Custom inference engine to run Minimax-M2.X series of models on dual RTX Pro 6000 (sm120) - View it on GitHub