Example code illustrating using Uber's ludwig deep learning framework.
Objectives:
- Define Docker images for the Ludwig software stack, including both Tensorflow (cpu-enabled) and Tensorflow (gpu-enabled)
- Demonstrate running ludwig using command line execution in a Docker container
- Demonstrate running ludwig using Python api in a Docker container
- Generate modeling assessment visualizations, e.g., learning curves, confusion matrix, etc.
- Provide samples for various types of models: image classification, text analytics, sentiment analysis, time series forecasting, etc.
Directory | Description |
---|---|
bin | bash scripts for various function |
containers | Docker containers for ludwig software stack |
kaggle_titanic | Kaggle Titanic predictive competition data set |
mnist | Use of ludwig with mnist data set |
text_classification | Text classification model |
time_series | Time series forecasting temperature |
Preparatory steps:
- Create docker images with ludwig software stack. Run the following bash script.
bin/build_images tf_cpu
bin/build_images tf_gpu