语言大模型 / LLM

VALUE: Value-Aware Large Language Model for Query Rewriting via Weighted Trie in Sponsored Search

VALUE: 用于通过加权前缀树进行查询重写的值感知大语言模型

Xiao Zhang, Guanyu Chen, Boyang Zuo, Feng Li, Pengjie Wang, Jian Xu, Bo Zheng

机构 * Alibaba Group（阿里巴巴集团） ； Tsinghua University（清华大学） 纠错

专题命中 指令微调 ：large language model(title,abstract)；language model(title,abstract)；LLM(abstract)；SFT(abstract)

AI总结 VALUE通过加权前缀树整合价值感知，提升搜索引擎广告中查询重写的语义相关性和商业价值

SGuard-v1: Safety Guardrail for Large Language Models

JoonHo Lee, HyeonMin Cho, Jaewoong Yun, Hyunjae Lee, JunKyu Lee, Juree Seok

专题命中 指令微调 ：large language model(title,abstract)；language model(title,abstract)；LLM(abstract)；instruction tuning(abstract)

Comments Technical Report

When Long Helps Short: How Context Length in Supervised Fine-tuning Affects Behavior of Large Language Models

Yingming Zheng, Hanqi Li, Kai Yu, Lu Chen

机构 * X-LANCE Lab（X-LANCE实验室） ； MoE Key Lab of Artificial Intelligence（人工智能MoE关键实验室） ； AI Institute（人工智能研究院） ； School of Computer Science（计算机科学学院） ； Shanghai Jiao Tong University（上海交通大学） ； Shanghai Innovation Institution（上海创新机构） ； Jiangsu Key Lab of Language Computing（江苏语言计算关键实验室） ； Suzhou Laboratory（苏州实验室） 纠错

专题命中 指令微调 ：large language model(title,abstract)；language model(title,abstract)；LLM(abstract)；pretraining(abstract)

UpSafe$^\circ$C: Upcycling for Controllable Safety in Large Language Models

Yuhao Sun, Zhuoer Xu, Shiwen Cui, Kun Yang, Lingyun Yu, Yongdong Zhang, Hongtao Xie

专题命中 指令微调 ：large language model(title,abstract)；language model(title,abstract)；LLM(abstract)；post-training(abstract)

Investigating ReLoRA: Effects on the Learning Dynamics of Small Language Models

Yuval Weiss, David Demitri Africa, Paula Buttery, Richard Diehl Martinez

机构 * University of Cambridge（剑桥大学） 纠错

专题命中 指令微调 ：language model(title,abstract)；small language model(title,abstract)；LLM(abstract)；large language model(abstract)

Comments 12 Pages, 6 Tables, 8 Figures

The Thinking Therapist: Training Large Language Models to Deliver Acceptance and Commitment Therapy using Supervised Fine-Tuning and Odds Ratio Policy Optimization

Talha Tahir

机构 * Department of Psychiatry University of Toronto（精神病学系多伦多大学） 纠错

专题命中 指令微调 ：large language model(title,abstract)；language model(title,abstract)；LLM(abstract)；post-training(abstract)

FEVO: Financial Knowledge Expansion and Reasoning Evolution for Large Language Models

Bo Pang, Yalu Ouyang, Hangfei Xu, Ziqi Jia, Panpan Li, Shengzhao Wen, Lu Wang, Shiyong Li, Yanpeng Wang

机构 * AI Cloud Group, Baidu（百度AI云团队） ； Tsinghua University（清华大学） ； University of California, San Diego（加州大学圣地亚哥分校） 纠错

专题命中 指令微调 ：large language model(title,abstract)；language model(title,abstract)；LLM(abstract)；post-training(abstract)

RoSTE: An Efficient Quantization-Aware Supervised Fine-Tuning Approach for Large Language Models

Quan Wei, Chung-Yiu Yau, Hoi-To Wai, Yang Katie Zhao, Dongyeop Kang, Youngsuk Park, Mingyi Hong

专题命中 指令微调 ：large language model(title,abstract)；language model(title,abstract)；LLM(abstract)；post-training(abstract)

Comments accepted by ICML 2025

SyntheT2C: Generating Synthetic Data for Fine-Tuning Large Language Models on the Text2Cypher Task

