While model merging offers a way to combine different skills, it is often impractical due to mismatched neural architectures and the closed-source nature of top-performing models.
大多数人认为模型合并是整合不同AI模型能力的可行方法，但作者明确指出这种方法在实践中存在根本性限制，挑战了行业对模型合并解决方案的普遍信任。