SOLO: Pretraining Billion-Parameter Language Models with Shared-Output Local Learning Paper • 2609.35440 • Published Sep 28 • 1