asvyatkovskiy / ludwig-intro

Hands-on introduction to Ludwig. Based on https://uber.github.io/ludwig/examples/

Getting to know Ludwig

Ludwig is a Python tool for code-free deep learning.

1. Official Ludwig GitHub
2. Official Ludwig website

Getting started

To follow, one needs a laptop with Miniconda (a minimal version of Anaconda) and Ludwig installed.

Downloading and installing Miniconda

Please go to the Anaconda website. Download and install the latest Miniconda version for Python 3.6 for your operating system.

wget <http:// link to miniconda>
sh <miniconda .sh>

After that, type:

conda --help

Create isolated Miniconda environment

Copy following into a file named environment.yaml:

name: hello-ludwig
channels:
  - conda-forge
dependencies:
  - python=3.6
  - pip:
    - ludwig
    - jupyter

then do:

conda env create -f environment.yml
source activate hello-ludwig

Enable Miniconda kernel in Jupyter

To make newly created miniconda environment visible in the Jupyter, install ipykernel:

python -m ipykernel install --user --name hello-ludwig --display-name "Hello, Ludwig"

Read official Ludwig user guide

https://uber.github.io/ludwig/user_guide/

We will be looking into text classification and time series forecasting examples closely following https://uber.github.io/ludwig/examples/

#export KMP_DUPLICATE_LIB_OK=True