语言大模型 / LLM

Self-Evolved Diverse Data Sampling for Efficient Instruction Tuning

Shengguang Wu, Keming Lu, Benfeng Xu, Junyang Lin, Qi Su, Chang Zhou

专题命中 指令微调 ：instruction tuning(title,abstract)；large language model(abstract)；language model(abstract)；分类 cs.CL、cs.LG

LLM as an Art Director (LaDi): Using LLMs to improve Text-to-Media Generators

Allen Roush, Emil Zakirov, Artemiy Shirokov, Polina Lunina, Jack Gane, Alexander Duffy, Charlie Basil, Aber Whitcomb, Jim Benedetto, Chris DeWolfe

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

Comments 12 pages, System Demonstration/Industry paper. Preprint

Active Instruction Tuning: Improving Cross-Task Generalization by Training on Prompt Sensitive Tasks

Po-Nien Kung, Fan Yin, Di Wu, Kai-Wei Chang, Nanyun Peng

专题命中 指令微调 ：instruction tuning(title,abstract)；large language model(abstract)；language model(abstract)；分类 cs.CL、cs.AI

Comments EMNLP 2023 Main

Roles of Scaling and Instruction Tuning in Language Perception: Model vs. Human Attention

Changjiang Gao, Shujian Huang, Jixing Li, Jiajun Chen

专题命中 指令微调 ：instruction tuning(title,abstract)；large language model(abstract)；language model(abstract)；分类 cs.CL、cs.AI

InstructAlign: High-and-Low Resource Language Alignment via Continual Crosslingual Instruction Tuning

Samuel Cahyawijaya, Holy Lovenia, Tiezheng Yu, Willy Chung, Pascale Fung

专题命中 指令微调 ：instruction tuning(title,abstract)；large language model(abstract)；language model(abstract)；分类 cs.CL、cs.AI

Correction Focused Language Model Training for Speech Recognition

Yingyi Ma, Zhe Liu, Ozlem Kalinli

专题命中 指令微调 ：language model(title,abstract)；LLM(abstract)；large language model(abstract)；分类 cs.CL、cs.LG

AstroLLaMA: Towards Specialized Foundation Models in Astronomy

Tuan Dung Nguyen, Yuan-Sen Ting, Ioana Ciucă, Charlie O'Neill, Ze-Chang Sun, Maja Jabłońska, Sandor Kruk, Ernest Perkowski, Jack Miller, Jason Li, Josh Peek, Kartheik Iyer, Tomasz Różański, Pranav Khetarpal, Sharaf Zaman, David Brodrick, Sergio J. Rodríguez Méndez, Thang Bui, Alyssa Goodman, Alberto Accomazzi, Jill Naiman, Jesse Cranney, Kevin Schawinski, UniverseTBD

专题命中 指令微调 ：foundation model(title,abstract)；large language model(abstract)；language model(abstract)；分类 cs.CL、cs.LG

Comments 6 pages, 3 figures, submitted to IJCNLP-AACL 2023. Comments are welcome. The model can be found on Hugging Face - https://huggingface.co/universeTBD/astrollama

Adversarial Fine-Tuning of Language Models: An Iterative Optimisation Approach for the Generation and Detection of Problematic Content

Charles O'Neill, Jack Miller, Ioana Ciuca, Yuan-Sen Ting, Thang Bui

专题命中 指令微调 ：language model(title,abstract)；large language model(abstract)；prompting(abstract)；分类 cs.CL、cs.LG

Language models are weak learners

Hariharan Manikandan, Yiding Jiang, J Zico Kolter

专题命中 指令微调 ：language model(title,abstract)；LLM(abstract)；large language model(abstract)；分类 cs.AI、cs.LG

Comments 23 pages, 6 figures

Revisiting Automated Prompting: Are We Actually Doing Better?

Yulin Zhou, Yiren Zhao, Ilia Shumailov, Robert Mullins, Yarin Gal

专题命中 指令微调 ：prompting(title,abstract)；large language model(abstract)；language model(abstract)；分类 cs.CL、cs.LG

MIMIC-IT: Multi-Modal In-Context Instruction Tuning

Bo Li, Yuanhan Zhang, Liangyu Chen, Jinghao Wang, Fanyi Pu, Jingkang Yang, Chunyuan Li, Ziwei Liu

专题命中 指令微调 ：instruction tuning(title,abstract)；large language model(abstract)；language model(abstract)；分类 cs.CL、cs.AI

Comments Project page: https://otter-ntu.github.io/ Dataset & code: https://github.com/Luodian/otter Initial release, work in progress

Text-to-Audio Generation using Instruction-Tuned LLM and Latent Diffusion Model

Deepanway Ghosal, Navonil Majumder, Ambuj Mehrish, Soujanya Poria

专题命中 指令微调 ：LLM(title,abstract)；large language model(abstract)；language model(abstract)；分类 cs.CL、cs.AI

Comments https://github.com/declare-lab/tango

Language Model Self-improvement by Reinforcement Learning Contemplation

Jing-Cheng Pang, Pengyuan Wang, Kaiyuan Li, Xiong-Hui Chen, Jiacheng Xu, Zongzhang Zhang, Yang Yu

