Model merging is now widely used to enhance large language model performance, with merged models frequently ranking high on the Open LLM Leaderboard. The process typically involves merging a set of models by downloading their checkpoints and using a merge algorithm, which can be memory-intensive but is made manageable by the mergekit library. However, there's a need to merge different "adapters" obtained from the same model, such as LoRA checkpoints, which presents unique challenges due to varying merging requirements and the need for seamless integration. To address this, new merging methods have been developed targeting popular LoRA adapters in 🤗 PEFT, offering code examples and promising results for developers to explore.