Brevitas Quantization Workshop

from Monday, 19 October 2026 (09:00) to Tuesday, 20 October 2026 (20:35)
CERN (503/1-001)

        : Sessions
    /     : Talks
        : Breaks
19 Oct 2026
20 Oct 2026
AM
09:00
Presentations (until 09:40) (222/R-001)
09:00 Welcome - Lorenzo Moneta (CERN)  
09:10 Brevitas Introduction - Nick Fraser (AMD)  
09:40
Tutorials (until 11:10) (222/R-001)
09:40 Brevitas tutorial - Giuseppe Franco (AMD)  
11:10 --- Coffee break ---
11:30
Presentations (until 12:00)
11:30 PeRQ: Pushing the Limits of Block Rotations in Post-Training Quantization - Nicholas Fraser (AMD)  
09:30
Presentations (until 10:30) (503/1-001 - Council Chamber)
09:30 dMX: presenting learned mixed precision floating point quantization - Giuseppe Franco (AMD)  
10:00 PiSO: optimal scaling factor selection and composability with PTQ algorithms - Juan Pablo Garcia Amboag Pablo Monteagudo Lago (AMD)  
10:30 --- Coffee break ---
11:00
Presentations (until 11:30) (503/1-001 - Council Chamber)
11:00 PQuantML introduction - Roope Oskari Niemi  
11:30
Tutorials (until 12:00) (503/1-001 - Council Chamber)
11:30 PQuantML tutorial - Roope Oskari Niemi  
PM
12:00 --- Lunch break ---
13:30
Presentations (until 15:30)
13:30 LHCb: use case - Mohamed Elashri (University of Cincinnati)  
14:00 CMS: use case - Christopher Edward Brown (CERN)  
14:30 Application of low-level compression and quantization in ALICE online processing - Christian Sonnabend (CERN, Heidelberg University (DE))  
15:00 ATLAS: use case - Diego Olmos Patiño (Universidad de La Serena (CL))  
15:30 --- Coffee break ---
16:00
Presentations (until 16:30)
16:00 QONNX intro - Marius Köppel (ETH Zurich (CH))  
16:30
Tutorials (until 17:15)
16:30 QONNX tutorial - Marius Köppel (ETH Zurich (CH))  
17:15
ATLAS excursion (until 18:30)
12:00 --- Lunch break ---
13:30
Presentations (until 15:00) (503/1-001 - Council Chamber)
13:30 AIE4ML introduction - Dimitrios Danopoulos (CERN)  
14:00 CERN openlab: A transformer implementation on AI-engines - Elias Leutgeb (CERN)  
14:30 CERN openlab: Anomaly detection in ATLAS: lessons from Run 3 - Paula Martinez Suarez (CERN)  
15:00 --- Coffee break ---
15:30
Presentations (until 17:15) (503/1-001 - Council Chamber)
15:30 A Hyperparameter Optimization Framework for Preparing ML Models for Real-Time Trigger Deployment - Anastasiia Petrovych (CERN)  
15:45 Accelerating ML Inference on heterogeneous architectures using SOFIE - Sanjiban Sengupta (CERN, University of Manchester)  
16:00 A2Q and A2Q+: these works show how to use low-precision accumulators with guaranteed overflow avoidance - Ian Colbert (AMD)  
16:30 HGQ - Chang Sun (California Institute of Technology (US))