专题命中 指令微调 ：language model(title,abstract)；LLM(abstract)；large language model(abstract)；分类 cs.CL、cs.LG

Self-Agreement: A Framework for Fine-tuning Language Models to Find Agreement among Diverse Opinions

Shiyao Ding, Takayuki Ito

专题命中 指令微调 ：language model(title,abstract)；LLM(abstract)；large language model(abstract)；分类 cs.CL、cs.AI

InstructUIE: Multi-task Instruction Tuning for Unified Information Extraction

Xiao Wang, Weikang Zhou, Can Zu, Han Xia, Tianze Chen, Yuansen Zhang, Rui Zheng, Junjie Ye, Qi Zhang, Tao Gui, Jihua Kang, Jingsheng Yang, Siyuan Li, Chunsai Du

专题命中 指令微调 ：instruction tuning(title,abstract)；large language model(abstract)；language model(abstract)；分类 cs.CL、cs.AI

Instruction Tuning with GPT-4

Baolin Peng, Chunyuan Li, Pengcheng He, Michel Galley, Jianfeng Gao

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

Comments 8 pages. Work in progress. Project page: https://instruction-tuning-with-gpt-4.github.io

Bidirectional Language Models Are Also Few-shot Learners

Ajay Patel, Bryan Li, Mohammad Sadegh Rasooli, Noah Constant, Colin Raffel, Chris Callison-Burch

专题命中 指令微调 ：language model(title,abstract)；large language model(abstract)；prompting(abstract)；分类 cs.CL、cs.LG

Comments To appear at ICLR 2023

Fine-tuning language models to find agreement among humans with diverse preferences

Michiel A. Bakker, Martin J. Chadwick, Hannah R. Sheahan, Michael Henry Tessler, Lucy Campbell-Gillingham, Jan Balaguer, Nat McAleese, Amelia Glaese, John Aslanides, Matthew M. Botvinick, Christopher Summerfield

专题命中 指令微调 ：language model(title,abstract)；LLM(abstract)；large language model(abstract)；分类 cs.CL、cs.LG

Is One Token All It Takes? Graph Pooling Tokens for LLM-based GraphQA

一个token足够吗？基于LLM的图QA的图池化token

Ankit Grover, Lodovico Giaretta, Rémi Bourgerie, Sarunas Girdzijauskas

专题命中 指令微调 ：LLM(title,abstract)；large language model(abstract)；language model(abstract)；分类 cs.LG

AI总结 本文提出通过多token池化和全局注意力机制改进图与LLM的接口，通过实验表明LoRA可稳定特定层次投影，使压缩表示在WebQSP上达到73%的Hit@1，揭示图QA基准存在表示饱和问题。

Comments Accepted at LREC, KG-LLM Workshop 2026

When Fine-Tuning Fails and when it Generalises: Role of Data Diversity and Mixed Training in LLM-based TTS

当微调失效且泛化时：数据多样性与混合训练在基于大语言模型的文本到语音系统中的作用

Anupam Purwar, Aditya Choudhary

机构 * Sprinklr AI 纠错

专题命中 指令微调 ：LLM(title,abstract)；large language model(abstract)；language model(abstract)；分类 cs.AI

AI总结 本文研究了LoRA微调在语音克隆任务中的有效性，发现其在语音质量和信噪比提升方面优于基模型，尤其在数据多样性支持下表现更佳。

Comments We finetune the Qwen 0.5B backbone in an LLM TTS with LoRA to raise MOS speaker similarity and SNR. It works best with diverse training audio with uniform data it can amplify noise so tune decoding and use GGUF quantization for low latency stable quality

On the Difficulty of Selecting Few-Shot Examples for Effective LLM-based Vulnerability Detection

在有效基于LLM的漏洞检测中选择少样本示例的难度

Md Abdul Hannan, Ronghao Ni, Chi Zhang, Limin Jia, Ravi Mangal, Corina S. Pasareanu

机构 * Colorado State University（科罗拉多州立大学） ； Carnegie Mellon University（卡内基梅隆大学） 纠错

专题命中 指令微调 ：LLM(title,abstract)；large language model(abstract)；language model(abstract)；分类 cs.LG

AI总结 本文研究了在基于LLM的漏洞检测中选择少样本示例的两种标准，并发现不同编程语言对示例选择的响应不同。

Comments Workshop on LLM Assisted Security and Trust Exploration (LAST-X) 2026

Fine-Tuned Language Models Generate Stable Inorganic Materials as Text

Nate Gruver, Anuroop Sriram, Andrea Madotto, Andrew Gordon Wilson, C. Lawrence Zitnick, Zachary Ulissi

机构 * NYU（纽约大学） ； Meta FAIR 纠错

专题命中 指令微调 ：language model(title,abstract)；large language model(abstract)；prompting(abstract)；分类 cs.LG

Comments ICLR 2024. Code available at: https://github.com/facebookresearch/crystal-llm

