kserve/test/e2e/predictor/test_predictive.py

# Copyright 2025 The KServe Authors.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#    http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

import os

import numpy
import pytest
from kubernetes import client
from kubernetes.client import V1ResourceRequirements

from kserve import (
    KServeClient,
    V1beta1InferenceService,
    V1beta1InferenceServiceSpec,
    V1beta1ModelFormat,
    V1beta1ModelSpec,
    V1beta1PredictorSpec,
    constants,
)

from ..common.utils import KSERVE_TEST_NAMESPACE, predict_isvc


@pytest.mark.predictor
@pytest.mark.asyncio(scope="session")
async def test_predictive_sklearn_v1(rest_v1_client, network_layer):
    service_name = "isvc-predictive-sklearn"
    predictor = V1beta1PredictorSpec(
        min_replicas=1,
        model=V1beta1ModelSpec(
            model_format=V1beta1ModelFormat(name="sklearn"),
            runtime="kserve-predictiveserver",
            storage_uri="gs://kfserving-examples/models/sklearn/1.0/model",
            resources=V1ResourceRequirements(
                requests={"cpu": "50m", "memory": "128Mi"},
                limits={"cpu": "100m", "memory": "256Mi"},
            ),
        ),
    )

    isvc = V1beta1InferenceService(
        api_version=constants.KSERVE_V1BETA1,
        kind=constants.KSERVE_KIND_INFERENCESERVICE,
        metadata=client.V1ObjectMeta(
            name=service_name, namespace=KSERVE_TEST_NAMESPACE
        ),
        spec=V1beta1InferenceServiceSpec(predictor=predictor),
    )

    kserve_client = KServeClient(
        config_file=os.environ.get("KUBECONFIG", "~/.kube/config")
    )
    kserve_client.create(isvc)
    kserve_client.wait_isvc_ready(service_name, namespace=KSERVE_TEST_NAMESPACE)
    res = await predict_isvc(
        rest_v1_client,
        service_name,
        "./data/iris_input.json",
        network_layer=network_layer,
    )
    assert res["predictions"] == [1, 1]
    kserve_client.delete(service_name, KSERVE_TEST_NAMESPACE)


@pytest.mark.predictor
@pytest.mark.asyncio(scope="session")
async def test_predictive_xgboost_v1(rest_v1_client, network_layer):
    service_name = "isvc-predictive-xgboost"
    predictor = V1beta1PredictorSpec(
        min_replicas=1,
        model=V1beta1ModelSpec(
            model_format=V1beta1ModelFormat(name="xgboost"),
            runtime="kserve-predictiveserver",
            storage_uri="gs://kfserving-examples/models/xgboost/1.5/model",
            resources=V1ResourceRequirements(
                requests={"cpu": "50m", "memory": "128Mi"},
                limits={"cpu": "100m", "memory": "256Mi"},
            ),
        ),
    )

    isvc = V1beta1InferenceService(
        api_version=constants.KSERVE_V1BETA1,
        kind=constants.KSERVE_KIND_INFERENCESERVICE,
        metadata=client.V1ObjectMeta(
            name=service_name, namespace=KSERVE_TEST_NAMESPACE
        ),
        spec=V1beta1InferenceServiceSpec(predictor=predictor),
    )

    kserve_client = KServeClient(
        config_file=os.environ.get("KUBECONFIG", "~/.kube/config")
    )
    kserve_client.create(isvc)
    kserve_client.wait_isvc_ready(service_name, namespace=KSERVE_TEST_NAMESPACE)
    res = await predict_isvc(
        rest_v1_client,
        service_name,
        "./data/iris_input.json",
        network_layer=network_layer,
    )
    assert res["predictions"] == [1, 1]
    kserve_client.delete(service_name, KSERVE_TEST_NAMESPACE)


@pytest.mark.predictor
@pytest.mark.asyncio(scope="session")
async def test_predictive_lightgbm_v1(rest_v1_client, network_layer):
    service_name = "isvc-predictive-lightgbm"
    predictor = V1beta1PredictorSpec(
        min_replicas=1,
        model=V1beta1ModelSpec(
            model_format=V1beta1ModelFormat(name="lightgbm"),
            runtime="kserve-predictiveserver",
            storage_uri="gs://kfserving-examples/models/lightgbm/iris",
            resources=V1ResourceRequirements(
                requests={"cpu": "50m", "memory": "128Mi"},
                limits={"cpu": "100m", "memory": "256Mi"},
            ),
        ),
    )

    isvc = V1beta1InferenceService(
        api_version=constants.KSERVE_V1BETA1,
        kind=constants.KSERVE_KIND_INFERENCESERVICE,
        metadata=client.V1ObjectMeta(
            name=service_name, namespace=KSERVE_TEST_NAMESPACE
        ),
        spec=V1beta1InferenceServiceSpec(predictor=predictor),
    )

    kserve_client = KServeClient(
        config_file=os.environ.get("KUBECONFIG", "~/.kube/config")
    )
    kserve_client.create(isvc)
    kserve_client.wait_isvc_ready(service_name, namespace=KSERVE_TEST_NAMESPACE)

    res = await predict_isvc(
        rest_v1_client,
        service_name,
        "./data/iris_input_v3.json",
        network_layer=network_layer,
    )
    assert numpy.argmax(res["predictions"][0]) == 0
    kserve_client.delete(service_name, KSERVE_TEST_NAMESPACE)


