This repository contains the code and implementation details for the AXOL1TL v5 along with the MLOps stategies associated to it. AXOL1TL v5 is a real-time anomaly detection framework for the CMS Level-1 Trigger. This trigger is designed to operate under extreme throughput constraints (~40 MHz) using self-supervised learning.
Complete documentation, motivation, methodology, and visualizations are available here:
https://dc250601.github.io/TechStudent24/
Please refer to the linked page for a full description of the codebase.