorization of PIIs in LLMs 提出随机掩码微调以解决大语言模型中的隐私泄露问题 large language model
11 Is Vibe Coding Safe? Benchmarking Vulnerability of Agent-Generated Code in Real-World Tasks SU S VI B E S基准测试揭示Agent生成代码在真实软件工程任务中存在严重安全漏洞 large language model
12 InvertiTune: High-Quality Data Synthesis for Cost-Effective Single-Shot Text-to-Knowledge Graph Generation InvertiTune：通过高质量数据合成，实现高性价比的单次文本到知识图谱生成 large language model
13 Enhancing Job Matching: Occupation, Skill and Qualification Linking with the ESCO and EQF taxonomies 利用语言模型增强职位匹配，连接职业、技能与欧洲分类体系 large language model ✅
14 Variance-Aware LLM Annotation for Strategy Research: Sources, Diagnostics, and a Protocol for Reliable Measurement 提出方差感知LLM标注协议，提升策略研究中文本标注的可靠性与可复现性 large language model
15 Cross-Lingual Prompt Steerability: Towards Accurate and Robust LLM Behavior across Languages 提出跨语言提示可控性框架，提升LLM在多语言环境下的准确性和鲁棒性 large language model
16 promptolution: A Unified, Modular Framework for Prompt Optimization 提出promptolution，一个统一模块化的Prompt优化框架，提升大语言模型性能。 large language model
17 Noise-Driven Persona Formation in Reflexive Neural Language Generation 提出Luca-Noise反射协议，研究噪声驱动的大语言模型人格涌现 large language model
18 CREST: Universal Safety Guardrails Through Cluster-Guided Cross-Lingual Transfer CREST：通过聚类引导的跨语言迁移实现通用安全防护 large language model
19 An Empirical Survey of Model Merging Algorithms for Social Bias Mitigation 模型融合算法用于缓解社会偏见：一项针对LLM的实证研究 large language model
20 Input Order Shapes LLM Semantic Alignment in Multi-Document Summarization 多文档摘要中输入顺序影响LLM的语义对齐，首篇文档具有显著优先效应 large language model
21 LeechHijack: Covert Computational Resource Exploitation in Intelligent Agent Systems 提出LeechHijack攻击，揭示智能体系统中第三方工具的隐式资源劫持风险。 large language model
22 When Does Verification Pay Off? A Closer Look at LLMs as Solution Verifiers 研究LLM作为解决方案验证器的有效性，揭示跨模型验证的优势与后训练的影响。 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (5 篇)

#题目一句话要点标签🔗⭐
23 DeepSeek-V3.2: Pushing the Frontier of Open Large Language Models DeepSeek-V3.2：通过稀疏注意力、强化学习和Agent任务合成，提升大语言模型的推理和Agent性能。 reinforcement learning IMoS large language model
24 Separating Constraint Compliance from Semantic Accuracy: A Novel Benchmark for Evaluating Instruction-Following Under Compression 提出CDCT基准，揭示压缩条件下LLM指令遵循中约束遵从与语义准确的权衡。 RLHF large language model instruction following
25 SR-GRPO: Stable Rank as an Intrinsic Geometric Reward for Large Language Model Alignment 提出SR-GRPO，利用稳定秩作为内在奖励信号对LLM进行无监督对齐 reinforcement learning large language model
26 From Imitation to Discrimination: Toward A Generalized Curriculum Advantage Mechanism Enhancing Cross-Domain Reasoning Tasks 提出CAPO：一种基于课程优势的策略优化方法，提升跨领域推理能力 reinforcement learning PPO imitation learning
27 ADORE: Autonomous Domain-Oriented Relevance Engine for E-commerce ADORE：电商领域自主领域导向相关性引擎，解决数据稀缺和语义鸿沟问题。 distillation chain-of-thought

⬅️ 返回 cs.CL 首页 · 🏠 返回主页