ai exp1 Simplify the alexnet model through the int8 quantization method to reduce the image recognition time on the GPU exp2 Simplify the yolo v3 model by pruning to reduce the pattern recognition time on the GPU