@pytest.mark.predictor
@pytest.mark.asyncio(scope="session")
async def test_predictive_sklearn_v2(rest_v2_client, network_layer):
    service_name = "isvc-predictive-sklearn-v2"
    protocol_version = "v2"
    predictor = V1beta1PredictorSpec(
        min_replicas=1,
        model=V1beta1ModelSpec(
            model_format=V1beta1ModelFormat(name="sklearn"),
            runtime="kserve-predictiveserver",
            protocol_version=protocol_version,
            storage_uri="gs://kfserving-examples/models/sklearn/1.0/model",
            resources=V1ResourceRequirements(
                requests={"cpu": "50m", "memory": "128Mi"},
                limits={"cpu": "100m", "memory": "512Mi"},
            ),
            readiness_probe=client.V1Probe(
                http_get=client.V1HTTPGetAction(
                    path=f"/v2/models/{service_name}/ready", port=8080
                ),
                initial_delay_seconds=30,
            ),
        ),
    )

    isvc = V1beta1InferenceService(
        api_version=constants.KSERVE_V1BETA1,
        kind=constants.KSERVE_KIND_INFERENCESERVICE,
        metadata=client.V1ObjectMeta(
            name=service_name, namespace=KSERVE_TEST_NAMESPACE
        ),
        spec=V1beta1InferenceServiceSpec(predictor=predictor),
    )

    kserve_client = KServeClient(
        config_file=os.environ.get("KUBECONFIG", "~/.kube/config")
    )
    kserve_client.create(isvc)
    kserve_client.wait_isvc_ready(service_name, namespace=KSERVE_TEST_NAMESPACE)

    res = await predict_isvc(
        rest_v2_client,
        service_name,
        "./data/iris_input_v2.json",
        network_layer=network_layer,
    )
    assert res.outputs[0].data == [1, 1]

    kserve_client.delete(service_name, KSERVE_TEST_NAMESPACE)


@pytest.mark.predictor
@pytest.mark.asyncio(scope="session")
async def test_predictive_xgboost_v2(rest_v2_client, network_layer):
    service_name = "isvc-predictive-xgboost-v2"
    protocol_version = "v2"
    predictor = V1beta1PredictorSpec(
        min_replicas=1,
        model=V1beta1ModelSpec(
            model_format=V1beta1ModelFormat(name="xgboost"),
            runtime="kserve-predictiveserver",
            protocol_version=protocol_version,
            storage_uri="gs://kfserving-examples/models/xgboost/1.5/model",
            resources=V1ResourceRequirements(
                requests={"cpu": "50m", "memory": "128Mi"},
                limits={"cpu": "100m", "memory": "512Mi"},
            ),
            readiness_probe=client.V1Probe(
                http_get=client.V1HTTPGetAction(
                    path=f"/v2/models/{service_name}/ready", port=8080
                ),
                initial_delay_seconds=30,
            ),
        ),
    )

    isvc = V1beta1InferenceService(
        api_version=constants.KSERVE_V1BETA1,
        kind=constants.KSERVE_KIND_INFERENCESERVICE,
        metadata=client.V1ObjectMeta(
            name=service_name, namespace=KSERVE_TEST_NAMESPACE
        ),
        spec=V1beta1InferenceServiceSpec(predictor=predictor),
    )

    kserve_client = KServeClient(
        config_file=os.environ.get("KUBECONFIG", "~/.kube/config")
    )
    kserve_client.create(isvc)
    kserve_client.wait_isvc_ready(service_name, namespace=KSERVE_TEST_NAMESPACE)

    res = await predict_isvc(
        rest_v2_client,
        service_name,
        "./data/iris_input_v2.json",
        network_layer=network_layer,
    )
    assert res.outputs[0].data == [1, 1]

    kserve_client.delete(service_name, KSERVE_TEST_NAMESPACE)


@pytest.mark.predictor
@pytest.mark.asyncio(scope="session")
async def test_predictive_lightgbm_v2(rest_v2_client, network_layer):
    service_name = "isvc-predictive-lightgbm-v2"
    protocol_version = "v2"
    predictor = V1beta1PredictorSpec(
        min_replicas=1,
        model=V1beta1ModelSpec(
            model_format=V1beta1ModelFormat(name="lightgbm"),
            runtime="kserve-predictiveserver",
            protocol_version=protocol_version,
            storage_uri="gs://kfserving-examples/models/lightgbm/iris",
            resources=V1ResourceRequirements(
                requests={"cpu": "50m", "memory": "128Mi"},
                limits={"cpu": "100m", "memory": "512Mi"},
            ),
            readiness_probe=client.V1Probe(
                http_get=client.V1HTTPGetAction(
                    path=f"/v2/models/{service_name}/ready", port=8080
                ),
                initial_delay_seconds=30,
            ),
        ),
    )

    isvc = V1beta1InferenceService(
        api_version=constants.KSERVE_V1BETA1,
        kind=constants.KSERVE_KIND_INFERENCESERVICE,
        metadata=client.V1ObjectMeta(
            name=service_name, namespace=KSERVE_TEST_NAMESPACE
        ),
        spec=V1beta1InferenceServiceSpec(predictor=predictor),
    )

    kserve_client = KServeClient(
        config_file=os.environ.get("KUBECONFIG", "~/.kube/config")
    )
    kserve_client.create(isvc)
    kserve_client.wait_isvc_ready(service_name, namespace=KSERVE_TEST_NAMESPACE)

    res = await predict_isvc(
        rest_v2_client,
        service_name,
        "./data/iris_input_v2.json",
        network_layer=network_layer,
    )
    # LightGBM returns probability predictions in v2 format
    # We verify the highest probability class matches expected result
    assert numpy.argmax(res.outputs[0].data[0]) == 0

    kserve_client.delete(service_name, KSERVE_TEST_NAMESPACE)