我正在尝试将输入变压器与模型一起使用。以下是部署 yaml:
apiVersion: machinelearning.seldon.io/v1
kind: SeldonDeployment
metadata:
name: two-testing
spec:
predictors:
- componentSpecs:
- spec:
containers:
- name: transformer
image: transformer_image
ports:
- containerPort: 7100
name: http
protocol: TCP
- name: model
image: model_image
ports:
- containerPort: 7200
name: http
protocol: TCP
graph:
name: transformer
type: TRANSFORMER
children:
- name: model
type: MODEL
children: []
endpoint:
service_port: 7300
name: model
replicas: 1
我发送请求的端点 api 是: 'http://0.0.0.0:3000/api/v1.0/predictions' , 3000 是我转发端口但请求直接命中模型容器的本地端口因此我得到一个错误。有人可以告诉我应该将请求发送到哪个端点 api,以便模型通过输入转换器然后进入模型。