This repository contains the PyTorch implementation of the paper:

Null-Space Filtering for Data-Free Continual Model Merging: Preserving Stability, Promoting Plasticity (ICLR 2026)

Abstract

Data-free continual model merging (DFCMM) aims to fuse independently fine-tuned models into a single backbone that evolves with incoming tasks without accessing task data. This paper revisits two fundamental desiderata for DFCMM: stability, avoiding interference with earlier tasks, and plasticity, adapting faithfully to each new task. This poses a challenge that existing approaches fail to address: how to bridge data-level desiderata with parameter-space optimization to ensure stability and plasticity in the absence of task data. To this end, we propose NUFILT (Null-space Filtering), a data-free framework that directly links these desiderata into parameter-space optimization. Our key observation is that task vectors approximately align with representation subspaces, providing structural surrogates for enforcing stability and plasticity. Accordingly, we design a null-space projector that preserves prior responses by filtering overlapping components of new task vectors, ensuring stability. We further introduce a lightweight LoRA adapter that injects complementary task-specific signals to enable plasticity. The adapter is trained with a projection-based surrogate loss that preserves consistency with prior knowledge while introducing novel directions. This joint filtering-adaptation process enables the backbone to absorb new knowledge while retaining existing behaviors, with updates fused back in a layer-wise linear fashion without extra parameters or inference cost. Theoretically, we establish approximate subspace alignment guarantees that justify null-space filtering. Empirically, NUFILT achieves state-of-the-art performance with minimal forgetting on both vision and NLP benchmarks, improving average accuracy by 4-7% over OPCM and WUDI-Merging, while narrowing the gap to fine-tuning and reducing computation overhead.

Overview

Overview of the NUFILT procedure. 1) Filtering: the new task vector is processed through a null-space projector that suppresses activations from previous tasks, ensuring stability to past knowledge. 2) Adapting: within the filter, a lightweight LoRA adapter refines the update for the current task using a data-free objective. 3) Fusing: the filter, task vector, and LoRA module are merged back into the backbone, keeping the parameter count and inference cost unchanged.

Introduction to DFCMM

Data-Free Continual Model Merging (DFCMM) aims to continually absorb new task knowledge into a shared backbone without revisiting any task data.

• Stability: preserve the behaviors learned from previous tasks and avoid destructive interference during merging.
• Plasticity: faithfully incorporate the knowledge carried by each incoming task vector.
• Challenge: without task data, stability and plasticity must be enforced directly in parameter space rather than through data-driven objectives.

NUFILT addresses this challenge with two complementary components:

• Null-space filtering removes overlapping components of new task vectors in representation-aligned subspaces, helping preserve prior responses.
• LoRA-based adaptation injects complementary task-specific signals through a lightweight adapter trained with a projection-based surrogate loss.

Together, these components enable continual model merging that remains stable on earlier tasks while staying plastic enough to acquire new knowledge.

Installation

install the latest version in development

pip install -e . # install the package in editable mode

Project Structure

The project is structured as follows:

• fusion_bench/: the main package of the benchmark.
  • method: contains the implementation of the fusion methods.

    naming convention: fusion_bench/method/{method_name}/{variant}.py contains the implementation of the specific method or its variants. For example, fusion_bench/method/regmean/clip_regmean.py contains the implementation of the RegMean algorithm for CLIP vision models.

  • modelpool: contains the implementation of the model pool, responsible for managing the models and dataset to be loaded.
  • taskpool: contains the implementation of the task pool, responsible for evaluating the performance of models returned by the algorithm.
• config/: configuration files for the benchmark. We use Hydra to manage the configurations.
  • method: configuration files for the fusion methods.

    naming convention: config/method/{method_name}/{variant}.yaml contains the configuration for the specific method or its variants.

  • modelpool: configuration files for the model pool.
  • taskpool: configuration files for the task pool.
  • model: configuration files for the models.
  • dataset: configuration files for the datasets.
• examples/: example scripts for running some of the experiments.

  naming convention: examples/{method_name}/ contains the files such as bash scripts and jupyter notebooks for the specific method.

How to run the experiments

We provide bash scripts to reproduce the results in the paper.
All scripts are located in the examples/nufilt folder.

Reproducing Tables

• bash examples/nufilt/baseline.sh

• bash examples/nufilt/nufilt.sh

• bash examples/nufilt/t5_base.sh

Acknowledgements

This project is based on FusionBench. We thank the authors for their valuable contribution.

In this repository, we only keep the essential code required to reproduce the results in the NUFILT paper.
For the full benchmark codebase and additional functionalities, please refer to FusionBench.

Citation

If you find our work useful, please consider citing:

@inproceedings{
qiu2026nullspace,
title={Null-Space Filtering for Data-free Continual Model Merging: Preserving Stability, Promoting Plasticity},
author={Zihuan Qiu and Lei Wang and Yang Cao and Runtong ZHANG and Bing Su and Yi Xu and Fanman Meng and Linfeng Xu and Qingbo Wu and Hongliang Li},
booktitle={The Fourteenth International Conference on Learning Representations},
year={2026},
url={https://openreview.net/forum?id=HDIf3fYqPP}
}