Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Provide example for InferenceGraph Ensemble voting #405

Open
asd981256 opened this issue Oct 15, 2024 · 0 comments · May be fixed by #415
Open

Provide example for InferenceGraph Ensemble voting #405

asd981256 opened this issue Oct 15, 2024 · 0 comments · May be fixed by #415

Comments

@asd981256
Copy link

Describe the change you'd like to see
A clear and concise description of what you want to happen.

We are deploying ensemble learning model with InferenceGraph Ensemble node, and would like to provide our python code and yaml file for document example.

This example would be helpful for understanding how to deploy another node in InferenceGraph and how ensemble node gathers parallel isvc outputs.

And, it can be adapted by anyone who has similar model output format, which is a list of probabilities of each label, to use our code directly for convenience. (v1 for now)

For example, will provide yaml and python code similar as following after we finsih our work and change our classifiers to more simple ones.

apiVersion: serving.kserve.io/v1beta1
kind: InferenceService
metadata:
  name: avg-vote
spec:
  predictor:
    containers:
      - name: avg-vote
        image: {avg-vote-image}
        args:
          - --model_name=avg-vote
---
apiVersion: serving.kserve.io/v1beta1
kind: InferenceService
metadata:
  name: dummy-classifier-1
spec:
  predictor:
    containers:
      - name: dummy-classifier-1
        image: {dummy-classifier-1-image}
        args:
          - --model_name=dummy-classifier-1
---
apiVersion: serving.kserve.io/v1beta1
kind: InferenceService
metadata:
  name: dummy-classifier-2
spec:
  predictor:
    containers:
      - name: dummy-classifier-2
        image: {dummy-classifier-2-image}
        args:
          - --model_name=dummy-classifier-2
apiVersion: "serving.kserve.io/v1alpha1"
kind: "InferenceGraph"
metadata:
  name: "ensemble-2-avg-vote"
spec:
  nodes:
    root:
      routerType: Sequence
      steps:
        - nodeName: ensemble-2
          name: ensemble-2
        - serviceName: avg-vote
          name: avg-vote
          data: $response
    ensemble-2:
      routerType: Ensemble
      steps:
        - serviceName: dummy-classifier-1
          name: dummy-classifier-1
        - serviceName: dummy-classifier-2
          name: dummy-classifier-2
import argparse
from typing import Dict, Union
import numpy as np
from kserve import (
    Model,
    ModelServer,
    model_server,
    InferRequest,
    InferOutput,
    InferResponse,
    logging,
)
from kserve.errors import InvalidInput
from kserve.model_server import app
from kserve.utils.utils import get_predict_response

class AvgVoting(Model):
    def __init__(self, name: str):
       super().__init__(name)
       self.model = None
       self.ready = False
       self.load()

    def load(self):
        self.ready = True

    def predict(self, payload: Union[Dict, InferRequest], headers: Dict[str, str] = None) -> Union[Dict, InferResponse]:
        # assume payload as ensemble node output: {"sklearn-iris":{"predictions":[1,1]},"xgboost-iris":{"predictions":[1,1]}}
        tmp = []
        for isvcName, output in payload.items():
          prediction = output['predictions']
          tmp.append(prediction)

        result = [sum(x)/len(tmp) for x in zip(*tmp)] # assume same number of label
        return get_predict_response(payload, result, self.name)

parser = argparse.ArgumentParser(parents=[model_server.parser])
args, _ = parser.parse_known_args()

if __name__ == "__main__":
    if args.configure_logging:
       logging.configure_logging(args.log_config_file)

    model = AvgVoting(args.model_name)
    app.add_middleware(
        CORSMiddleware,
        allow_origins=["*"],
        allow_credentials=True,
        allow_methods=["*"],
        allow_headers=["*"],
    )
    ModelServer().start([model])

Additional context
Add any other context or screenshots about the feature request here.
Should be related to #288
any suggestion or telling somebody already doing this that i didnt notice would be nice.

@asd981256 asd981256 linked a pull request Oct 31, 2024 that will close this issue
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging a pull request may close this issue.

1 participant