Use GPU Model Service

Brief introduction

The Xiaomi Cloud-ML Model Service also supports GPU-based services and is used the same way as the training tasks. See Using GPU Training.

Use GPU Model Service

GPU resources can be used by adding GPU parameters when the user creates a model service.

cloudml models create -n linear -v v1 -u fds://cloud-ml/linear -c 1 -M 1G -g 1

Parameter introduction

• -GThis parameter indicates the number of GPUs used. The platform will select a resource-sufficient machine for scheduling. In addition, the startup GPU container image is also different from the CPU container image.