Can I deploy PyTorch trained models on Seldon?

284 views Asked by At

I have tried deploying PyTorch model on seldon by using TRITON server. with yaml

apiVersion: machinelearning.seldon.io/v1
kind: SeldonDeployment
metadata:
  name: cifar10
spec:
  annotations:
    seldon.io/engine-seldon-log-messages-externally: "true"
  name: resnet32
  predictors:
  - graph:
      implementation: TRITON_SERVER
      logger:
        mode: all
      modelUri: s3://<Bucket-name>/sample-model
      serviceAccountName: <service-account-name>
      name: cifar10
      type: MODEL
    name: default
    replicas: 1
  protocol: kfserving

accesing the path /v2/models/cifar10, it shows error

{"error":"Request for unknown model: 'cifar10' is not found"}
0

There are 0 answers