cs.CL（2025-08-23）

📊 共 14 篇论文 | 🔗 5 篇有代码

🎯 兴趣领域导航

#	题目	一句话要点	标签	🔗
1	Unbiased Reasoning for Knowledge-Intensive Tasks in Large Language Models via Conditional Front-Door Adjustment	提出条件前门调整框架以解决大语言模型的偏见问题	large language model chain-of-thought
2	Quantifying Language Disparities in Multilingual Large Language Models	提出框架以量化多语言大模型中的语言差异	large language model
3	Planning for Success: Exploring LLM Long-term Planning Capabilities in Table Understanding	利用大型语言模型提升表格理解能力以解决复杂问题	large language model chain-of-thought
4	Token Homogenization under Positional Bias	研究标记均质化与位置偏差的关系	large language model
5	Linguistic Neuron Overlap Patterns to Facilitate Cross-lingual Transfer on Low-resource Languages	提出BridgeX-ICL以解决低资源语言跨语言学习问题	large language model	✅
6	GRAID: Synthetic Data Generation with Geometric Constraints and Multi-Agentic Reflection for Harmful Content Detection	提出GRAID以解决有害内容检测中的数据稀缺问题	large language model
7	ObjexMT: Objective Extraction and Metacognitive Calibration for LLM-as-a-Judge under Multi-Turn Jailbreaks	提出ObjexMT以解决多轮对话中的目标提取与元认知校准问题	large language model	✅
8	QFrCoLA: a Quebec-French Corpus of Linguistic Acceptability Judgments	提出QFrCoLA数据集以评估法语语言模型的语言判断能力	large language model

#	题目	一句话要点	标签	🔗
9	DeAR: Dual-Stage Document Reranking with Reasoning Agents via LLM Distillation	提出DeAR以解决文档重排序中的推理与评分平衡问题	distillation large language model chain-of-thought	✅
10	KL-Regularised Q-Learning: A Token-level Action-Value perspective on Online RLHF	提出KL正则化Q学习以优化语言模型的强化学习	reinforcement learning PPO RLHF
11	Decoding Alignment: A Critical Survey of LLM Development Initiatives through Value-setting and Data-centric Lens	通过价值设定与数据中心视角审视大型语言模型的对齐问题	reinforcement learning RLHF large language model
12	Being Kind Isn't Always Being Safe: Diagnosing Affective Hallucination in LLMs	提出AHaBench与AHaPairs以解决大语言模型的情感幻觉问题	DPO direct preference optimization large language model	✅
13	Dream to Chat: Model-based Reinforcement Learning on Dialogues with User Belief Modeling	提出对话世界模型以解决用户信念建模问题	reinforcement learning world model
14	Learning from Diverse Reasoning Paths with Routing and Collaboration	提出QR-Distill以解决知识蒸馏中的路径质量问题	distillation large language model	✅