cs.CL（2026-01-05）

📊 共 19 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) (14) 支柱二：RL算法与架构 (RL & Architecture) (4 🔗2) 支柱一：机器人控制 (Robot Control) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (14 篇)

#题目一句话要点标签🔗⭐
1 Power-of-Two Quantization-Aware-Training (PoT-QAT) in Large Language Models (LLMs) 提出面向LLM的二次幂量化感知训练(PoT-QAT)以加速边缘设备推理 large language model
2 Toward Global Large Language Models in Medicine 构建GlobMed多语言医学数据集与LLM，提升低资源语言医疗场景性能 large language model
3 From XAI to Stories: A Factorial Study of LLM-Generated Explanation Quality 系统性研究LLM生成解释质量的影响因素，揭示XAI在时间序列预测中的局限性 large language model chain-of-thought
4 Routing by Analogy: kNN-Augmented Expert Assignment for Mixture-of-Experts 提出kNN-MoE，通过检索增强路由解决MoE模型在分布偏移下的脆弱性问题 large language model
5 Cost-Efficient Cross-Lingual Retrieval-Augmented Generation for Low-Resource Languages: A Case Study in Bengali Agricultural Advisory 提出一种低成本跨语言RAG框架，解决孟加拉语农业咨询难题 large language model
6 Not All Needles Are Found: How Fact Distribution and Don't Make It Up Prompts Shape Literal Extraction, Logical Inference, and Hallucination Risks in Long-Context LLMs 研究长文本LLM在事实分布和反幻觉提示下的信息抽取、逻辑推理及幻觉风险 large language model
7 Tackling the Inherent Difficulty of Noise Filtering in RAG 提出RAG噪声过滤微调方法，提升LLM在噪声环境下的鲁棒性 large language model
8 K-EXAONE Technical Report LG AI Research 提出 K-EXAONE：一个支持六种语言的236B参数规模多语言大模型。 foundation model
9 Robust Persona-Aware Toxicity Detection with Prompt Optimization and Learned Ensembling 提出基于提示优化和集成学习的鲁棒的、考虑个体差异的毒性检测方法 large language model
10 Confidence Estimation for LLMs in Multi-turn Interactions 针对多轮交互中LLM的置信度估计问题，提出一种新的评估框架和置信度探测方法。 large language model
11 Towards Multi-Level Transcript Segmentation: LoRA Fine-Tuning for Table-of-Contents Generation 提出基于LoRA微调的多层次转录分割方法，用于生成目录。 large language model
12 Judging with Personality and Confidence: A Study on Personality-Conditioned LLM Relevance Assessment 提出人格条件LLM相关性评估方法，提升评估可靠性与人对齐程度。 large language model
13 Can LLMs Track Their Output Length? A Dynamic Feedback Mechanism for Precise Length Regulation 提出动态反馈机制以解决LLMs输出长度控制问题 large language model
14 A Training-Free Large Reasoning Model-based Knowledge Tracing Framework for Unified Prediction and Prescription 提出Thinking-KT：基于无训练大模型推理的知识追踪统一框架 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🔗⭐
15 Agentic Memory: Learning Unified Long-Term and Short-Term Memory Management for Large Language Model Agents 提出Agentic Memory，解决LLM Agent长期和短期记忆统一管理问题 reinforcement learning large language model
16 CD4LM: Consistency Distillation and aDaptive Decoding for Diffusion Language Models 提出CD4LM框架，通过一致性蒸馏和自适应解码加速扩散语言模型的并行生成。 distillation large language model ✅
17 Hidden State Poisoning Attacks against Mamba-based Language Models 针对Mamba语言模型的隐藏状态投毒攻击研究 Mamba SSM state space model
18 DermoGPT: Open Weights and Open Data for Morphology-Grounded Dermatological Reasoning MLLMs DermoGPT：提出基于形态学的皮肤科推理多模态大语言模型，并开源数据与权重 reinforcement learning large language model multimodal ✅

🔬 支柱一：机器人控制 (Robot Control) (1 篇)

#题目一句话要点标签🔗⭐
19 Emergent Introspective Awareness in Large Language Models 通过注入概念表征并测量模型自报告状态，探索大语言模型内省意识 manipulation large language model

⬅️ 返回 cs.CL 首页 · 🏠 返回主页