3 ms·MiniMax M2 Tech Blog 3: Why Did M2 End Up as a Full Attention Model?1 points by logicprog 11mo ago