Train Machine Learning models with MLflow, Deploy with Seldon

MLflow

MLflow: the open-source platform for the machine leaning lifecycle, 管理machine learning整个生命周期的一款开源产品,主要提供了三种服务:

• MLflow Tracking: 记录并维护了machine learning的代码,数据,matrics,config,results…并结合UI展示
• MLflow Projects: 将machine learning的model带包成一个docker image,实现run anywhere
• MLflow Models: 标准化machine learning的model及其configuration files,实现与其他平台共同开发/部署

几乎支持市面上的所有Machine Learning frameworks, TensorFlow/PyTorch/Spark/SKlearn/R…

开源,并有着Databricks/Microsoft等一众公司的committer.

Seldon

Seldon: the open-source platform to help deploy machine learning models, 主要focus在model的deployment

• 可以deploy市面上几乎所有的machine learning model
• 不仅可以deploy在both in cloud and on-promise
• expose metrics/HTTP trance等monitoring信息

Example: Train ML model using MLflow, Deploy using Seldon

Train Model with MLflow

import os
import warnings
import sys

import pandas as pd
import numpy as np
from sklearn.metrics import mean_squared_error, mean_absolute_error, r2_score
from sklearn.model_selection import train_test_split
from sklearn.linear_model import ElasticNet

import mlflow
import mlflow.sklearn

def eval_metrics(actual, pred):
    rmse = np.sqrt(mean_squared_error(actual, pred))
    mae = mean_absolute_error(actual, pred)
    r2 = r2_score(actual, pred)
    return rmse, mae, r2

if __name__ == "__main__":
    warnings.filterwarnings("ignore")
    np.random.seed(40)

    # Read the wine-quality csv file (make sure you're running this from the root of MLflow!)
    wine_path = os.path.join(os.path.dirname(os.path.abspath(__file__)), "wine-quality.csv")
    data = pd.read_csv(wine_path)

    # Split the data into training and test sets. (0.75, 0.25) split.
    train, test = train_test_split(data)

    # The predicted column is "quality" which is a scalar from [3, 9]
    train_x = train.drop(["quality"], axis=1)
    test_x = test.drop(["quality"], axis=1)
    train_y = train[["quality"]]
    test_y = test[["quality"]]

    alpha = float(sys.argv[1]) if len(sys.argv) > 1 else 0.5
    l1_ratio = float(sys.argv[2]) if len(sys.argv) > 2 else 0.5

    mlflow.set_experiment('test')

    with mlflow.start_run():
        lr = ElasticNet(alpha=alpha, l1_ratio=l1_ratio, random_state=42)
        lr.fit(train_x, train_y)
        predicted_qualities = lr.predict(test_x)
        (rmse, mae, r2) = eval_metrics(test_y, predicted_qualities)
        print("Elasticnet model (alpha=%f, l1_ratio=%f):" % (alpha, l1_ratio))
        print("  RMSE: %s" % rmse)
        print("  MAE: %s" % mae)
        print("  R2: %s" % r2)
        mlflow.log_param("alpha", alpha)
        mlflow.log_param("l1_ratio", l1_ratio)
        mlflow.log_metric("rmse", rmse)
        mlflow.log_metric("r2", r2)
        mlflow.log_metric("mae", mae)
        mlflow.sklearn.log_model(lr, "model")

在对应的Model Storage下，可以看到MLmodel文件这个文件内包含了很多信息：模型本身model.pkl，模型产生的env conda.yaml… 之后Seldon会读取这部分信息去做deploy

artifact_path: model
flavors:
  python_function:
    data: model.pkl
    env: conda.yaml
    loader_module: mlflow.sklearn
    python_version: 3.6.5
  sklearn:
    pickled_model: model.pkl
    serialization_format: cloudpickle
    sklearn_version: 0.21.3
run_id: 26f04f36493b4982a064bb8d6e9d9b30

Deploy ML model with Seldon

Prerequisites:

• a k8s cluster
• helm installed

curl https://raw.githubusercontent.com/helm/helm/master/scripts/get > get_helm.sh
chmod 777 get_helm.sh
./get_helm.sh
helm init #install Tiller, a deployment/service/pod of Tiller will be installed automatically in **kube-system** NS
#### create account for Tiller
kubectl create serviceaccount --namespace kube-system tiller
kubectl create clusterrolebinding tiller-cluster-rule --clusterrole=cluster-admin --serviceaccount=kube-system:tiller
kubectl patch deploy --namespace kube-system tiller-deploy -p '{"spec":{"template":{"spec":{"serviceAccount":"tiller"}}}}'

install Seldon

helm install \
    seldon-core-operator \
    --name seldon-core \
    --repo https://storage.googleapis.com/seldon-charts \
    --namespace seldon-system \
    --set usagemetrics.enabled=true \
    --set ambassador.enabled=true

install Ambassador (k8s cloud native gateway)

helm install stable/ambassador --name ambassador --set crds.keep=false
kubectl rollout status deployment.apps/ambassador

Port forwarding:

#### run below command in another terminal
kubectl port-forward $(kubectl get pods -l app.kubernetes.io/name=ambassador -o jsonpath='{.items[0].metadata.name}') 8003:8080

install Seldon Analytics

# install Seldon Analytics with prometheus and grafana
helm install seldon-core-analytics --name seldon-core-analytics \
     --repo https://storage.googleapis.com/seldon-charts \
     --set grafana_prom_admin_password=password \
     --set persistence.enabled=false

#### run below command in another terminal
kubectl port-forward \
    $(kubectl get pods \
        -l app=grafana-prom-server -o jsonpath='{.items[0].metadata.name}') \
    3000:3000

Deploy Model

apiVersion: machinelearning.seldon.io/v1alpha2
kind: SeldonDeployment
metadata:
  name: test
spec:
  name: rex
  predictors:
  - graph:
      children: []
      implementation: MLFLOW_SERVER
      modelUri: s3://mlflow/xxx/artifacts/model
      envSecretRefName: s3-secret
      name: classifier
    name: default
    replicas: 1