Repo contains two notebooks which demonstrate functionality of MLFlow tracking:
/notebooks/exp_mlflow_local_autolog.ipynb: basic use with autlog() from MLflow
/notebooks/exp_mlflow_serve_custom.ipynb: custom configuration of tracking parameters
More detailed information about model can be found in this repo and blog post
conda create -n mlflow_tracking-env python=3.9
conda activate mlflow_tracking-env
pip install -r requirements.txt
Option 1: without using tracking server
cd /notebooks #mlflow should be always launched from the folder with notebooks/scripts
mlflow ui
Go to http://127.0.0.1:5000 which will open UI for tracked experiments.
Option 2: with local tracking server
cd /notebooks #mlflow should be always launched from the folder with notebooks/scripts
mlflow server --backend-store-uri sqlite:///backend.db --default-artifact-root ./artifacts_local