Evolutionary Optimization of Model Merging Recipes

Takuya Akiba, Makoto Shing, Yujin Tang +2

Large language models (LLMs) have become increasingly capable, but their development often requires substantial computational resources. While model merging has emerged as a cost-e…