L1DeepMET

Hardware-aware deep learning framework for Level-1 Missing Transverse Energy reconstruction on FPGAs for CMS and the next generation HL-LHC.

This project is an extensive refactor and extension of https://github.com/ucsd-hep-ex/L1METML

Quick Start

# Setup
micromamba env create -f environment.yaml
micromamba activate l1deepmet
pip install -e .

# Run preprocessing
dvc repro preprocess

Pipeline

1. Preprocess - Load ROOT files, clean data, create train/val/test splits
2. Train - Neural network training (coming soon)
3. Evaluate - Performance assessment (coming soon)

Documentation

• Data Preprocessing - Pipeline details
• DVC Workflow - Experiment management