CAT Merging: A Training-Free Approach for Resolving Conflicts in Model Merging

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Sun, Wenju, Li, Qingyong, Geng, Yangli-ao, Li, Boyang
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866916736826081280
author Sun, Wenju
Li, Qingyong
Geng, Yangli-ao
Li, Boyang
author_facet Sun, Wenju
Li, Qingyong
Geng, Yangli-ao
Li, Boyang
contents Multi-task model merging offers a promising paradigm for integrating multiple expert models into a unified model without additional training. Existing state-of-the-art techniques, such as Task Arithmetic and its variants, merge models by accumulating task vectors -- the parameter differences between pretrained and finetuned models. However, task vector accumulation is often hindered by knowledge conflicts, leading to performance degradation. To address this challenge, we propose Conflict-Aware Task Merging (CAT Merging), a novel training-free framework that selectively trims conflict-prone components from the task vectors. CAT Merging introduces several parameter-specific strategies, including projection for linear weights and masking for scaling and shifting parameters in normalization layers. Extensive experiments on vision, language, and vision-language tasks demonstrate that CAT Merging effectively suppresses knowledge conflicts, achieving average accuracy improvements of up to 2.5% (ViT-B/32) and 2.0% (ViT-L/14) over state-of-the-art methods.
format Preprint
id arxiv_https___arxiv_org_abs_2505_06977
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle CAT Merging: A Training-Free Approach for Resolving Conflicts in Model Merging
Sun, Wenju
Li, Qingyong
Geng, Yangli-ao
Li, Boyang
Artificial Intelligence
Machine Learning
Multi-task model merging offers a promising paradigm for integrating multiple expert models into a unified model without additional training. Existing state-of-the-art techniques, such as Task Arithmetic and its variants, merge models by accumulating task vectors -- the parameter differences between pretrained and finetuned models. However, task vector accumulation is often hindered by knowledge conflicts, leading to performance degradation. To address this challenge, we propose Conflict-Aware Task Merging (CAT Merging), a novel training-free framework that selectively trims conflict-prone components from the task vectors. CAT Merging introduces several parameter-specific strategies, including projection for linear weights and masking for scaling and shifting parameters in normalization layers. Extensive experiments on vision, language, and vision-language tasks demonstrate that CAT Merging effectively suppresses knowledge conflicts, achieving average accuracy improvements of up to 2.5% (ViT-B/32) and 2.0% (ViT-L/14) over state-of-the-art methods.
title CAT Merging: A Training-Free Approach for Resolving Conflicts in Model Merging
topic Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2505.06977