cs.CL（2025-03-06）

📊 共 52 篇论文 | 🔗 9 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (40 🔗5) 支柱二：RL算法与架构 (RL & Architecture) (11 🔗3) 支柱一：机器人控制 (Robot Control) (1 🔗1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (40 篇)

#题目一句话要点标签🔗⭐
1 Keeping Yourself is Important in Downstream Tuning Multimodal Large Language Model 系统性评测多模态大语言模型微调策略，解决任务专精和知识遗忘问题。 large language model multimodal ✅
2 Knowledge-Decoupled Synergetic Learning: An MLLM based Collaborative Approach to Few-shot Multimodal Dialogue Intention Recognition 提出知识解耦协同学习(KDSL)，解决电商场景下少样本多模态对话意图识别难题。 large language model multimodal
3 Benchmarking Large Language Models on Multiple Tasks in Bioinformatics NLP with Prompting Bio-benchmark：基于Prompt的大语言模型在生物信息学NLP多任务上的综合评测框架 large language model chain-of-thought
4 Large Language Models in Bioinformatics: A Survey 综述：大型语言模型赋能生物信息学，推动基因组、RNA、蛋白质和单细胞数据分析 large language model multimodal
5 TRACT: Regression-Aware Fine-tuning Meets Chain-of-Thought Reasoning for LLM-as-a-Judge TRACT：融合回归感知微调与思维链推理，提升LLM作为裁判的性能 large language model chain-of-thought
6 HILGEN: Hierarchically-Informed Data Generation for Biomedical NER Using Knowledgebases and Large Language Models HILGEN：结合知识库和LLM的分层生物医学命名实体识别数据生成方法 large language model
7 Are Large Language Models Good In-context Learners for Financial Sentiment Analysis? 探索大语言模型在金融情感分析中作为上下文学习器的能力 large language model
8 Codebook Reduction and Saturation: Novel observations on Inductive Thematic Saturation for Large Language Models and initial coding in Thematic Analysis 提出基于DSPy的归纳主题饱和度测量方法，用于评估LLM在主题分析中的表现 large language model
9 Uncovering inequalities in new knowledge learning by large language models across different languages 揭示大语言模型跨语言新知识学习中的不平等现象，关注低资源语言的劣势。 large language model
10 Biases in Large Language Model-Elicited Text: A Case Study in Natural Language Inference 揭示大型语言模型生成文本中的偏见：自然语言推理案例研究 large language model
11 IFIR: A Comprehensive Benchmark for Evaluating Instruction-Following in Expert-Domain Information Retrieval 提出IFIR：一个综合性的专家领域信息检索指令跟随能力评测基准。 instruction following
12 Can Large Language Models Predict Antimicrobial Resistance Gene? 利用大型语言模型预测抗生素耐药基因，探索DNA序列分析新范式 large language model ✅
13 TableLoRA: Low-rank Adaptation on Table Structure Understanding for Large Language Models TableLoRA：面向大语言模型的表格结构理解低秩自适应方法 large language model
14 Shaping Shared Languages: Human and Large Language Models' Inductive Biases in Emergent Communication 通过人与大语言模型交互，探索涌现语言中归纳偏置的影响，促进人机对齐。 large language model
15 Dynamic-KGQA: A Scalable Framework for Generating Adaptive Question Answering Datasets Dynamic-KGQA：一种可扩展的自适应问答数据集生成框架 large language model foundation model
16 Chart-HQA: A Benchmark for Hypothetical Question Answering in Charts 提出Chart-HQA基准，用于评估多模态大语言模型在图表中的假设性问题回答能力。 large language model multimodal
17 Measuring temporal effects of agent knowledge by date-controlled tool use 提出基于日期控制工具的LLM Agent知识时效性评估方法 large language model chain-of-thought
18 Biological Sequence with Language Model Prompting: A Survey 综述：基于语言模型提示的生物序列分析与应用 large language model multimodal
19 Beyond RAG: Task-Aware KV Cache Compression for Comprehensive Knowledge Reasoning 提出任务感知KV缓存压缩方法，提升LLM在知识推理任务中的效率与准确性 large language model
20 Factorio Learning Environment 提出Factorio学习环境FLE，用于评估LLM在长期规划、程序合成和资源优化方面的能力。 large language model
21 Enough Coin Flips Can Make LLMs Act Bayesian 通过足够多的抛硬币示例，可使大语言模型表现出贝叶斯行为 large language model
22 HybridNorm: Towards Stable and Efficient Transformer Training via Hybrid Normalization 提出HybridNorm混合归一化方法，提升Transformer训练稳定性和效率 large language model ✅
23 Shifting Long-Context LLMs Research from Input to Output 呼吁NLP研究从长文本输入转向长文本输出，填补LLM在长文本生成方面的能力空白。 large language model
24 Solving Word-Sense Disambiguation and Word-Sense Induction with Dictionary Examples 利用字典示例和LLM解决低资源语言的词义消歧和词义归纳问题 large language model
25 M2S: Multi-turn to Single-turn jailbreak in Red Teaming for LLMs 提出M2S框架，将多轮对抗性提示转化为单轮提示，提升LLM红队测试效率。 large language model
26 Temporal Alignment of LLMs through Cycle Encoding for Long-Range Time Representations Ticktack：通过周期编码对LLM进行时间对齐，解决长时程时间表示问题 large language model
27 Disparities in LLM Reasoning Accuracy and Explanations: A Case Study on African American English 揭示LLM在非洲裔美国英语推理任务中的准确性和解释性差异 large language model ✅
28 ReasonGraph: Visualisation of Reasoning Paths ReasonGraph：用于可视化和分析LLM推理路径的Web平台 large language model
29 Safety is Not Only About Refusal: Reasoning-Enhanced Fine-tuning for Interpretable LLM Safety 提出Rational框架，通过推理增强微调提升LLM安全性和可解释性 large language model
30 DP-GTR: Differentially Private Prompt Protection via Group Text Rewriting DP-GTR：通过分组文本重写实现差分隐私提示保护 large language model
31 DB-Explore: Automated Database Exploration and Instruction Synthesis for Text-to-SQL DB-Explore：通过数据库探索与指令合成增强Text-to-SQL模型的数据库理解能力 large language model
32 UIPE: Enhancing LLM Unlearning by Removing Knowledge Related to Forgetting Targets UIPE：通过移除相关知识增强LLM的遗忘学习 large language model
33 LLM-guided Plan and Retrieval: A Strategic Alignment for Interpretable User Satisfaction Estimation in Dialogue 提出PRAISE框架，利用LLM指导的规划与检索，提升对话系统中用户满意度估计的可解释性与准确性。 large language model
34 START: Self-taught Reasoner with Tools START：一种自学习工具增强推理器，提升复杂推理任务性能 chain-of-thought
35 Better Process Supervision with Bi-directional Rewarding Signals 提出BiRM双向奖励模型，提升LLM推理过程监督的准确性和有效性 large language model
36 Guiding LLMs to Generate High-Fidelity and High-Quality Counterfactual Explanations for Text Classification 提出分类器引导的LLM方法，无需微调即可生成高质量文本分类对抗样本 large language model
37 More Documents, Same Length: Isolating the Challenge of Multiple Documents in RAG 研究揭示RAG中多文档数量对LLM性能的负面影响，并发现Qwen2.5具有更强的多文档处理能力。 large language model ✅
38 Layer-Specific Scaling of Positional Encodings for Superior Long-Context Modeling 提出层特异性位置编码缩放方法，缓解长文本建模中的“中间信息丢失”问题。 large language model
39 In-depth Analysis of Graph-based RAG in a Unified Framework 统一框架下对图结构RAG方法进行深入分析，并发现新的更优变体。 large language model
40 LLMs Can Generate a Better Answer by Aggregating Their Own Responses 提出生成式自聚合（GSA）方法，提升LLM在复杂任务中的答案质量。 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (11 篇)

