AI RESEARCH
AutoAdapt: An Automated Domain Adaptation Framework for LLMs
arXiv CS.LG
•
ArXi:2603.08181v1 Announce Type: new Large language models (LLMs) excel in open domains but struggle in specialized settings with limited data and evolving knowledge. Existing domain adaptation practices rely heavily on manual trial-and-error processes, incur significant hyperparameter complexity, and are highly sensitive to data and user preferences, all under the high cost of LLM