Ziije Zhong, Linqing Zhong, Zhaoze Sun, Qingyun Jin, Zengchang Qin, Xiaofan Zhang

专题命中 指令微调 ：large language model(title,abstract)；language model(title,abstract)；LLM(abstract)；SFT(abstract)

Comments COLING 2025 paper. 21 pages, 15 figures, 8 tables

Unlock the Correlation between Supervised Fine-Tuning and Reinforcement Learning in Training Code Large Language Models

Jie Chen, Xintian Han, Yu Ma, Xun Zhou, Liang Xiang

专题命中 指令微调 ：large language model(title,abstract)；language model(title,abstract)；LLM(abstract)；pretraining(abstract)

AssistRAG: Boosting the Potential of Large Language Models with an Intelligent Information Assistant

Yujia Zhou, Zheng Liu, Zhicheng Dou

专题命中 指令微调 ：large language model(title,abstract)；language model(title,abstract)；LLM(abstract)；SFT(abstract)

Comments Accepted by NeurIPS 2024 (poster)

LEAF: Learning and Evaluation Augmented by Fact-Checking to Improve Factualness in Large Language Models

Hieu Tran, Junda Wang, Yujan Ting, Weijing Huang, Terrence Chen

专题命中 指令微调 ：large language model(title,abstract)；language model(title,abstract)；LLM(abstract)；SFT(abstract)

Comments 22 pages, 9 figures

Predicting the Big Five Personality Traits in Chinese Counselling Dialogues Using Large Language Models

Yang Yan, Lizhi Ma, Anqi Li, Jingsong Ma, Zhenzhong Lan

专题命中 指令微调 ：large language model(title,abstract)；language model(title,abstract)；LLM(abstract)；preference optimization(abstract)

Preserving Knowledge in Large Language Model with Model-Agnostic Self-Decompression

Zilun Zhang, Yutao Sun, Tiancheng Zhao, Leigang Sha, Ruochen Xu, Kyusong Lee, Jianwei Yin

专题命中 指令微调 ：large language model(title,abstract)；language model(title,abstract)；LLM(abstract)；instruction tuning(abstract)

Paying More Attention to Source Context: Mitigating Unfaithful Translations from Large Language Model

Hongbin Zhang, Kehai Chen, Xuefeng Bai, Yang Xiang, Min Zhang

专题命中 指令微调 ：large language model(title,abstract)；language model(title,abstract)；LLM(abstract)；instruction tuning(abstract)

Comments Accepted by ACL2024 Findings

Psychometric Predictive Power of Large Language Models

Tatsuki Kuribayashi, Yohei Oseki, Timothy Baldwin

专题命中 指令微调 ：large language model(title,abstract)；language model(title,abstract)；LLM(abstract)；instruction tuning(abstract)

Comments 23 pages; Findings of NAACL 2024

Aligning Modalities in Vision Large Language Models via Preference Fine-tuning

Yiyang Zhou, Chenhang Cui, Rafael Rafailov, Chelsea Finn, Huaxiu Yao

专题命中 指令微调 ：large language model(title,abstract)；language model(title,abstract)；LLM(abstract)；RLHF(abstract)

FP8-LM: Training FP8 Large Language Models

Houwen Peng, Kan Wu, Yixuan Wei, Guoshuai Zhao, Yuxiang Yang, Ze Liu, Yifan Xiong, Ziyue Yang, Bolin Ni, Jingcheng Hu, Ruihang Li, Miaosen Zhang, Chen Li, Jia Ning, Ruizhe Wang, Zheng Zhang, Shuguang Liu, Joe Chau, Han Hu, Peng Cheng

专题命中 指令微调 ：large language model(title,abstract)；language model(title,abstract)；LLM(abstract)；foundation model(abstract)

Are Large Language Models Really Robust to Word-Level Perturbations?

