Analyzing experiments#

To follow the provided examples, make sure you installed SQLTrack with the full optional deptendencies: pip install sqltrack[full].

Important

Whichever platform you choose and what you do with it is up to you. These examples are meant to guide and inspire.

Jupyter notebooks#

Our first example is a Jupyter notebook (see examples/notebook.ipynb) that recreates an mlflow-like experience through SQLTrack, but with some important improvements, like displaying the best metric value for a run instead of the latest.

[Optional] Self-signed SSL certificate#

You can create a SSL self-signed certificate to use Juypter Lab with HTTPS:

openssl req -x509 -newkey rsa:4096 -keyout jupyter.key -out jupyter.crt -sha256 -days 365 -nodes

Start Jupyter Lab with your certificate:

jupyter-lab [options...] --certfile jupyter.crt --keyfile jupyter.key

Plotly Dash#

Todo

Create a custom UI with Dash.