RADLADS: Rapid Attention Distillation to Linear Attention Decoders at Scale Paper • 2505.03005 • Published May 5, 2025 • 36
