cs.CL（2024-10-07）

📊 共 39 篇论文 | 🔗 3 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (33 🔗2) 支柱二：RL算法与架构 (RL & Architecture) (6 🔗1)

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

#题目一句话要点标签🔗⭐
1 MINER: Mining the Underlying Pattern of Modality-Specific Neurons in Multimodal Large Language Models 提出MINER框架以解决多模态大语言模型的可解释性问题 large language model multimodal
2 Rational Metareasoning for Large Language Models 提出基于元推理的LLM优化方法，降低推理成本并保持性能 large language model chain-of-thought
3 Grounding Partially-Defined Events in Multimodal Data 提出MultiVENT-G基准，用于多模态数据中部分定义事件的定位与理解。 multimodal
4 Data Advisor: Dynamic Data Curation for Safety Alignment of Large Language Models Data Advisor：面向大语言模型安全对齐的动态数据管理方法 large language model
5 Narrative-of-Thought: Improving Temporal Reasoning of Large Language Models via Recounted Narratives 提出Narrative-of-Thought，提升大语言模型在时间推理任务上的性能 large language model ✅
6 Representing the Under-Represented: Cultural and Core Capability Benchmarks for Developing Thai Large Language Models 为泰语大语言模型开发，提出文化和核心能力基准测试集Thai-H6和ThaiCLI large language model
7 Attribute Controlled Fine-tuning for Large Language Models: A Case Study on Detoxification 提出基于属性控制的LLM微调框架，以提升模型输出安全性 large language model
8 Mitigating the Risk of Health Inequity Exacerbated by Large Language Models 提出EquityGuard框架，用于检测和缓解大语言模型在医疗应用中加剧的健康不公平风险。 large language model
9 Efficient Inference for Large Language Model-based Generative Recommendation 提出AtSpeed框架，加速基于大语言模型的生成式推荐系统推理。 large language model ✅
10 Investigating large language models for their competence in extracting grammatically sound sentences from transcribed noisy utterances 评估大型语言模型从含噪语音转录中提取语法正确句子的能力 large language model
11 Explanation sensitivity to the randomness of large language models: the case of journalistic text classification 揭示LLM随机性对解释性的影响：以新闻文本分类为例 large language model
12 Initialization of Large Language Models via Reparameterization to Mitigate Loss Spikes 提出WeSaR重参数化方法，缓解大语言模型预训练中的Loss Spikes问题 large language model
13 Formality is Favored: Unraveling the Learning Preferences of Large Language Models on Data with Conflicting Knowledge 揭示大语言模型在冲突知识数据上的学习偏好：偏向形式化文本 large language model
14 Superficial Safety Alignment Hypothesis 提出浅层安全对齐假设(SSAH)，揭示LLM安全对齐的关键神经元组件。 large language model instruction following
15 SFTMix: Elevating Language Model Instruction Tuning with Mixup Recipe SFTMix：利用Mixup提升语言模型指令微调效果，无需高质量数据集。 large language model instruction following
16 $\textbf{Only-IF}$:Revealing the Decisive Effect of Instruction Diversity on Generalization 揭示指令多样性对LLM泛化能力的关键影响，指导指令调优数据收集 large language model instruction following
17 On Instruction-Finetuning Neural Machine Translation Models 提出指令微调方法，将LLM的指令遵循能力迁移至更小的NMT模型，实现定制化翻译。 large language model instruction following
18 A Recipe For Building a Compliant Real Estate Chatbot 构建合规房地产聊天机器人，解决歧视性行为，媲美GPT-4o large language model instruction following
19 TurtleBench: Evaluating Top Language Models via Real-World Yes/No Puzzles 提出TurtleBench，通过真实用户谜题评估大语言模型的推理能力 large language model chain-of-thought
20 RevisEval: Improving LLM-as-a-Judge via Response-Adapted References RevisEval：通过响应适配的参考文本提升LLM作为评估器的可靠性 large language model instruction following
21 Fill In The Gaps: Model Calibration and Generalization with Synthetic Data 提出基于合成数据的模型校准方法，提升泛化能力并保持模型精度。 large language model
22 Causal Micro-Narratives 提出一种基于主题本体的因果微叙事分类方法，应用于通货膨胀叙事分析。 large language model
23 Cookbook: A framework for improving LLM generative abilities via programmatic data generating templates Cookbook：通过程序化数据生成模板提升LLM生成能力 large language model
24 ScienceAgentBench: Toward Rigorous Assessment of Language Agents for Data-Driven Scientific Discovery ScienceAgentBench：面向数据驱动科学发现的语言智能体严格评估基准 large language model
25 Reasoning Paths Optimization: Learning to Reason and Explore From Diverse Paths 提出推理路径优化方法以提升复杂问题求解能力 large language model
26 Debate, Deliberate, Decide (D3): A Cost-Aware Adversarial Framework for Reliable and Interpretable LLM Evaluation 提出D3框架，通过多智能体辩论实现可靠、可解释且低成本的大语言模型评估。 large language model
27 Exploring the Personality Traits of LLMs through Latent Features Steering 通过潜在特征引导探索LLM的个性特质 large language model
28 Post-hoc Study of Climate Microtargeting on Social Media Ads with LLMs: Thematic Insights and Fairness Evaluation 利用大型语言模型分析社交媒体气候微定向广告，揭示主题偏好与公平性问题。 large language model
29 Mirror-Consistency: Harnessing Inconsistency in Majority Voting 提出Mirror-Consistency，利用大语言模型投票不一致性提升推理能力和置信度校准 large language model
30 Rationale-Aware Answer Verification by Pairwise Self-Evaluation 提出REPS方法，通过成对自评估提升答案验证器对推理过程合理性的判断能力 large language model
31 PECAN: LLM-Guided Dynamic Progress Control with Attention-Guided Hierarchical Weighted Graph for Long-Document QA PECAN：基于LLM引导的动态进度控制和注意力引导分层加权图的长文档问答方法 large language model
32 The LLM Effect: Are Humans Truly Using LLMs, or Are They Being Influenced By Them Instead? 研究表明LLM辅助分析虽提速，但可能引入锚定偏差，影响分析深度。 large language model
33 MathHay: An Automated Benchmark for Long-Context Mathematical Reasoning in LLMs 提出MathHay：一个用于评估LLM长文本数学推理能力的自动化基准。 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (6 篇)

#题目一句话要点标签🔗⭐
34 Falcon Mamba: The First Competitive Attention-free 7B Language Model Falcon Mamba 7B：首个具有竞争力的无注意力7B语言模型 Mamba large language model ✅
35 ReasoningRank: Teaching Student Models to Rank through Reasoning-Based Knowledge Distillation ReasoningRank：通过基于推理的知识蒸馏提升学生模型排序能力 distillation large language model
36 As Simple as Fine-tuning: LLM Alignment via Bidirectional Negative Feedback Loss 提出双向负反馈损失BNF，简化LLM对齐过程并提升数学数据集上的稳定性。 reinforcement learning PPO RLHF
37 Contrastive Learning to Improve Retrieval for Real-world Fact Checking 提出对比式事实核查重排序器(CFR)，提升真实世界事实核查的检索性能 contrastive learning distillation
38 LLMs Are In-Context Bandit Reinforcement Learners 研究表明：大型语言模型是上下文情境中的Bandit强化学习器 reinforcement learning large language model
39 FAME: Towards Factual Multi-Task Model Editing 提出FAME数据集与SKEME方法，提升LLM在多任务模型编辑中的实用性 world model large language model

⬅️ 返回 cs.CL 首页 · 🏠 返回主页