CPU and GPU Model Training and Deployment

CPU Model Training and Deployment

If the model is to be deployed on a CPU device, it is recommended to select Lite (better with CPU deployment) .

../../../_images/instance_cpu_training.png

Select CPU for Deployment device.

../../../_images/instance_cpu_deployment.png

GPU Model Training and Deployment

If the model is to be deployed on a GPU device, it is recommended to select Normal (better with GPU deployment) .

../../../_images/instance_gpu_training.png

There are two model types to choose from.

  • dlkmo: The model optimization takes about 5 minutes. This model type is for the situation where the GPU graphics card models are different for model training and model deployment.

  • dlkmt: No optimization is required for this type. This model type is for the situation where the GPU graphics card models are the same for model training and model deployment.

../../../_images/instance_gpu_deployment.png