Griffin: Mixing Gated Linear Recurrences with Local Attention for Efficient Language Models

Open in new window