AI RESEARCH
Boosting Large Language Models with Mask Fine-Tuning
arXiv CS.AI
•
ArXi:2503.22764v2 Announce Type: replace-cross The large language model (LLM) is typically integrated into the mainstream optimization protocol. No work has questioned whether maintaining the model integrity is \textit{indispensable} for promising performance. In this work, we