AI RESEARCH
PEFT-Arena: Understanding Parameter-Efficient Finetuning from a Stability-Plasticity Perspective
arXiv CS.LG
•
ArXi:2605.28819v1 Announce Type: new Parameter-efficient finetuning (PEFT) has become the standard approach for adapting large language models, yet evaluations largely emphasize downstream accuracy while overlooking the retention of pretrained capabilities. We argue that PEFT should be assessed through the stability-plasticity dilemma: the trade-off between target-task adaptation and resistance to forgetting. We