cs.CL（2025-04-10）

📊 共 37 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) (30 🔗3) 支柱二：RL算法与架构 (RL & Architecture) (5 🔗1) 支柱一：机器人控制 (Robot Control) (1) 支柱六：视频提取与匹配 (Video Extraction) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (30 篇)

#题目一句话要点标签🔗⭐
1 Unveiling the Impact of Multimodal Features on Chinese Spelling Correction: From Analysis to Design 提出NamBert模型，有效融合多模态特征，提升中文拼写纠错性能 large language model multimodal ✅
2 CollEX -- A Multimodal Agentic RAG System Enabling Interactive Exploration of Scientific Collections CollEx：一种多模态Agentic RAG系统，用于交互式探索科学收藏 multimodal
3 Capybara-OMNI: An Efficient Paradigm for Building Omni-Modal Language Models Capybara-OMNI：一种高效构建全模态语言模型的范式 large language model multimodal instruction following
4 How do Large Language Models Understand Relevance? A Mechanistic Interpretability Perspective 通过机制可解释性分析，揭示大语言模型理解相关性的内在机制 large language model
5 Pangu Ultra: Pushing the Limits of Dense Large Language Models on Ascend NPUs 盘古Ultra：在昇腾NPU上突破稠密大语言模型的极限 large language model
6 The KL3M Data Project: Copyright-Clean Training Resources for Large Language Models KL3M数据项目：构建版权清晰的大语言模型训练资源 large language model
7 On the Temporal Question-Answering Capabilities of Large Language Models Over Anonymized Data 提出RATA数据集，研究LLM在匿名时序数据上的推理能力，并验证集成方法的需求。 large language model
8 ConceptFormer: Towards Efficient Use of Knowledge-Graph Embeddings in Large Language Models 提出ConceptFormer以高效整合知识图谱嵌入至大型语言模型 large language model
9 Evaluating Large Language Models on Multiword Expressions in Multilingual and Code-Switched Contexts 评估大型语言模型在多语言和代码切换环境中处理多词表达的能力 large language model
10 MuSaRoNews: A Multidomain, Multimodal Satire Dataset from Romanian News Articles MuSaRoNews：一个用于罗马尼亚语新闻文章的多领域、多模态讽刺数据集 multimodal
11 Cluster-Driven Expert Pruning for Mixture-of-Experts Large Language Models 提出C-Prune，通过聚类驱动的专家剪枝压缩MoE大语言模型。 large language model
12 Efficient Tuning of Large Language Models for Knowledge-Grounded Dialogue Generation KEDiT：一种高效微调大型语言模型用于知识驱动对话生成的方法 large language model
13 LLM4Ranking: An Easy-to-use Framework of Utilizing Large Language Models for Document Reranking LLM4Ranking：易用的LLM文档重排序框架，支持多种模型与方法 large language model ✅
14 TALE: A Tool-Augmented Framework for Reference-Free Evaluation of Large Language Models 提出TALE框架，通过工具增强LLM评估，无需预先标注的参考答案 large language model
15 A System for Comprehensive Assessment of RAG Frameworks 提出SCARF：一个全面的RAG框架评估系统，解决现有评估方法的局限性。 large language model
16 Has the Creativity of Large-Language Models peaked? An analysis of inter- and intra-LLM variability 大型语言模型创造力评估：模型间差异显著，模型内变异性高，创造力水平未见显著提升 large language model
17 MALIBU Benchmark: Multi-Agent LLM Implicit Bias Uncovered 提出MALIBU基准，揭示多智能体LLM系统中存在的隐性偏见 large language model
18 Token Level Routing Inference System for Edge Devices 提出边缘设备Token级路由推理系统，提升小模型性能并降低资源消耗。 large language model
19 Zero-Shot Cross-Domain Code Search without Fine-Tuning 提出CodeBridge，一种无需微调的零样本跨领域代码搜索方法 large language model
