cs.CL（2024-10-24）

📊 共 39 篇论文 | 🔗 8 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (33 🔗6) 支柱二：RL算法与架构 (RL & Architecture) (4 🔗1) 支柱六：视频提取与匹配 (Video Extraction) (1) 支柱一：机器人控制 (Robot Control) (1 🔗1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (33 篇)

#题目一句话要点标签🔗⭐
1 PRISM: A Methodology for Auditing Biases in Large Language Models PRISM：一种用于审计大型语言模型偏见的任务型探查方法 large language model
2 Watermarking Large Language Models and the Generated Content: Opportunities and Challenges 综述大型语言模型及其生成内容的水印技术：机遇与挑战 large language model
3 A Survey of Multimodal Sarcasm Detection 综述性研究：全面回顾2018-2023年多模态讽刺检测方法与未来方向。 multimodal
4 Delving into the Reversal Curse: How Far Can Large Language Models Generalize? 揭示大语言模型中的反转诅咒现象，探究其泛化能力边界 large language model ✅
5 Task Calibration: Calibrating Large Language Models on Inference Tasks 提出任务校准(TC)方法，通过任务重构提升大语言模型在推理任务上的性能。 large language model
6 Infinity-MM: Scaling Multimodal Performance with Large-Scale and High-Quality Instruction Data Infinity-MM：通过大规模高质量指令数据提升多模态模型性能 multimodal ✅
7 CCI3.0-HQ: a large-scale Chinese dataset of high quality designed for pre-training large language models 提出CCI3.0-HQ高质量中文预训练数据集，提升小模型性能。 large language model ✅
8 ChineseSafe: A Chinese Benchmark for Evaluating Safety in Large Language Models 提出ChineseSafe中文安全基准，评估大型语言模型识别中文语境下不安全内容的能力。 large language model ✅
9 Large Language Models Reflect the Ideology of their Creators 大型语言模型反映其创建者的意识形态倾向 large language model
10 AdaEDL: Early Draft Stopping for Speculative Decoding of Large Language Models via an Entropy-based Lower Bound on Token Acceptance Probability AdaEDL：基于熵的下界概率自适应停止推测解码，提升大语言模型推理效率。 large language model
11 DeCoRe: Decoding by Contrasting Retrieval Heads to Mitigate Hallucinations 提出DeCoRe，通过对比检索头解码来缓解大语言模型的幻觉问题 large language model instruction following
12 BioMistral-NLU: Towards More Generalizable Medical Language Understanding through Instruction Tuning BioMistral-NLU：通过指令微调提升医学语言理解的泛化能力 large language model instruction following
13 Distill Visual Chart Reasoning Ability from LLMs to MLLMs 提出CIT方法，利用代码作为媒介，从LLM蒸馏视觉图表推理能力到MLLM large language model multimodal ✅
14 WAFFLE: Finetuning Multi-Modal Model for Automated Front-End Development WAFFLE：微调多模态模型，实现自动化前端开发 large language model
15 Read-ME: Refactorizing LLMs as Router-Decoupled Mixture of Experts with System Co-Design Read-ME：通过系统协同设计，将LLM重构为解耦路由的混合专家模型，提升推理效率。 large language model ✅
16 Lived Experience Not Found: LLMs Struggle to Align with Experts on Addressing Adverse Drug Reactions from Psychiatric Medication Use 提出Psych-ADR基准和ADRA框架，评估LLM在精神药物不良反应处理中的专家对齐度 large language model
17 Dynamic Vocabulary Pruning in Early-Exit LLMs 提出动态词汇表剪枝方法，加速早退LLM的推理效率并保持性能。 large language model
18 From Blind Solvers to Logical Thinkers: Benchmarking LLMs' Logical Integrity on Faulty Mathematical Problems FaultyMath：评估LLM在错误数学问题上的逻辑一致性，揭示其“盲解”本质 large language model
19 From English-Centric to Effective Bilingual: LLMs with Custom Tokenizers for Underrepresented Languages 提出一种低成本、模型无关的双语LLM构建方法，提升低资源语言性能。 large language model
