[short] Simple linear attention language models balance the recall-throughput tradeoff

Arxiv Papers - [short] Simple linear attention language models balance the recall-throughput tradeoff