Adapter Tuning

Follow this topic Ask about this topic

Adapter tuning studies how to specialize pretrained language, vision, speech, and multimodal models by training small added modules while leaving most or all of the backbone frozen. Common approaches include bottleneck adapters, low-rank updates such as LoRA, prompts, and input- or feature-space modules; researchers also vary where adapters are placed, how their rank is allocated, and whether experts are routed by task or input. These methods are evaluated on settings such as language understanding and reasoning, few-shot vision, speech recognition, retrieval, and continual learning, with attention to trainable parameters, accuracy, inference cost, and forgetting. The central tension is how to give adapters enough capacity and flexibility to match full fine-tuning while keeping them efficient, composable, and reliable across tasks and domains.

Latest papers

76 papers3 papers in the last four weeks, down 57% on the four weeks before. 0.0% of all new papers.

Introspection Adapters: Training LLMs to Report Their Learned Behaviors
Apr 18, 2026Keshav Shenoy, Li Yang, Abhay Sheshadri +4LLM Auditing Language Model Introspection
Anthropic Fellows · Ashwood Centre on AI Science and Policy, University of Cambridge · Anthropic
What and Where to Adapt: Structure-Semantics Co-Tuning for Machine Vision Compression via Synergistic Adapters
Apr 11, 2026Shaobo Liu, Haobo Xiong, Kai Liu +1Entropy Modeling Learned Image Compression
Xidian University
Learning to Synergize Semantic and Geometric Priors for Limited-Data Wheat Disease Segmentation
Apr 7, 2026Shijie Wang, Zijian Wang, Yadan Luo +3Image Segmentation Promptable Image Segmentation
The University of Queensland, Australia · The University of Adelaide, Australia
Lightweight Prompt-Guided CLIP Adaptation for Monocular Depth Estimation
Apr 1, 2026Reyhaneh Ahani Manghotay, Jie LiangPEFT of VLMs Monocular Depth Estimation
Simon Fraser University, Burnaby, Canada · School of Engineering Science, Simon Fraser University, Burnaby, BC, Canada · Eastern Institute of Technology, Ningbo, China +1
HeBA: Heterogeneous Bottleneck Adapters for Robust Vision-Language Models
Mar 17, 2026Md Jahidul IslamVLM Adaptation Adapter Tuning
Department of Electrical and Electronic Engineering, Bangladesh University of Engineering and Technology, Dhaka, Bangladesh
AdapterTune: Zero-Initialized Low-Rank Adapters for Frozen Vision Transformers
Mar 16, 2026Salim KhazemAdapter Tuning Low-Rank Adaptation
Talan Research Center, Paris, France
SK-Adapter: Skeleton-Based Structural Control for Native 3D Generation
Mar 14, 2026Anbang Wang, Yuzhuo Ao, Shangzhe Wu +13D Foundation Models 3D Asset Generation
The Hong Kong University of Science and Technology · University of Cambridge
A Simple Efficiency Incremental Learning Framework via Vision-Language Model with Nonlinear Multi-Adapters
Mar 11, 2026Haihua Luo, Xuming Ran, Jiangrong Shen +4Incremental Learning Continual Learning for VLMs
Faculty of Information Technology, University of Jyväskylä, Jyväskylä, Finland · Equal contribution · National University of Singapore, Singapore, Singapore +2
Multi-Kernel Gated Decoder Adapters for Robust Multi-Task Thyroid Ultrasound under Cross-Center Shift
Mar 9, 2026Maziar Sabouri, Nourhan Bayasi, Arman RahmimThyroid Nodule Detection Ultrasound Image Segmentation
Department of Astronomy and Physics, University of British Columbia, Vancouver, Canada · Department of Basic and Translational Research, BC Cancer Research Institute, Vancouver, Canada · Department of Radiology, University of British Columbia, Vancouver, Canada +1
Spectral Surgery: Training-Free Refinement of LoRA via Gradient-Guided Singular Value Reweighting
Mar 4, 2026Zailong Tian, Yanzhe Chen, Zhuoheng Han +1Low-Rank Adaptation Adapter Tuning
School of Computing and Information Systems, Singapore Management University · School of Computing, National University of Singapore · State Key Laboratory for Multimedia Information Processing, Peking University
CeRA: Breaking the Linear Ceiling of Low-Rank Adaptation with Non-linearity Retained at Inference
Feb 26, 2026Hung-Hsuan ChenNonlinear LoRA Adapter Tuning
Computer Science and Information Engineering National Central University Taoyuan, Taiwan
SCOPE: Structured Prototype-Guided Adaptation for EEG Foundation Models with Limited Labels
Feb 19, 2026Jingying Ma, Feng Wu, Yucheng Xing +5EEG Foundation Models Few-Shot Domain Adaptation
Saw Swee Hock School of Public Health, National University of Singapore, Singapore · Guangzhou Research Translation and Innovation Institute, National University of Singapore, Guangzhou, China · College of Computing and Data Science, Nanyang Technological University, Singapore +3
Improving the Performance of Compressed Transformers with Post-Hoc Embedding Alignment
Feb 16, 2026Rohit Raj Rai, Abhishek Dhaka, Amit AwekarAdapter Tuning Representation Alignment
Indian Institute of Technology Guwahati, Guwahati, Assam, India · B.K. Birla Institute of Engineering & Technology, Pilani, Rajasthan, India
GRASP: Guided Residual Adapters with Sample-wise Partitioning
Dec 1, 2025Felix Nützel, Mischa Dombrowski, Bernhard KainzImage Generation Adapter Tuning
Friedrich-Alexander-Universität Erlangen-Nürnberg, GER · Imperial College London, UK
MetaTT: A Global Tensor-Train Adapter for Parameter-Efficient Fine-Tuning
Jun 10, 2025Javier Lopez-Piqueres, Pranav Deshpande, Archan Ray +3Transformer Tensor Decomposition
Global Technology Applied Research, JPMorgan Chase, New York, NY 10001, USA
Forecast-PEFT: Parameter-Efficient Fine-Tuning for Pre-trained Motion Forecasting Models
Jul 28, 2024Jifeng Wang, Kaouther Messaoud, Yuejiang Liu +2Parameter-Efficient Fine-Tuning Adapter Tuning
University of Bonn, Bonn, Germany · VITA Laboratory, EPFL, Lausanne, Switzerland · Stanford University, Stanford, USA

