This is a sample project that ilustrates how to use MLflow and Jupyter to solve a classification problem.
- Download the famous MNIST dataset from Kaggle and place then under the
data
folder. - Ensure you have installed
jupyter
,matplotlib
,mlflow
,pandas
,numpy
andsklearn
on your python distribution. - Navigate to the root folder of this project and, using the console, run jupyter and MLflow:
jupyter notebook
mlflow ui
- Open Jupyter on a brower tab. Most likely the URL would be something like
http://localhost:8888/notebooks/MNIST data.ipynb
. - Open MLflow on a different brower tab. In this case the URL would be something like
http://127.0.0.1:5000/
. - Run all the cells on the Jupter notebook:
Top menu > Cells > Run all
. Wait the notebook to complete. - Inspect the results of each of the experiments in the MLflow interface.