Griffin: Mixing Gated Linear Recurrences with Local Attention for Efficient Language Models Paper • 2402.19427 • Published Feb 29 • 52 • 5