• Author: Fantine Huot
• DOI: 10.5281/zenodo.6344771

After cloning the repository, run the following commands to initialize and update the submodules.

git submodule init
git submodule update

You can run the project from an interactive bash session within the provided Docker container:

docker run --gpus all -it fantine/ml_framework:latest bash

If you do not have root permissions to run Docker, Singularity might be a good alternative for you. Refer to containers/README.md for more details.

• bin: Scripts to run machine learning jobs.
• catalog: Earthquake and background noise database.
• config: Configuration files.
• containers: Details on how to use containers for this project.
• docs: Documentation.
• hptuning: Hyperparameter tuning for machine learning.
• log: Directory for log files.
• ml_framework: Machine learning framework.
• preprocessing: Data preprocessing steps.
• processing_utils: Processing utility functions.
• tfrecords: Utility functions for converting input files to TFRecords.

Set the DATAPATH variable inside config/datapath.sh to the data or scratch directory to which you want write data files.

This repository provides a parameterized, modular framework for creating and running ML models.

• Convert input data to TensorFlow records
• Machine learning training and inference
• Hyperparameter tuning