cs.CL（2025-10-03）

📊 共 20 篇论文 | 🔗 1 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (18 🔗1) 支柱二：RL算法与架构 (RL & Architecture) (2)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (18 篇)

#	题目	一句话要点	标签	🔗	⭐
1	TS-Reasoner: Aligning Time Series Foundation Models with LLM Reasoning	提出TS-Reasoner，对齐时间序列基础模型与LLM推理能力，解决时间序列理解与推理难题。	large language model foundation model
2	CCD-Bench: Probing Cultural Conflict in Large Language Model Decision-Making	CCD-Bench：评估大语言模型在跨文化冲突情境下的决策能力	large language model
3	Omni-Embed-Nemotron: A Unified Multimodal Retrieval Model for Text, Image, Audio, and Video	Omni-Embed-Nemotron：统一多模态检索模型，支持文本、图像、音频和视频检索	multimodal
4	Cache-to-Cache: Direct Semantic Communication Between Large Language Models	提出Cache-to-Cache (C2C)，实现大语言模型间基于KV-Cache的直接语义通信，提升性能和效率。	large language model	✅
5	Beyond the Final Layer: Intermediate Representations for Better Multilingual Calibration in Large Language Models	提出语言感知的层集成方法LACE，提升大语言模型在多语言环境下的校准性能。	large language model
6	Listening or Reading? Evaluating Speech Awareness in Chain-of-Thought Speech-to-Text Translation	评估思维链语音到文本翻译中的语音感知能力，发现其主要依赖文本转录。	chain-of-thought
7	Grounding Large Language Models in Clinical Evidence: A Retrieval-Augmented Generation System for Querying UK NICE Clinical Guidelines	提出RAG系统，利用LLM高效查询英国NICE临床指南，提升医疗决策效率。	large language model
8	TRepLiNa: Layer-wise CKA+REPINA Alignment Improves Low-Resource Machine Translation in Aya-23 8B	TRepLiNa通过层间CKA+REPINA对齐提升Aya-23 8B在低资源机器翻译中的性能	large language model multimodal
9	Reactive Transformer (RxT) -- Stateful Real-Time Processing for Event-Driven Reactive Language Models	提出Reactive Transformer (RxT)，用于事件驱动的实时状态语言建模，解决长对话中的计算瓶颈。	large language model
10	What is a protest anyway? Codebook conceptualization is still a first-order concern in LLM-era classification	强调LLM时代文本分类中概念化重要性，避免因忽略概念定义导致偏差	large language model
11	Fine-Tuning on Noisy Instructions: Effects on Generalization and Performance	通过噪声指令微调提升大语言模型泛化性和鲁棒性	large language model
12	Scalable multilingual PII annotation for responsible AI in LLMs	提出一种可扩展的多语种PII标注框架，用于提升LLM的负责任AI能力	large language model
13	FocusAgent: Simple Yet Effective Ways of Trimming the Large Context of Web Agents	FocusAgent：利用轻量级LLM检索，有效精简Web Agent上下文，提升效率与安全性	large language model
14	When Names Disappear: Revealing What LLMs Actually Understand About Code	揭示LLM代码理解的局限性：命名消失后的语义推理能力评估	large language model
15	Implicit Values Embedded in How Humans and LLMs Complete Subjective Everyday Tasks	评估LLM在日常任务中体现的隐含价值观，揭示其与人类价值观的差异	large language model
16	Topic Modeling as Long-Form Generation: Can Long-Context LLMs revolutionize NTM via Zero-Shot Prompting?	提出基于长文本生成范式的LLM主题建模方法，通过零样本提示超越传统NTM模型。	large language model
17	Neural Correlates of Language Models Are Specific to Human Language	验证语言模型与人脑活动的关联性，并强调人类语言的独特性	large language model
18	EditLens: Quantifying the Extent of AI Editing in Text	EditLens：量化文本中AI编辑程度，区分人写、AI生成和混合文本	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	🔗	⭐
19	Self-Anchor: Large Language Model Reasoning via Step-by-step Attention Alignment	Self-Anchor：通过逐步注意力对齐增强大语言模型推理能力	reinforcement learning large language model
20	Reward Models are Metrics in a Trench Coat	强调奖励模型与评估指标的关联，提升大语言模型后训练效果	reinforcement learning large language model

⬅️ 返回 cs.CL 首页 · 🏠 返回主页