IMU-1: Sample-Efficient Pre-training of Small Language Models Paper • 2602.02522 • Published about 1 month ago • 6