Apr 18, 2026J/K move · Enter open · S save

Introspection Adapters: Training LLMs to Report Their Learned Behaviors

Keshav Shenoy, Li Yang, Abhay Sheshadri, Sören Mindermann+3

Anthropic Fellows · Ashwood Centre on AI Science and Policy, University of Cambridge · Anthropic

LLM Auditing Language Model Introspection Language Model Self-Assessment Adapter Tuning Language Model Probing

Open paper AskPDF arXiv

When model developers or users fine-tune an LLM, this can induce behaviors that are unexpected, deliberately harmful, or hard to detect. It would be far easier to audit LLMs if they could simply describe their behaviors in natural language. Here, we study a scalable approach to rapidly identify learned behaviors of many LLMs derived from a shared base LLM. Given a model

M

, our method works by finetuning models

M_i

from

M

with implanted behaviors

b_i

; the

(M_i, b_i)

pairs serve as labeled training data. We then train an introspection adapter (IA): a single LoRA adapter jointly trained across the finetunes

M_i

to cause them to verbalize their implanted behaviors. We find that this IA induces self-description of learned behaviors even in finetunes of

M

that were trained in very different ways from the

M_i

. For example, IAs generalize to AuditBench, achieving state-of-the-art at identifying explicitly hidden concerning behaviors. IAs can also be used to detect encrypted finetuning API attacks. They scale favorably with model size and training data diversity. Overall, our results suggest that IAs are a scalable, effective, and practically useful approach to auditing fine-tuned LLMs.