Haoyu Wang, Guozheng Ma, Cong Yu, Ning Gui, Linrui Zhang, Zhiqi Huang, Suwei Ma, Yongzhe Chang, Sen Zhang, Li Shen, Xueqian Wang, Peilin Zhao, Dacheng Tao

专题命中 指令微调 ：large language model(title,abstract)；language model(title,abstract)；LLM(abstract)；SFT(abstract)

KeyPooling: Measuring Where LLM API Relay Paths Collapse Prompt Cache Isolation

KeyPooling：测量LLM API中继路径在提示缓存隔离中的崩溃位置

Bowen Sun, Yixi Cai, Xiaogeng Liu, Zhengyue Zhao, Yinzhi Cao, Chaowei Xiao

专题命中 指令微调 ：LLM(title,title_cn)；large language model(abstract)；language model(abstract)

AI总结 该研究提出KeyPooling测量方法，发现LLM API中继服务默认未将客户绑定到上游凭证，存在跨客户缓存读取漏洞，并提出防御契约及优化方案。

Refine After Generation: Toward Correct and Concise Patches in LLM-based Program Repair

生成后优化：面向基于大语言模型的程序修复中正确且简洁的补丁

Wenqiang Luo, Jacky Keung, Xiaoyu Shi, Yicheng Sun, Boyang Yang, Zhou Yang, Haoye Tian

专题命中 指令微调 ：LLM(title,summary_cn)；large language model(abstract)；language model(abstract)；preference optimization(abstract)

AI总结 针对基于LLM的程序修复中补丁冗余问题，提出RECAP适配器，在保留或提升解决率的同时，大幅降低补丁大小，实现更优的大小-正确性权衡。

GradAlign: Gradient-Aligned Data Selection for LLM Reinforcement Learning

GradAlign: 用于大语言模型强化学习的梯度对齐数据选择

Ningyuan Yang, Weihua Du, Weiwei Sun, Sean Welleck, Yiming Yang

机构 * Institute for Interdisciplinary Information Sciences (IIIS), Tsinghua University（交叉信息学院（IIIS）、清华大学） ； Language Technologies Institute (LTI), Carnegie Mellon University（语言技术研究所（LTI）、卡内基梅隆大学） 纠错

专题命中 指令微调 ：LLM(title,abstract)；SFT(abstract,abstract_cn)；large language model(abstract)；language model(abstract)

AI总结 GradAlign通过梯度对齐数据选择方法提升大语言模型强化学习的训练稳定性与性能。

Comments 20 pages. Accepted by COLM 2026

LLM-Empowered Multimodal Fusion Framework for Autonomous Driving: Semantic Enhancement and Channel-Adaptive Design

LLM赋能的自动驾驶多模态融合框架：语义增强与信道自适应设计

Wen Wang, Yaping Sun, Yejun He, Hao Chen, Zhiyong Chen, Xiaodong Xu, Nan Ma, Shuguang Cui

机构 * National Natural Science Foundation of China（国家自然科学基金） ； Shenzhen Natural Science Foundation（深圳市自然科学基金） ； Shenzhen Key Laboratory Evaluation（深圳市重点实验室评估） 纠错

专题命中 指令微调 ：LLM(title,title_cn)；large language model(abstract)；language model(abstract)

AI总结 提出LM-SCIP框架，以LLM为核心推理中枢，通过信道自适应语义模块动态融合视觉与雷达特征，实现低信噪比下的视觉主导回退和高信噪比下的协同融合，在nuScenes和VIRAT数据集上显著提升定位精度。

Comments 6 pages, 4 figures. Accepted by 2026 IEEE 37th International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC)

OASIF: An Efficient Obfuscation-Aware Self-Improving Framework for LLM-Based Assembly Code Instruction Following and Comprehension

OASIF：一种高效的混淆感知自改进框架，用于基于LLM的汇编代码指令遵循与理解

Xinyi Wang, Rongze Chen, Ke Wang, Qiyuan Chen, Yanming Liu, Xiang Li, Chunfu Jia

专题命中 指令微调 ：LLM(title,title_cn)；large language model(abstract)；language model(abstract)

AI总结 提出OASIF框架，通过令牌高效汇编编码器、轻量投影器和三阶段训练（特征空间对齐、监督指令微调、在线自进化强化学习），提升LLM在商业级混淆汇编代码上的指令遵循与理解能力，在VMISA-Bench上取得显著成功率提升。

