Model Assembly Learning with Heterogeneous Layer Weight Merging
This addresses the problem of efficiently combining pre-trained models for AI practitioners, offering a more flexible approach than previous methods that required identical architectures.
The paper tackles the problem of model merging by introducing Model Assembly Learning (MAL), a novel paradigm that iteratively integrates parameters from diverse models in a model zoo to enhance a base model's capabilities, allowing merging of heterogeneous architectures and selective parameters across layers.
Model merging acquires general capabilities without extra data or training by combining multiple models' parameters. Previous approaches achieve linear mode connectivity by aligning parameters into the same loss basin using permutation invariance. In this paper, we introduce Model Assembly Learning (MAL), a novel paradigm for model merging that iteratively integrates parameters from diverse models in an open-ended model zoo to enhance the base model's capabilities. Unlike previous works that require identical architectures, MAL allows the merging of heterogeneous architectures and selective parameters across layers. Specifically, the base model can incorporate parameters from different layers of multiple pre-trained models. We systematically investigate the conditions and fundamental settings of heterogeneous parameter merging, addressing all possible mismatches in layer widths between the base and target models. Furthermore, we establish key laws and provide practical guidelines for effectively implementing MAL.