3 ms·How Minimax-01 Achieves 1M Token Context Length with Linear Attention (MIT)2 points by research_pie 2y ago