cs.CL（2026-03-09）

📊 共 20 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) (12 🔗2) 支柱二：RL算法与架构 (RL & Architecture) (7 🔗2) 支柱五：交互与反应 (Interaction & Reaction) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (12 篇)

#题目一句话要点标签🔗⭐
1 BRIDGE: Benchmark for multi-hop Reasoning In long multimodal Documents with Grounded Evidence 提出BRIDGE基准，用于评估长多模态文档中多跳推理能力。 large language model multimodal
2 Using Multimodal and Language-Agnostic Sentence Embeddings for Abstractive Summarization 提出SBARThez框架，利用多模态和语言无关的句子嵌入改进抽象摘要生成，提升事实一致性。 multimodal
3 Learning Multiple Utterance-Level Attribute Representations with a Unified Speech Encoder 提出统一语音编码器框架，学习多重语句级属性表示，提升跨语言检索和说话人识别性能。 foundation model multimodal
4 Is continuous CoT better suited for multi-lingual reasoning? 连续CoT提升多语言推理能力，尤其在低资源和零样本场景下表现突出 chain-of-thought
5 How Much Do LLMs Hallucinate in Document Q&A Scenarios? A 172-Billion-Token Study Across Temperatures, Context Lengths, and Hardware Platforms RIKER评估框架揭示LLM在文档问答中幻觉比例随上下文长度显著增加 large language model
6 CCR-Bench: A Comprehensive Benchmark for Evaluating LLMs on Complex Constraints, Control Flows, and Real-World Cases 提出CCR-Bench基准，评估LLM在复杂约束、控制流和真实场景下的指令遵循能力。 large language model
7 Fanar-Sadiq: A Multi-Agent Architecture for Grounded Islamic QA Fanar-Sadiq：一个面向伊斯兰问答的多智能体架构，解决幻觉和来源错误问题。 large language model
8 Sandpiper: Orchestrated AI-Annotation for Educational Discourse at Scale Sandpiper：用于大规模教育对话的协同式AI标注系统 large language model
9 Adaptive Loops and Memory in Transformers: Think Harder or Know More? 提出自适应循环与记忆Transformer，提升数学推理与常识任务性能 chain-of-thought
10 LAMUS: A Large-Scale Corpus for Legal Argument Mining from U.S. Caselaw using LLMs LAMUS：利用LLM构建美国判例法大规模法律论证挖掘语料库 chain-of-thought ✅
11 NCL-UoR at SemEval-2026 Task 5: Embedding-Based Methods, Fine-Tuning, and LLMs for Word Sense Plausibility Rating 针对词义合理性评级，提出基于结构化提示和决策规则的大语言模型方法。 large language model ✅
12 EvoScientist: Towards Multi-Agent Evolving AI Scientists for End-to-End Scientific Discovery EvoScientist：面向端到端科学发现的多智能体进化AI科学家框架 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (7 篇)

#题目一句话要点标签🔗⭐
13 SmartThinker: Progressive Chain-of-Thought Length Calibration for Efficient Large Language Model Reasoning SmartThinker：通过渐进式CoT长度校准提升大语言模型推理效率 reward design large language model chain-of-thought ✅
14 Revealing Behavioral Plasticity in Large Language Models: A Token-Conditional Perspective 提出Token条件强化学习(ToCoRL)，实现大语言模型行为模式的精准控制与迁移。 reinforcement learning large language model
15 High-Fidelity Pruning for Large Language Models 提出基于信息熵的LLM高保真剪枝方法，提升部署效率 distillation large language model ✅
16 TildeOpen LLM: Leveraging Curriculum Learning to Achieve Equitable Language Representation TildeOpen LLM：利用课程学习实现公平的语言表征 curriculum learning large language model
17 Toward Robust LLM-Based Judges: Taxonomic Bias Evaluation and Debiasing Optimization 提出JudgeBiasBench基准，并优化LLM评判偏见，提升自动化评估可靠性。 reinforcement learning contrastive learning large language model
18 ConflictBench: Evaluating Human-AI Conflict via Interactive and Visually Grounded Environments 提出ConflictBench，用于评估人机交互中基于视觉环境的冲突对齐问题 world model large language model
19 Aligning to Illusions: Choice Blindness in Human and AI Feedback 提出选择盲目性研究以挑战人类反馈在RLHF中的假设 reinforcement learning RLHF

🔬 支柱五：交互与反应 (Interaction & Reaction) (1 篇)

#题目一句话要点标签🔗⭐
20 AdaCultureSafe: Adaptive Cultural Safety Grounded by Cultural Knowledge in Large Language Models AdaCultureSafe：基于文化知识自适应提升大语言模型的文化安全性 ReMoS large language model

⬅️ 返回 cs.CL 首页 · 🏠 返回主页