# VITIS AI Models for LHC Decision-Making

This repository serves as an archive of functional neural network models designed to run on the **VITIS AI VEK280** engine. The goal is to explore machine learning-driven decision-making for the **Large Hadron Collider (LHC)**.

I'm conducting this research on a remote server, and this repository provides a centralized location to pull the latest compiled models and run experiments efficiently.

## Included Models

- **ResNet-50**  
- **ResNet-18**  
  Both are compiled into `.xmodel` format and are ready to run using the `resnet_pt` runtime executable.

- **EfficientNet B0**  
  A custom model I built in PyTorch. *(Not yet compiled into `.xmodel` format)*

> **Note:** To run any models using the VITIS AI runtime, you must be inside the correct **container** or working from a **VITIS-compatible environment**.

## Research Objective

The main objective of this project is to develop a **lightweight feedforward neural network** optimized for processing **particle collision data** from the LHC. This model will be:

- Fully quantized and built in C++  
- Designed to bypass DPU usage  
- Optimized specifically for deployment on the **VEK280 platform**

All models in this repository are **convolutional neural networks (CNNs)** unless otherwise noted.