LLM Agents as Static Level-k Players in Behavioural Games

LLM智能体作为行为博弈中的静态k级玩家

Po Han Teo

专题命中 指令微调 ：LLM(title,title_cn)；large language model(abstract)；language model(abstract)

AI总结 通过p-beauty竞赛和公共物品博弈实验，发现LLM在不同部署设置下可部分复现人类选择的分散性，但无法模拟人类的策略过程，表现为静态的、按模型规模检索类别的k级玩家，缺乏信念更新和逆向归纳能力。

Comments 30 pages, 3 figures, 6 tables

Utility-Diversity Aware Online Batch Selection for LLM Supervised Fine-tuning

面向LLM监督微调的效用-多样性感知在线批次选择

Heming Zou, Yixiu Mao, Yun Qu, Qi Wang, Xiangyang Ji

机构 * University of Science and Technology of China（中国科学技术大学） 纠错

专题命中 指令微调 ：LLM(title,title_cn)；SFT(abstract,abstract_cn)；large language model(abstract)；language model(abstract)

AI总结 提出UDS框架，利用logits矩阵核范数和轻量记忆缓冲实现高效在线批次选择，兼顾数据效用与多样性，无需外部资源，在多个基准上优于现有方法并降低训练时间。

Comments ICML 2026 accepted paper

DV-SFT: Direct Vision Supervision for Fine-Grained Visual Understanding

DV-SFT: 直接视觉监督用于细粒度视觉理解

Jianfei Zhao, Feng Zhang, Xin Sun, Chong Feng, Bing Wang, Zhixing Tan

机构 * School of Computer Science and Technology, Beijing Institute of Technology（北京理工大学计算机科学与技术学院） ； Zhongguancun Academy（中关村学院） ； Beihang University（北航） ； Zhongguancun Laboratory（中关村实验室） ； Southeast Academy of Information Technology, Beijing Institute of Technology（北京理工大学信息科学技术东南学院） 纠错

专题命中 指令微调 ：SFT(title,title_cn)；large language model(abstract)；language model(abstract)

AI总结 提出DV-SFT方法，通过为视觉令牌构建显式令牌级监督信号，利用OCR场景中的直接视觉-文本对应关系，在不修改模型架构或增加前向传播的情况下，显著提升多模态大语言模型的细粒度视觉理解能力。

Comments Under Review

Boosting Automatic Java-to-Cangjie Translation with Multi-Stage LLM Training and Error Repair

通过多阶段LLM训练和错误修复提升Java到Cangjie的自动翻译

Xinyue Liang, Jingxuan Zhang, Lin Li, Jun Zhang, Junhao Chen

专题命中 指令微调 ：LLM(title,title_cn)；large language model(abstract)；language model(abstract)

AI总结 本文提出多阶段LLM训练框架，结合编译器反馈和错误修复，提升Java到Cangjie的翻译准确性与语义一致性，实验显示在有限平行数据下功能等价性提升6.06%。

SafeTune: Mitigating Data Poisoning in LLM Fine-Tuning for RTL Code Generation

SafeTune: 缓解LLM微调中RTL代码生成的数据污染

Mahshid Rezakhani, Nowfel Mashnoor, Kimia Azar, Hadi Kamali

专题命中 指令微调 ：LLM(title,title_cn)；large language model(abstract)；language model(abstract)

AI总结 SafeTune通过结合图神经网络和语义验证模块，有效过滤数据污染，提升LLM微调的鲁棒性和可靠性，不改变底层模型架构。

Comments This paper will be presented at IEEE VLSI Test Symposium (VTS) 2026. 5 pages, 2 figures, 4 tables

Three Birds, One Stone: Solving the Communication-Memory-Privacy Trilemma in LLM Fine-tuning Over Wireless Networks with Zeroth-Order Optimization

三鸟一石：在无线网络上通过零阶优化解决LLM微调中的通信-内存-隐私三重困境

Zhijie Cai, Yuhao Zheng, Haolong Chen, Dongzhu Liu, Bin Wang, Guangxu Zhu

专题命中 指令微调 ：LLM(title,title_cn)；large language model(abstract)；language model(abstract)

AI总结 本文提出pAirZero框架，结合零阶优化与空中计算，解决LLM微调中的通信、内存和隐私三重问题，实现低开销的安全高效微调。