LigateProject / ligen_autotuning

Tools for LiGen autotuning

Parallel Asynchronous Bayesian Optimization-Machine Learning (PAMaliboo)

The PAMaliboo submodule contains a library for parallel optimization via BO techniques hybridized with ML models, including classes specifically tailored for the LiGen optimization.

The ligen_simulated_campaign.py script is an example of statistically robust exploration campaign for finding the best LiGen configuration.

aMLLibrary

aMLLibrary is an open-source, high-level Python package that uses supervised ML techniques to train regression models on a given dataset. It is also included as a submodule in this repository.

The ligen_aml_config_files folder contains examples of text configuration files used for LiGen autotuning of aMLLibrary.

Plotting tools

This repository also contains some Python scripts which produce some plots to help visualize the LiGen datasets. They can be executed as simply as:

python3 analysis.py

Produced plots will be saved in the plots folder of this repository.