cs.CL（2024-09-25）

📊 共 28 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) (25 🔗6) 支柱二：RL算法与架构 (RL & Architecture) (2) 支柱一：机器人控制 (Robot Control) (1 🔗1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (25 篇)

#题目一句话要点标签🔗⭐
1 How to Connect Speech Foundation Models and Large Language Models? What Matters and What Does Not 研究语音基础模型与大语言模型连接方式，分析各组件对语音转录任务的影响 large language model foundation model
2 Mitigating the Bias of Large Language Model Evaluation 针对LLM评估偏见，提出校准与对比训练方法，提升评估公平性。 large language model instruction following
3 Speech Recognition Rescoring with Large Speech-Text Foundation Models 利用语音-文本大模型进行语音识别重打分，显著提升ASR性能 large language model foundation model
4 FineZip : Pushing the Limits of Large Language Models for Practical Lossless Text Compression FineZip：利用LLM进行高效无损文本压缩，显著提升压缩速度 large language model
5 From Deception to Detection: The Dual Roles of Large Language Models in Fake News 研究大型语言模型在假新闻生成与检测中的双重角色 large language model
6 Scaling Behavior for Large Language Models regarding Numeral Systems: An Example using Pythia 研究数字系统对大语言模型的影响，发现十进制在训练数据效率上更优 large language model
7 Internalizing ASR with Implicit Chain of Thought for Efficient Speech-to-Speech Conversational LLM 提出隐式思维链的语音LLM，提升端到端语音对话效率 chain-of-thought
8 AutoLLM-CARD: Towards a Description and Landscape of Large Language Models 提出AutoLLM-CARD以解决LLM信息过载问题 large language model ✅
9 Pruning Multilingual Large Language Models for Multilingual Inference 通过剪枝多语言大模型中的关键特征提升非英语语言的零样本推理性能 large language model
10 Enabling Auditory Large Language Models for Automatic Speech Quality Evaluation 利用听觉大语言模型实现自动语音质量评估 large language model ✅
11 Evaluating and Enhancing Large Language Models for Novelty Assessment in Scholarly Publications 提出SchNovel基准和RAG-Novelty方法，评估并提升LLM在学术论文新颖性评估中的能力 large language model
12 HDFlow: Enhancing LLM Complex Problem-Solving with Hybrid Thinking and Dynamic Workflows HDFlow：结合混合思维与动态工作流增强LLM的复杂问题求解能力 large language model chain-of-thought ✅
13 DiaSynth: Synthetic Dialogue Generation Framework for Low Resource Dialogue Applications DiaSynth：用于低资源对话应用的高质量合成对话生成框架 large language model chain-of-thought
14 Discovering the Gems in Early Layers: Accelerating Long-Context LLMs with 1000x Input Token Reduction GemFilter：利用早期层过滤加速长文本LLM，实现千倍输入token缩减 large language model ✅
15 Programming Every Example: Lifting Pre-training Data Quality Like Experts at Scale ProX：通过编程每个样本，提升大规模预训练数据质量，媲美专家水平 large language model ✅
16 Decoding Large-Language Models: A Systematic Overview of Socio-Technical Impacts, Constraints, and Emerging Questions 系统性综述大型语言模型：社会技术影响、约束与新兴问题 large language model
17 Adaptive Self-Supervised Learning Strategies for Dynamic On-Device LLM Personalization 提出自适应自监督学习策略，用于动态的端侧LLM个性化 large language model
18 Investigating OCR-Sensitive Neurons to Improve Entity Recognition in Historical Documents 通过识别并中和OCR敏感神经元，提升Transformer在历史文档实体识别中的性能 large language model
19 Zero-Shot Detection of LLM-Generated Text using Token Cohesiveness 提出TOCSIN，利用token内聚性零样本检测LLM生成文本，提升检测性能。 large language model ✅
20 A Few Hypocrites: Few-Shot Learning and Subtype Definitions for Detecting Hypocrisy Accusations in Online Climate Change Debates 提出气候辩论中伪善指控检测任务，并利用少样本学习和LLM进行有效识别。 large language model
21 E-SQL: Direct Schema Linking via Question Enrichment in Text-to-SQL E-SQL：通过问题增强实现Text-to-SQL中的直接模式链接 large language model
22 RoleBreak: Character Hallucination as a Jailbreak Attack in Role-Playing Systems RoleBreak：通过角色扮演系统中的角色幻觉漏洞进行越狱攻击 large language model
23 Beyond Turing Test: Can GPT-4 Sway Experts' Decisions? GPT-4能否影响专家决策？一项基于读者反应的LLM评估研究 large language model
24 Cross-Lingual and Cross-Cultural Variation in Image Descriptions 大规模跨语言图像描述研究揭示文化和语言对视觉感知的差异 multimodal
25 Understanding the Cognitive Complexity in Language Elicited by Product Images 提出一种衡量产品图像引发语言认知复杂度的可扩展方法，并验证其有效性。 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (2 篇)

#题目一句话要点标签🔗⭐
26 Holistic Automated Red Teaming for Large Language Models through Top-Down Test Case Generation and Multi-turn Interaction HARM：通过自顶向下测试用例生成和多轮交互，实现对大型语言模型的全面自动化红队测试。 reinforcement learning large language model
27 On Extending Direct Preference Optimization to Accommodate Ties 扩展直接偏好优化(DPO)以处理并列偏好，提升翻译与推理任务性能 DPO direct preference optimization

🔬 支柱一：机器人控制 (Robot Control) (1 篇)

#题目一句话要点标签🔗⭐
28 Claim-Guided Textual Backdoor Attack for Practical Applications 提出Claim-Guided Backdoor Attack以解决文本后门攻击的实际应用问题 manipulation large language model ✅

⬅️ 返回 cs.CL 首页 · 🏠 返回主页