GNN Tracking Hyperparameter Optimization 2.0

This repository hosts submission scripts and framework for hyperparameter optimization of the models defined in the main library. It is a successor to the original hyperparameter optimization repository that became obsolete after the main library was refactored to use pytorch lightning.

✨ Features

• Configure your model with only yaml files
• Submission files for SLURM
• Upload your results to the Weights & Biases platform

🗺️ Layout of this repository

• scripts: This is the main content: All the config files and submission scripts to start training are here
• bin: Helper scripts that might be worth adding to your $PATH (i.e., that apply to all experiments/subprojects)
• src/hpo2: Python package with additional helper files that don't have a place in the main library (currently empty)

🚀 Installation

First, follow the instructions from the main library to set up the mamba environment and install the main library. Then run

pip3 install --editable '.[dev,testing]'

for this library.

For the helper scripts, install xonsh:

pip3 install xonsh
xonsh link_bin.xsh

📖 Further reads

• wandb-osh: package to trigger wandb syncs on compute nodes without internet