# Model Merging in LLMs, MLLMs, and Beyond: Methods, Theories, Applications and Opportunities

Model merging broadens the scope of machine learning.

Model merging in machine learning is gaining traction as a crucial technique that allows for the enhancement of models without direct access to raw data or substantial computational resources. However, observing a gap in literature, this survey contributes significantly by examining the current methods, theories, and practical applications of model merging, alongside prospective areas for research. Furthermore, it introduces a novel taxonomic structure while emphasizing both the challenges and prospects that lie ahead in the domain of model merging.

Read more: [
Research and Development

](https://arxiv.org/abs/2408.07666)