cs.CL（2025-12-23）

📊 共 15 篇论文 | 🔗 4 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (8 🔗1) 支柱二：RL算法与架构 (RL & Architecture) (6 🔗3) 支柱一：机器人控制 (Robot Control) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (8 篇)

#	题目	一句话要点	标签	🔗	⭐
1	Retrieval-augmented Prompt Learning for Pre-trained Foundation Models	提出RetroPrompt，通过检索增强提示学习提升预训练模型泛化能力	foundation model multimodal
2	M$^3$KG-RAG: Multi-hop Multimodal Knowledge Graph-enhanced Retrieval-Augmented Generation	提出M$^3$KG-RAG，通过多跳多模态知识图谱增强检索增强生成，提升MLLM在视听领域的推理和 grounding 能力。	large language model multimodal
3	Can LLMs Solve My Grandma's Riddle? Evaluating Multilingual Large Language Models on Reasoning Traditional Bangla Tricky Riddles	BanglaRiddleEval：评估多语言大模型在孟加拉语谜语推理上的能力	large language model chain-of-thought	✅
4	Making Large Language Models Efficient Dense Retrievers	提出EffiR框架，通过MLP压缩提升LLM密集检索器的效率，保持性能。	large language model
5	Cube Bench: A Benchmark for Spatial Visual Reasoning in MLLMs	提出Cube Bench：用于评估多模态大语言模型空间视觉推理能力的魔方基准测试。	large language model multimodal
6	Schoenfeld's Anatomy of Mathematical Reasoning by Language Models	提出ThinkARM框架，剖析语言模型数学推理过程中的认知结构与步骤	large language model
7	AI Security Beyond Core Domains: Resume Screening as a Case Study of Adversarial Vulnerabilities in Specialized LLM Applications	揭示LLM在简历筛选等专业应用中对抗性漏洞，并提出有效防御方法	large language model
8	Coherence in the brain unfolds across separable temporal regimes	利用LLM提取的漂移和位移信号，揭示大脑中语言连贯性处理的时域分离机制	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	🔗	⭐
9	FaithLens: Detecting and Explaining Faithfulness Hallucination	提出FaithLens，用于检测并解释大语言模型中的忠实性幻觉。	reinforcement learning large language model
10	Fun-Audio-Chat Technical Report	Fun-Audio-Chat：通过双分辨率语音表示和核心鸡尾酒训练，提升语音交互大模型性能	DPO instruction following
11	Multi-hop Reasoning via Early Knowledge Alignment	提出早期知识对齐(EKA)模块，提升迭代RAG多跳推理性能与效率。	reinforcement learning large language model	✅
12	Memory-T1: Reinforcement Learning for Temporal Reasoning in Multi-session Agents	提出Memory-T1框架，利用强化学习解决多轮对话Agent中的时序推理难题。	reinforcement learning	✅
13	Distilling to Hybrid Attention Models via KL-Guided Layer Selection	提出基于KL散度引导的层选择方法，用于将Softmax注意力Transformer蒸馏为混合注意力模型。	linear attention distillation
14	SpidR: Learning Fast and Stable Linguistic Units for Spoken Language Models Without Supervision	SpidR：无需监督，学习快速稳定的语音单元用于语音语言模型	representation learning distillation	✅

🔬 支柱一：机器人控制 (Robot Control) (1 篇)

#	题目	一句话要点	标签	🔗	⭐
15	AprielGuard	提出AprielGuard，统一安全风险与对抗威胁，提升LLM安全防护能力	manipulation large language model

⬅️ 返回 cs.CL 首页 · 🏠 返回主页