/machine_learning_serving

Primary LanguagePythonMIT LicenseMIT

machine learning serving

image01

Install

git clone 
cd machine_learning_serving
python3.8 -m venv venv
. venv/bin/activate
pip install -U pip
pip install -r requirements.txt

MLflow

cd mlflow_tutorial
mlflow ui --backend-store-uri sqlite:///store.db

#Open new terminal
cd machine_learning_serving
. venv/bin/activate
cd mlflow_tutorial
python train.py 0.3 0.5

#open browser http://localhost:5000

Airflow

. venv/bin/activate
cd airflow_tutorial
export AIRFLOW_HOME=$(pwd)/airflow
export PYTHONPATH=$(pwd):$PYTHONPATH
export OBJC_DISABLE_INITIALIZE_FORK_SAFETY=YES

mkdir airflow/dags
cp dags/ml_pipeline.py airflow/dags/ml_pipeline.py
airflow dags list

airflow db init
airflow users create \
    --username admin \
    --firstname Peter \
    --lastname Parker \
    --role Admin \
    --email spiderman@superhero.org



airflow webserver --port 8080
#open browser http://localhost:8080 and enable ml_pipeline

#Open new terminal
cd machine_learning_serving
. venv/bin/activate
cd airflow_tutorial
python serve.py

#Open new terminal
cd machine_learning_serving
. venv/bin/activate
cd airflow_tutorial
export AIRFLOW_HOME=$(pwd)/airflow
export PYTHONPATH=$(pwd):$PYTHONPATH
export OBJC_DISABLE_INITIALIZE_FORK_SAFETY=YES
airflow scheduler

References