REPRO.md

🔁 Reproducibility Instructions

This document provides the full set of instructions to reproduce our project results from scratch, including data setup, environment configuration, training, and evaluation.

---

🧱 Project Structure

.
├── data/                   # Contains raw and processed datasets
├── src/                    # All source code (models, training, evaluation)
├── requirements.txt        # Python dependencies
├── README.md               # README file
├── REPRO.md                # This file
├── XXXXX
├── XXXXX

---

⚙️ Environment Setup

Setup project by running the following commands:

# Example -- overwrite if needed
conda create -n XXXXX python=XXXX
conda activate XXXXX
pip install -r requirements.txt

---

📂 Download & Prepare Datasets

Place your datasets in the XXXX/ directory.

Example Dataset

mkdir -p data/example_dataset
cd data/example_dataset
wget xxxxx
python -m src.preprocess_example_dataset.py xxxx
cd ../..

---

⚙️ Configuration

Set your parameters in the config file before training. Example:

---

🚀 5. Training

Baselines

Run the following command to train the baseline:

python XXXX

To perform inference:

python XXXX

Alternatively, execute the following slurm jobs:

sbatch job_scripts/train_xxxxx.job
sbatch job_scripts/infer_xxxxx.job

---

📈 Evaluation

After training, evaluate all models with:

python XXXX

---

📎 Misc. Notes (optional)

---

📦 Dependencies / References

This project repository uses the following frameworks / refers to the following papers:

• XXX
• XXX