20 Does Differential Privacy Impact Bias in Pretrained NLP Models? 研究表明：在预训练NLP模型中，差分隐私训练会加剧模型对特定群体的偏见。 large language model
21 Unleashing LLM Reasoning Capability via Scalable Question Synthesis from Scratch 提出ScaleQuest，通过可扩展的自生问题合成方法提升LLM的数学推理能力 large language model
22 Prompting and Fine-Tuning of Small LLMs for Length-Controllable Telephone Call Summarization 利用Prompt和微调小LLM实现长度可控的电话呼叫总结 large language model
23 Label Set Optimization via Activation Distribution Kurtosis for Zero-shot Classification with Generative Models 提出LOADS方法，通过激活分布峰度优化标签集，提升生成模型零样本分类性能。 large language model
24 Does Data Contamination Detection Work (Well) for LLMs? A Survey and Evaluation on Detection Assumptions 系统性评估LLM数据污染检测方法，揭示现有方法在实际应用中的局限性 large language model
25 Bridge-Coder: Unlocking LLMs' Potential to Overcome Language Gaps in Low-Resource Code Bridge-Coder：利用LLM弥合低资源代码语言的语言差距 large language model
26 LLMs for Extremely Low-Resource Finno-Ugric Languages 针对极低资源芬兰-乌戈尔语族，提出LLM构建、指令调优与评测的完整方案 large language model
27 Are LLMs Better than Reported? Detecting Label Errors and Mitigating Their Effect on Model Performance 利用大语言模型检测并纠正数据集中的标签错误，提升模型性能 large language model
28 An LLM Agent for Automatic Geospatial Data Analysis 提出GeoAgent框架，解决LLM在复杂地理空间数据分析中逻辑错误和幻觉问题。 large language model
29 Why Does the Effective Context Length of LLMs Fall Short? 提出STRING，通过转移旋转位置嵌入显著提升LLM长文本处理能力 large language model
30 A Systematic Survey on Instructional Text: From Representation Formats to Downstream NLP Tasks 系统性综述指令文本：从表示格式到下游NLP任务，填补复杂指令理解领域空白。 large language model
31 ToolFlow: Boosting LLM Tool-Calling Through Natural and Coherent Dialogue Synthesis ToolFlow：通过自然连贯的对话合成，提升LLM的工具调用能力 large language model
32 MoMQ: Mixture-of-Experts Enhances Multi-Dialect Query Generation across Relational and Non-Relational Databases 提出MoMQ框架以解决多方言查询生成问题 large language model
33 Assessing the Creativity of LLMs in Proposing Novel Solutions to Mathematical Problems 提出CreativeMath基准，评估LLM在数学问题上提出创新解法的能力 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🔗⭐
34 GCoder: Improving Large Language Model for Generalized Graph Problem Solving GCoder：提升大语言模型在通用图问题求解中的能力 reinforcement learning large language model ✅
35 Taipan: Efficient and Expressive State Space Language Models with Selective Attention Taipan：结合选择性注意力的高效表达型状态空间语言模型，解决长文本建模难题。 Mamba SSM state space model
36 AlignCap: Aligning Speech Emotion Captioning to Human Preferences AlignCap：通过对齐人类偏好提升语音情感描述的准确性和泛化性 distillation large language model
37 Dialog2Flow: Pre-training Soft-Contrastive Action-Driven Sentence Embeddings for Automatic Dialog Flow Extraction 提出Dialog2Flow，通过预训练软对比动作驱动的句子嵌入，实现自动对话流程提取。 representation learning 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🔗⭐
38 Retrieving Implicit and Explicit Emotional Events Using Large Language Models 利用大型语言模型进行隐式和显式情感事件检索 motion retrieval large language model

🔬 支柱一：机器人控制 (Robot Control) (1 篇)

#题目一句话要点标签🔗⭐
39 From Imitation to Introspection: Probing Self-Consciousness in Language Models 首个探索语言模型自我意识的研究，利用因果结构游戏定义并评估核心概念。 manipulation ✅

⬅️ 返回 cs.CL 首页 · 🏠 返回主页