GLAM: Efficient Continual Learning at Scale via Grouped LoRA Adapter Merging Paper • 2509.13211 • Published 27 days ago • 5