20 Automated Construction of a Knowledge Graph of Nuclear Fusion Energy for Effective Elicitation and Retrieval of Information 提出一种自动构建核聚变能源知识图谱的方法，用于高效的信息提取和检索。 large language model
21 Proactive User Information Acquisition via Chats on User-Favored Topics 提出PIVOT任务，通过用户偏好话题聊天主动获取用户信息，并构建数据集。 large language model
22 Synthetic Fluency: Hallucinations, Confabulations, and the Creation of Irish Words in LLM-Generated Translations 研究LLM在爱尔兰语翻译中产生幻觉性词汇的现象，揭示其对低资源语言的影响。 large language model
23 SaRoHead: Detecting Satire in a Multi-Domain Romanian News Headline Dataset SaRoHead：构建多领域罗马尼亚语新闻标题讽刺检测数据集并提出有效检测方法 large language model
24 Defense against Prompt Injection Attacks via Mixture of Encodings 提出混合编码防御机制，提升LLM抵抗提示注入攻击能力并保持NLP任务性能 large language model
25 Beyond LLMs: A Linguistic Approach to Causal Graph Generation from Narrative Texts 提出一种基于语言学特征的因果图生成框架，提升从叙事文本中提取因果关系的能力。 large language model
26 Model Utility Law: Evaluating LLMs beyond Performance through Mechanism Interpretable Metric 提出模型利用率指标MUI，通过神经元激活比例评估LLM，揭示性能与效率的Utility Law。 large language model ✅
27 LSR-MCTS: Alleviating Long Range Dependency in Code Generation 提出LSR-MCTS算法，缓解代码生成中长程依赖问题，提升代码质量。 large language model
28 AI Coding with Few-Shot Prompting for Thematic Analysis 利用少量样本提示，GPT-3.5 Turbo实现主题分析的AI自动编码。 large language model
29 Enhancing Time Series Forecasting via Multi-Level Text Alignment with LLMs 提出一种基于多层次文本对齐的LLM时间序列预测方法，提升预测精度和可解释性。 large language model
30 Revisiting Prompt Optimization with Large Reasoning Models-A Case Study on Event Extraction 研究表明，大型推理模型在事件抽取任务中仍受益于提示优化 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🔗⭐
31 SFT or RL? An Early Investigation into Training R1-Like Reasoning Large Vision-Language Models 揭示SFT诱导伪推理路径对LVLM强化学习的负面影响，提出VLAA-Thinking数据集和VLAA-Thinker模型。 reinforcement learning distillation multimodal ✅
32 Supervised Optimism Correction: Be Confident When LLMs Are Sure 提出监督乐观校正(SOC)方法，解决LLM中Beam Search的过度乐观问题 reinforcement learning offline reinforcement learning large language model
33 SD$^2$: Self-Distilled Sparse Drafters 提出SD$^2$，通过自蒸馏稀疏化草稿模型提升LLM推断效率，尤其在通用辅助生成场景下。 distillation large language model
34 Findings of the BabyLM Challenge: Sample-Efficient Pretraining on Developmentally Plausible Corpora BabyLM挑战赛：探索在有限数据下高效预训练语言模型的方法 curriculum learning large language model
35 Seed1.5-Thinking: Advancing Superb Reasoning Models with Reinforcement Learning Seed1.5-Thinking：通过强化学习提升卓越推理模型，实现更广泛应用 reinforcement learning

🔬 支柱一：机器人控制 (Robot Control) (1 篇)

#题目一句话要点标签🔗⭐
36 Benchmarking Adversarial Robustness to Bias Elicitation in Large Language Models: Scalable Automated Assessment with LLM-as-a-Judge 提出CLEAR-Bias基准测试框架，评估大型语言模型对抗偏见诱导的鲁棒性 manipulation large language model

🔬 支柱六：视频提取与匹配 (Video Extraction) (1 篇)

#题目一句话要点标签🔗⭐
37 Redefining Machine Translation on Social Network Services with Large Language Models RedTrans：利用大型语言模型重新定义社交网络服务上的机器翻译 HuMoR large language model

⬅️ 返回 cs.CL 首页 · 🏠 返回主页