Tag
The Laya model runs offline on Apple's M4 chip using CoreML, achieving 45 decisions per second in inference.