cs.CL（2025-12-27）

📊 共 14 篇论文

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (12) 支柱二：RL算法与架构 (RL & Architecture) (2)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (12 篇)

#	题目	一句话要点	标签	🔗	⭐
1	Chain-of-thought Reviewing and Correction for Time Series Question Answering	提出T3LLM框架，通过显式纠错机制提升时间序列问答的推理能力	large language model chain-of-thought
2	Structured Prompting and LLM Ensembling for Multimodal Conversational Aspect-based Sentiment Analysis	提出结构化提示与LLM集成方法，用于多模态对话场景下的细粒度情感分析。	large language model multimodal
3	Fragile Knowledge, Robust Instruction-Following: The Width Pruning Dichotomy in Llama-3.2	Llama-3.2宽度剪枝揭示：参数知识退化，指令跟随能力增强	instruction following
4	Conformal Prediction Sets for Next-Token Prediction in Large Language Models: Balancing Coverage Guarantees with Set Efficiency	提出VACP框架，在LLM的Next-Token预测中平衡覆盖率保证与集合效率。	large language model
5	Exploring the Vertical-Domain Reasoning Capabilities of Large Language Models	探索大型语言模型在垂直领域（会计）的推理能力，为企业数字化转型提供基准。	large language model
6	Hallucination Detection and Evaluation of Large Language Model	提出HHEM框架，高效检测大语言模型幻觉，并结合分段检索提升摘要任务性能。	large language model
7	Syntactic Framing Fragility: An Audit of Robustness in LLM Ethical Decisions	提出句法框架脆弱性(SFF)评估框架，揭示LLM在伦理决策中对句法变异的敏感性。	large language model chain-of-thought
8	Beg to Differ: Understanding Reasoning-Answer Misalignment Across Languages	揭示多语言大模型推理与答案错位问题，提出跨语言推理评估框架	large language model chain-of-thought
9	Topic Segmentation Using Generative Language Models	提出基于生成式语言模型的篇章分割方法，利用重叠递归提示策略提升分割效果。	large language model
10	SagaScale: A Realistic, Scalable, and High-Quality Long-Context Benchmark Built from Full-Length Novels	SagaScale：基于完整小说的真实、可扩展、高质量长文本基准	large language model
11	Learning When Not to Attend Globally	提出All-or-Here Attention，使LLM动态决定何时关注全局上下文以提升效率。	large language model
12	Mitigating Social Desirability Bias in Random Silicon Sampling	通过心理学引导提示减少LLM中的社会期望偏差	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	🔗	⭐
13	Evaluating GRPO and DPO for Faithful Chain-of-Thought Reasoning in LLMs	评估GRPO和DPO在提升LLM中思维链推理忠实度的能力	DPO direct preference optimization large language model
14	ADMEDTAGGER: an annotation framework for distillation of expert knowledge for the Polish medical language	ADMEDTAGGER：利用多语言LLM蒸馏波兰语医学文本标注知识	distillation large language model

⬅️ 返回 cs.CL 首页 · 🏠 返回主页