cs.CL（2024-10-13）

📊 共 17 篇论文 | 🔗 2 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (16 🔗2) 支柱二：RL算法与架构 (RL & Architecture) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (16 篇)

#题目一句话要点标签🔗⭐
1 Empowering Dysarthric Speech: Leveraging Advanced LLMs for Accurate Speech Correction and Multimodal Emotion Analysis 利用大型语言模型进行构音障碍语音校正和多模态情感分析 large language model multimodal
2 ELF-Gym: Evaluating Large Language Models Generated Features for Tabular Prediction 提出ELF-Gym，用于评估大型语言模型生成表格数据预测的特征，并揭示其与人类专家特征工程的差距。 large language model
3 Safety-Aware Fine-Tuning of Large Language Models 提出安全感知微调框架SAFT，自动移除有害数据以提升LLM安全性。 large language model
4 Investigating Implicit Bias in Large Language Models: A Large-Scale Study of Over 50 LLMs 大规模研究揭示：大型语言模型中隐性偏见并未随模型增大而减少 large language model
5 Reverse Modeling in Large Language Models 揭示大语言模型逆向建模难题，提出基于损失差异的数据选择方法显著提升性能 large language model
6 Conversational Code Generation: a Case Study of Designing a Dialogue System for Generating Driving Scenarios for Testing Autonomous Vehicles 提出一种基于对话的自动驾驶场景生成系统，显著提升场景生成成功率。 large language model instruction following
7 'Quis custodiet ipsos custodes?' Who will watch the watchmen? On Detecting AI-generated peer-reviews 提出TF和RR模型，检测AI生成的同行评审，保障学术诚信。 large language model
8 ChroKnowledge: Unveiling Chronological Knowledge of Language Models in Multiple Domains 提出ChroKnowledge框架，用于评估大型语言模型在多领域的时间知识掌握程度。 large language model
9 Expanding Search Space with Diverse Prompting Agents: An Efficient Sampling Approach for LLM Mathematical Reasoning 提出基于多样化Prompting Agent的高效采样方法，提升LLM数学推理能力 large language model
10 Honest AI: Fine-Tuning "Small" Language Models to Say "I Don't Know", and Reducing Hallucination in RAG Honest AI：通过微调“小型”语言模型使其回答“我不知道”来减少RAG中的幻觉 large language model
11 Evaluating Gender Bias of LLMs in Making Morality Judgements GenMO数据集揭示LLM在道德判断中存在显著性别偏见，尤其偏袒女性 large language model
12 MisinfoEval: Generative AI in the Era of "Alternative Facts" MisinfoEval：利用生成式AI对抗“另类事实”时代的信息误导 large language model
13 Learning to Rank for Multiple Retrieval-Augmented Models through Iterative Utility Maximization 提出一种迭代效用最大化方法，为多个RAG模型学习排序，实现检索结果的个性化。 large language model
14 Reddit is all you need: Authorship profiling for Romanian 提出首个罗马尼亚语作者画像语料库，并探索LLM在该任务上的基线性能。 large language model
15 RMB: Comprehensively Benchmarking Reward Models in LLM Alignment RMB：全面评估LLM对齐中奖励模型的基准，揭示现有模型的泛化缺陷。 large language model ✅
16 LLM-Based Multi-Agent Systems are Scalable Graph Generative Models 提出GraphAgent-Generator，利用LLM零样本生成大规模、符合真实社会属性的动态图。 large language model ✅

🔬 支柱二：RL算法与架构 (RL & Architecture) (1 篇)

#题目一句话要点标签🔗⭐
17 Taming Overconfidence in LLMs: Reward Calibration in RLHF 提出PPO-M和PPO-C，解决RLHF训练LLM中的过度自信问题。 reinforcement learning PPO RLHF

⬅️ 返回 cs.CL 首页 · 🏠 返回主页