Imagen de Portada

Towards Minimizing Feature Drift in Model Merging: Layer-wise Task Vector Fusion for Adaptive Knowledge Integration

Fuente: arXiv

Guardado en:

Detalles Bibliográficos
Autores principales:	Sun, Wenju, Li, Qingyong, Wang, Wen, Liu, Yang, Geng, Yangli-ao, Li, Boyang
Formato:	Preprint
Publicado:	2025
Materias:	Machine Learning Artificial Intelligence
Acceso en línea:	Acceder al recurso
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

Ejemplares similares

Task Arithmetic in Trust Region: A Training-Free Model Merging Approach to Navigate Knowledge Conflicts
por: Sun, Wenju, et al.
Publicado: (2025)

CAT Merging: A Training-Free Approach for Resolving Conflicts in Model Merging
por: Sun, Wenju, et al.
Publicado: (2025)

Towards Plastic and Stable Exemplar-Free Incremental Learning: A Dual-Learner Framework with Cumulative Parameter Averaging
por: Sun, Wenju, et al.
Publicado: (2023)

Exemplar-free Class Incremental Learning via Discriminative and Comparable One-class Classifiers
por: Sun, Wenju, et al.
Publicado: (2022)

Extending Precipitation Nowcasting Horizons via Spectral Fusion of Radar Observations and Foundation Model Priors
por: Qin, Yuze, et al.
Publicado: (2026)

AdaLTM: Adaptive Layer-wise Task Vector Merging for Categorical Speech Emotion Recognition with ASR Knowledge Integration
por: Lee, Chia-Yu, et al.
Publicado: (2026)

Scalable Model Merging with Progressive Layer-wise Distillation
por: Xu, Jing, et al.
Publicado: (2025)

Layer-wise Model Merging for Unsupervised Domain Adaptation in Segmentation Tasks
por: Alcover-Couso, Roberto, et al.
Publicado: (2024)

Purifying Task Vectors in Knowledge-Aware Subspace for Model Merging
por: An, Bang, et al.
Publicado: (2025)

Rethinking Layer-wise Model Merging through Chain of Merges
por: Buzzega, Pietro, et al.
Publicado: (2025)

LGB: Language Model and Graph Neural Network-Driven Social Bot Detection
por: Zhou, Ming, et al.
Publicado: (2024)

AdaMerging: Adaptive Model Merging for Multi-Task Learning
por: Yang, Enneng, et al.
Publicado: (2023)

LARV: Data-Free Layer-wise Adaptive Rescaling Veneer for Model Merging
por: Wang, Xinyu, et al.
Publicado: (2026)

Enhancing Out-of-Distribution Detection with Multitesting-based Layer-wise Feature Fusion
por: Li, Jiawei, et al.
Publicado: (2024)

AdaTrans: Feature-wise and Sample-wise Adaptive Transfer Learning for High-dimensional Regression
por: He, Zelin, et al.
Publicado: (2024)

EMS: Adaptive Evict-then-Merge Strategy for Head-wise KV Cache Compression Based on Global-Local Importance
por: Li, Yingxin, et al.
Publicado: (2024)

Superpose Task-specific Features for Model Merging
por: Qiu, Haiquan, et al.
Publicado: (2025)

Harmonizing Diverse Models: A Layer-wise Merging Strategy for Consistent Generation
por: Peng, Xujun, et al.
Publicado: (2025)

SpecBound: Adaptive Bounded Self-Speculation with Layer-wise Confidence Calibration
por: Wen, Zhuofan, et al.
Publicado: (2026)

Task Singular Vectors: Reducing Task Interference in Model Merging
por: Gargiulo, Antonio Andrea, et al.
Publicado: (2024)

Task Vector Quantization for Memory-Efficient Model Merging
por: Kim, Youngeun, et al.
Publicado: (2025)

DLink: Distilling Layer-wise and Dominant Knowledge from EEG Foundation Models
por: Wang, Jingyuan, et al.
Publicado: (2026)

LayerPeeler: Autoregressive Peeling for Layer-wise Image Vectorization
por: Wu, Ronghuan, et al.
Publicado: (2025)

PM-MOE: Mixture of Experts on Private Model Parameters for Personalized Federated Learning
por: Feng, Yu, et al.
Publicado: (2025)

MergeNet: Knowledge Migration across Heterogeneous Models, Tasks, and Modalities
por: Li, Kunxi, et al.
Publicado: (2024)

Modeling Multi-Task Model Merging as Adaptive Projective Gradient Descent
por: Wei, Yongxian, et al.
Publicado: (2025)

Dataless Knowledge Fusion by Merging Weights of Language Models
por: Jin, Xisen, et al.
Publicado: (2022)

Tuning Layer Orbital Hall Effect via Spin Rotation in Ferromagnetic Transition Metal Dichalcogenides
por: Ji, Shilei, et al.
Publicado: (2025)

Merging without Forgetting: Continual Fusion of Task-Specific Models via Optimal Transport
por: Pan, Zecheng, et al.
Publicado: (2025)

Layer-wise Regularized Dropout for Neural Language Models
por: Ni, Shiwen, et al.
Publicado: (2024)

Segmentation-guided Layer-wise Image Vectorization with Gradient Fills
por: Zhou, Hengyu, et al.
Publicado: (2024)

Step-wise Adaptive Integration of Supervised Fine-tuning and Reinforcement Learning for Task-Specific LLMs
por: Chen, Jack, et al.
Publicado: (2025)

Universally Empowering Zeroth-Order Optimization via Adaptive Layer-wise Sampling
por: Wang, Fei, et al.
Publicado: (2026)

Channel-wise Vector Quantization
por: Song, Wei, et al.
Publicado: (2026)

FedLWS: Federated Learning with Adaptive Layer-wise Weight Shrinking
por: Shi, Changlong, et al.
Publicado: (2025)

Magnetic Anisotropy of Cr2Te3: Competition between Surface and Middle Layers
por: Yile Wang, et al.
Publicado: (2024)

Source‐Free Domain Adaptive Fundus Image Segmentation With Multiscale Feature Fusion and Stepwise Attention Integration
por: Mingtao Liu, et al.
Publicado: (2026)

Model Merging in the Essential Subspace
por: Li, Longhua, et al.
Publicado: (2026)

VecAttention: Vector-wise Sparse Attention for Accelerating Long Context Inference
por: Liu, Anmin, et al.
Publicado: (2026)

Integrating Intermediate Layer Optimization and Projected Gradient Descent for Solving Inverse Problems with Diffusion Models
por: Zheng, Yang, et al.
Publicado: (2025)