Efficient Telecom Specific LLM: TSLAM-Mini with QLoRA and Digital Twin Data

Vignesh Ethiraj, Divya Vijay, Sidhanth Menon, Heblin Berscilla

机构 * NetoAI Solutions Ltd（NetoAI解决方案有限公司） 纠错

专题命中 指令微调 ：LLM(title,abstract)；language model(abstract,comments)；large language model(abstract)；分类 cs.AI

Comments Introducing TSLAM-Mini, a specialized language model for telecommunications, demonstrating the efficacy of QLoRA fine-tuning and digital twin-synthesized data for enhanced network intelligence. Model available on: https://huggingface.co/NetoAISolutions/TSLAM-Mini-2B

Enhancing Code Translation in Language Models with Few-Shot Learning via Retrieval-Augmented Generation

Manish Bhattarai, Javier E. Santos, Shawn Jones, Ayan Biswas, Boian Alexandrov, Daniel O'Malley

专题命中 指令微调 ：language model(title,abstract)；LLM(abstract,comments)；large language model(abstract)；分类 cs.AI

Comments LLM for code translation

Vision-Flan: Scaling Human-Labeled Tasks in Visual Instruction Tuning

Zhiyang Xu, Chao Feng, Rulin Shao, Trevor Ashby, Ying Shen, Di Jin, Yu Cheng, Qifan Wang, Lifu Huang

专题命中 指令微调 ：instruction tuning(title,abstract)；language model(abstract)；pretraining(abstract)；分类 cs.CL

Comments 8 Pages, visual instruction tuning

Constitutional Midtraining: Content Presence Drives Alignment Gains

宪法式中间训练：内容存在驱动对齐增益

Desiree Cho, Cameron Tice, Bernie Hogan, Hunar Batra, Puria Radmard, Jun Zhao, Nigel Shadbolt

机构 * University of Oxford（牛津大学） ； Institute for Ethics in AI, University of Oxford（牛津大学人工智能伦理研究所） ； Geodesic Research ； Oxford Internet Institute, University of Oxford（牛津大学互联网研究院） 纠错

专题命中 指令微调 ：SFT(summary_cn,abstract)；post-training(abstract)；分类 cs.CL、cs.AI、cs.LG

AI总结 该研究提出宪法式中间训练方法，在中间训练中插入基于原则价值观的内容，可提升模型对齐的泛化性与持久性，削弱SFT灌输的敲诈倾向且不造成能力损失，为以SFT为中心的流程提供低成本补充方案。

Benchmarking LLMs on File System Design and Implementation

在文件系统设计与实现上对大语言模型（LLMs）进行基准测试：优势、不足与缺陷

Yuqi Xue, Daixuan Li, Jian Huang

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

AI总结 该研究提出针对文件系统特定任务的LLM基准框架phi-Bench，含505项六类任务，测试开源与专有LLMs，揭示模型效率、失败原因及缓解技术，将开源phi-Bench。

Mitigating Visual Degradation in MLLMs via Spatial-Spectral Visual Anchor Learning

通过空间-光谱视觉锚学习缓解多模态大语言模型（MLLMs）中的视觉退化

Qianlong Yang, Bowen Ye, Xianda Guo, Yanlun Peng, Wenke Huang, Hongyuan Zhang, Yulei Jia

机构 * China University of Petroleum (East China)（中国石油大学（华东）） ； Shanghai Jiao Tong University（上海交通大学） ； Wuhan University（武汉大学） ； Great Wall Motor（长城汽车） ； Nanyang Technological University（南洋理工大学） ； The University of Hong Kong（香港大学） 纠错

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

AI总结 针对MLLMs推理时的视觉表示退化问题，提出SSVAL方法，通过VAPI及辅助对齐损失实现稳定视觉锚，性能优于现有方法。

Comments This paper has been accepted by ACM MM 2026

Building Large-Scale English-Romanian Literary Translation Resources with Open Models

构建大规模英语-罗马尼亚文学翻译资源的开放模型

Mihai Nadas, Laura Diosan, Andreea Tomescu, Andrei Piscoran

机构 * Babeș-Bolyai University（巴克斯-波耶_ai大学） ； KlusAI Labs（KlusAI实验室） 纠错

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

AI总结 本文提出TF2框架，通过开放模型和大规模合成语料库构建英语-罗马尼亚文学翻译资源，实现高效、低成本的高质量翻译。

Comments 21 pages. Published version: Front. Artif. Intell. 9:1807431 (2026). Datasets and models released on Hugging Face

Journal ref Front. Artif. Intell. 9:1807431 (2026)

Parameter Efficient Multimodal Instruction Tuning for Romanian Vision Language Models

罗马尼亚视觉语言模型的参数高效多模态指令微调

George-Andrei Dima, Răzvan-Alexandru Smădu, Dumitru-Clementin Cercel

机构 * National University of Science and Technology（科学技术大学） 纠错

专题命中 指令微调 ：language model(title)；instruction tuning(title)；分类 cs.CL、cs.AI、cs.LG

AI总结 针对罗马尼亚语这种低资源语言，通过翻译并扩展Flickr30K数据集，采用参数高效的LoRA方法微调开源视觉语言模型来降低多模态NLP资源差距，模型在视觉问答等任务中能力提升且语法错误减少。