MUSH (Model Uncertainty Sensitivity to Hyperparameters)

Exploring the Sensitivity of Model Uncertainty to Hyperparameter Variations: A Systematic Analysis

Structure

• src/ - Source code (data, models, training, utils)
• notebooks/ - Jupyter notebooks for experiments
• data/ - Raw and processed data (not tracked by git)
• logs/ - Training logs
• outputs/ - Model checkpoints, predictions
• mlruns/ - MLflow experiment tracking
• scripts/ - Standalone scripts
• tests/ - Unit tests

Setup

See environment.yml for dependencies.

Usage

• Use Jupyter notebooks in notebooks/ for experiments.
• Place source code in src/.
• Use mlruns/ for MLflow tracking.