Brian Van Essen

LBANN

Van Essen is first author of "LBANN: Livermore big artificial neural network HPC toolkit", presented at a 2015 workshop on machine learning in high-performance computing environments. A 2016 LLNL article said Van Essen's group was working to improve LBANN's speed and efficiency by maximizing node utilization while reducing communication between nodes.⁴⁵

LBANN is an open-source deep learning training framework built to compose model, data and ensemble parallelism on supercomputers. A 2025 LLNL newsletter reported that the LBANN group, including Van Essen, had released a parallel LLaMA model in PyTorch optimized for AMD chips and demonstrated it on Tenaya, an early-access system for El Capitan.⁶⁷

Scientific applications

As principal investigator, Van Essen led a 2020 LBANN-based generative model for COVID-19 drug discovery, trained on more than a billion molecules on the Sierra supercomputer, which was a finalist for the Gordon Bell Special Prize and was presented at SC20. Van Essen was also part of the cognitive simulation team that applied machine learning to inertial confinement fusion experiments.⁸⁹

With Fred Streitz, Van Essen led LLNL's part of the ECP-CANDLE project, a collaboration of national laboratories and the National Cancer Institute applying deep learning to cancer research, whose LLNL work centered on scalable distributed training within LBANN.¹⁰

Work and contributions

• LBANN (Livermore Big Artificial Neural Network Toolkit) Open-source code

  Co-led, with Barry Chen, the Livermore team that developed LBANN in 2015, and is first author of the 2015 workshop paper introducing the toolkit.³⁴²

• ECP-CANDLE (LLNL contribution)

  Co-led LLNL's part of the ECP-CANDLE cancer deep learning project with Fred Streitz, which focused on scalable distributed deep learning tools within LBANN.¹⁰

Timeline

1. 2015Co-led the Livermore team that developed the LBANN toolkit and first-authored its workshop paper.³⁴
2. Jun 2016Led a group working to improve LBANN's speed and node utilization, as reported by LLNL.⁵
3. Nov 2020Presented an LBANN-based COVID-19 drug discovery model, a Gordon Bell Special Prize finalist, at SC20.⁸
4. Nov 2022Was part of the LLNL cognitive simulation team recognized by HPCwire at SC22.⁹
5. 2023The ECP-CANDLE software suite, with LLNL's part co-led by Van Essen, received an R&D 100 award.¹⁰
6. 2025The LBANN group demonstrated its parallel LLaMA model on Tenaya, an early-access El Capitan system.⁷

Honors and recognition

• R&D 100 Award (ECP-CANDLE, as co-lead of the LLNL team) · R&D World, 2023¹⁰
• HPCwire Editor's Choice Award, Best Use of HPC in Energy (LLNL cognitive simulation team) · HPCwire, 2022⁹

Where they work

• Lawrence Livermore National Laboratory 2 people in the wiki

Related people

• Kimberly S. BudilAlso at Lawrence Livermore National Laboratory
• Frédéric BastienML frameworks and libraries · Training systems
• Martín AbadiML frameworks and libraries · Training systems
• Sylvain GuggerML frameworks and libraries · Training systems
• James BradburyML frameworks and libraries · Training systems
• Clément FarabetTraining systems · ML frameworks and libraries
• Alban DesmaisonML frameworks and libraries · Training systems
• Pieter NoordhuisTraining systems · ML frameworks and libraries
• Luca AntigaML frameworks and libraries · Training systems
• Edward YangML frameworks and libraries · Training systems