cs.CL（2024-10-29）

📊 共 31 篇论文 | 🔗 7 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (26 🔗7) 支柱二：RL算法与架构 (RL & Architecture) (4) 支柱七：动作重定向 (Motion Retargeting) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (26 篇)

#题目一句话要点标签🔗⭐
1 Protecting Privacy in Multimodal Large Language Models with MLLMU-Bench 提出MLLMU-Bench基准，用于评估和提升多模态大语言模型的隐私保护能力。 large language model multimodal
2 Beyond Text: Optimizing RAG with Multimodal Inputs for Industrial Applications 针对工业应用，提出基于多模态输入的RAG优化方法，提升问答性能。 large language model multimodal
3 Enhancing Adversarial Attacks through Chain of Thought 提出基于思维链的GCG对抗攻击方法，提升LLM对抗攻击的迁移性和通用性 large language model chain-of-thought ✅
4 Auto-Intent: Automated Intent Discovery and Self-Exploration for Large Language Model Web Agents Auto-Intent：无需微调，自动发现意图并自探索的大语言模型Web Agent large language model
5 A Novel Psychometrics-Based Approach to Developing Professional Competency Benchmark for Large Language Models 提出基于心理测量学的LLM专业能力基准评估方法，应用于教育领域。 large language model
6 Do Large Language Models Align with Core Mental Health Counseling Competencies? CounselingBench：评估大型语言模型在心理健康咨询能力上的表现 large language model ✅
7 Anticipating Future with Large Language Model for Simultaneous Machine Translation 提出TAF：利用大语言模型预测未来词汇，提升同步机器翻译质量。 large language model ✅
8 MIMIC-IV-Ext-PE: Using a large language model to predict pulmonary embolism phenotype in the MIMIC-IV dataset 利用大型语言模型在MIMIC-IV数据集上预测肺栓塞表型 large language model
9 Multimodal Quantum Natural Language Processing: A Novel Framework for using Quantum Methods to Analyse Real Data 提出多模态量子自然语言处理框架，利用量子方法分析真实数据中的语言组合性。 multimodal
10 ProMQA: Question Answering Dataset for Multimodal Procedural Activity Understanding 提出ProMQA多模态问答数据集，用于评估程序性活动理解能力。 multimodal
11 Improving Math Problem Solving in Large Language Models Through Categorization and Strategy Tailoring 提出基于分类和策略定制的方法，提升大型语言模型在数学问题求解中的能力 large language model
12 Linear Chain Transformation: Expanding Optimization Dynamics for Fine-Tuning Large Language Models LinChain：通过线性链变换扩展优化动态，提升大语言模型微调性能 large language model
13 Personalization of Large Language Models: A Survey 综述：大型语言模型个性化研究，填补文本生成与下游应用间的空白。 large language model
14 SG-Bench: Evaluating LLM Safety Generalization Across Diverse Tasks and Prompt Types SG-Bench：提出一个综合性评测基准，评估LLM在不同任务和提示类型下的安全性泛化能力。 large language model chain-of-thought
15 Attention Speaks Volumes: Localizing and Mitigating Bias in Language Models 提出ATLAS方法，通过注意力机制干预缓解大语言模型中的偏见问题 large language model
16 Scaling LLM Inference with Optimized Sample Compute Allocation 提出OSCA算法，通过优化采样计算分配显著提升大语言模型推理效率。 large language model ✅
17 Toxicity of the Commons: Curating Open-Source Pre-Training Data 提出开放源代码数据过滤管道以减少有害输出 large language model
18 DISCERN: Decoding Systematic Errors in Natural Language for Text Classifiers DISCERN：利用自然语言解释解码文本分类器中的系统性误差 large language model
19 Benchmarking LLM Guardrails in Handling Multilingual Toxicity 构建多语言毒性测试基准，评估LLM安全防护机制的有效性与鲁棒性 large language model
20 The Impact of Inference Acceleration on Bias of LLMs 推理加速优化可能显著且不可预测地改变LLM的偏见 large language model
21 Distinguishing Ignorance from Error in LLM Hallucinations 区分LLM幻觉中的无知与错误，提升幻觉检测与缓解效果 large language model ✅
22 SceneGenAgent: Precise Industrial Scene Generation with Coding Agent SceneGenAgent：基于代码生成精确工业场景，解决LLM在工业场景应用的难题 large language model ✅
23 Self-Preference Bias in LLM-as-a-Judge 提出一种定量指标以评估LLM作为评估者时的自偏好偏差，揭示偏差源于对低困惑度文本的偏爱。 large language model
24 Leveraging LLMs for Hypothetical Deduction in Logical Inference: A Neuro-Symbolic Approach LINA：利用LLM进行假设演绎的神经符号逻辑推理方法 large language model ✅
25 Learning and Unlearning of Fabricated Knowledge in Language Models 研究语言模型中虚构知识的学习与遗忘，并提出多步稀疏更新方法缓解数据中毒。 large language model
26 A Bayesian Approach to Harnessing the Power of LLMs in Authorship Attribution 提出基于贝叶斯方法的LLM作者身份识别框架，实现卓越的单样本分类精度 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🔗⭐
27 Let's Be Self-generated via Step by Step: A Curriculum Learning Approach to Automated Reasoning with Large Language Models 提出LBS3，一种基于课程学习的自动化推理提示方法，提升大语言模型在复杂推理任务中的性能。 curriculum learning large language model chain-of-thought
28 Flow-DPO: Improving LLM Mathematical Reasoning through Online Multi-Agent Learning Flow-DPO：通过在线多智能体学习提升LLM数学推理能力 DPO direct preference optimization large language model
29 SimSiam Naming Game: A Unified Approach for Representation Learning and Emergent Communication 提出SimSiam+VAE，统一表征学习与涌现通信，提升模型性能。 representation learning
30 Choosy Babies Need One Coach: Inducing Mode-Seeking Behavior in BabyLlama with Reverse KL Divergence 利用反向KL散度诱导BabyLlama的Mode-Seeking行为，单教师优于多教师。 teacher-student distillation

🔬 支柱七：动作重定向 (Motion Retargeting) (1 篇)

#题目一句话要点标签🔗⭐
31 Natural Language Inference Improves Compositionality in Vision-Language Models 提出CECE方法，利用自然语言推理提升视觉-语言模型中的组合性推理能力 spatial relationship large language model

⬅️ 返回 cs.CL 首页 · 🏠 返回主页