#题目一句话要点标签🔗⭐
41 Mark Your LLM: Detecting the Misuse of Open-Source Large Language Models via Watermarking 提出基于水印技术的开源大语言模型滥用检测方法，应对IP侵犯和违规使用。 distillation large language model
42 DiffPO: Diffusion-styled Preference Optimization for Efficient Inference-Time Alignment of Large Language Models 提出DiffPO，通过扩散模型优化LLM推理时对齐，提升效率与效果 RLHF large language model
43 Implicit Cross-Lingual Rewarding for Efficient Multilingual Preference Alignment 提出隐式跨语言奖励方法，高效实现多语言偏好对齐。 DPO direct preference optimization large language model ✅
44 FuseChat-3.0: Preference Optimization Meets Heterogeneous Model Fusion FuseChat-3.0：融合异构大模型偏好优化，提升紧凑型目标模型性能 DPO direct preference optimization large language model ✅
45 L1: Controlling How Long A Reasoning Model Thinks With Reinforcement Learning 提出LCPO，通过强化学习控制推理语言模型的思考长度，实现计算成本与准确率的平衡。 reinforcement learning chain-of-thought ✅
46 TinyR1-32B-Preview: Boosting Accuracy with Branch-Merge Distillation 提出Branch-Merge蒸馏方法，提升小模型在数学、编程和科学等领域的性能。 distillation large language model
47 Balcony: A Lightweight Approach to Dynamic Inference of Generative Language Models Balcony：一种轻量级的生成语言模型动态推理方法，提升推理效率。 distillation large language model
48 An Information-theoretic Multi-task Representation Learning Framework for Natural Language Understanding 提出InfoMTL框架，通过信息论方法提升多任务学习中的自然语言理解能力。 representation learning
49 Memory Is All You Need: Testing How Model Memory Affects LLM Performance in Annotation Tasks 利用模型记忆提升LLM在文本标注任务中的性能 reinforcement learning large language model
50 Towards Data-Efficient Language Models: A Child-Inspired Approach to Language Learning 提出一种受儿童语言学习启发的数据高效语言模型训练方法 curriculum learning large language model
51 Revisiting the Othello World Model Hypothesis 通过扩展实验，为语言模型具备黑白棋世界模型能力提供更强证据 world model

🔬 支柱一：机器人控制 (Robot Control) (1 篇)

#题目一句话要点标签🔗⭐
52 An Empirical Study on Eliciting and Improving R1-like Reasoning Models 通过强化学习和工具使用提升类R1推理模型的性能 manipulation ✅

⬅️ 返回 cs.CL 首页 · 🏠 返回主页