Just a clone-able, runnable (🤞), minimal example of the Ludwig low code toolbox.
The example is the Auto MPG example used in the Ludwig documentation. That is itself a recreation of the basic regression TensorFlow example. Here, we include the dataset and explicit instructions.
The automobiles
directory contains:
- A cleaned csv of the Auto MPG data set. Rows with missing values have had the
?
character from the original dataset removed to avoid parsing errors. - A model specification file,
model_definition.yaml
. - A run script with the right Ludwig invocation,
experiment.sh
.
Assuming you have a working python3 installation and git on a unix operating system.
- Open a terminal.
- Clone this repo somewhere sensible with
git clone git@github.com:cjwallace/a-tiny-ludwig-example.git
. cd
into the repo.- Run
python3 -m venv venv
. - Run
source venv/bin/activate
. - Run
pip install ludwig
. cd
into theautomobiles
directory.- Run
./experiment.sh
.- If this doesn't work, you may need to run
chmod +x experiment.sh
first to give the script permission to execute. - Alternatively, copy the contents of
experiment.sh
and run it on the command line directly.
- If this doesn't work, you may need to run
- Sit back and relax.