1. Setup

git clone git@github.com:Run3-anomaly-tagging/AutoencoderTraining.git
cd AutoencoderTraining
python3 -m venv venv
cd ..
echo $PWD > AutoencoderTraining/venv/lib/python3.9/site-packages/project_root.pth
cd AutoencoderTraining
source venv/bin/activate
pip install -r requirements.txt

2. Config

Edit configs/dataset_config.json to match your data paths and cross-sections.

3. Check datasets

python sampling/explore_files.py

4. Create samples training dataset (HT merging)

python sampling/sample_strategy.py

5. Training

python training/dense_autoencoder/train.py 
python training/image_autoencoder/train.py 

On lpc-gpu nodes needs to be done with apptainer

./run_training_image.sh #The warning only concerns the initial directory for singularity image, it does not affect the results
./run_training_dense.sh

6. Evaluation

python evaluation/plot_reco_loss.py