Topic: Scientific ML

New CASC research puts sparse autoencoders and concept bottlenecks to work on foundation models.

LLNL researchers have posters and workshop papers accepted to the IEEE/CVF Conference on Computer Vision and Pattern Recognition on June 3–7.

Genesis is a coordinated national effort involving all 17 DOE national laboratories, alongside universities, industry and federal partners.

LLNL leaders, scientists, and engineers joined national voices at the Special Competitive Studies Project’s AI+ Expo, highlighting how AI is reshaping science, security, and energy innovation.

Highlights include turbulent flows, virtual reality, correctness checkers, and AI for chemistry.

LLNL researchers have several posters and papers accepted to the 13th International Conference on Learning Representations on April 23–27.

Six LLNL Computing researchers have been named Distinguished Members of Technical Staff in recognition of their extraordinary scientific and technical contributions.

LLNL scientists initiated and led a cross-disciplinary team that developed a machine learning model to distinguish opioids from other chemicals.

During the weeklong conference, attendees visiting the Department of Energy’s booth were treated to two technical demonstrations and a talk by LLNL staff.

LLNL’s presence, which included dozens of sessions, including tutorials, workshops, paper presentations and birds-of-a-feather meetings was felt across virtually every major event of the week.

Join LLNL at the 39th annual Conference on Neural Information Processing Systems on December 2–7.

Livermore researchers are engaged in efforts to apply correctness and formal methods to improve the reliability, reproducibility, and accuracy of the Laboratory’s high performance computing codes.

Scientists at LLNL and collaborators at AMD and Columbia University have achieved a milestone in biological computing: completing the largest and fastest protein structure prediction workflow ever run, using the full power of El Capitan.

LLNL is participating in the 37th annual International Conference for High Performance Computing, Networking, Storage, and Analysis (SC25) in St. Louis on November 16–21, 2025.

Building on our leadership in HPC and AI and our long open-source tradition, ElMerFold is a high performance framework for large-scale inference and distillation on LLNL supercomputers with OpenFold-specific optimizations.

LLNL researchers employed an AI-driven model to predict fusion ignition days ahead of the historic 2022 shot, according to a new study in Science.

A new CASC paper proposes unity and clarity around foundation models in computational science, offering an implementation framework inspired by finite element methods.

LLNL researchers have posters and workshop papers accepted to the 42nd International Conference on Machine Learning on July 13–19.

A new cancer drug candidate developed by LLNL, BridgeBio Oncology Therapeutics, and the Frederick National Laboratory for Cancer Research has demonstrated the ability to block tumor growth without triggering a common and debilitating side effect.

As the application of AI across industries accelerates the pace of development, so too must national security remain at the cutting edge, a task requiring extensive collaboration to deploy the nation’s most critical resources.

LLNL participates in the ISC High Performance Conference (ISC25) on June 10–13.

LLNL researchers have posters and workshop papers accepted to the 13th International Conference on Learning Representations on April 24–28.

Researchers from LLNL, in collaboration with other leading institutions, have successfully used an AI-driven platform to preemptively optimize an antibody to neutralize SARS-CoV-2 variants.

The February 28 event brought together over 1,400 Department of Energy scientists across multiple sites to explore how cutting-edge AI models could transform scientific research.

Highlights include ML techniques for computed tomography, a scalable Gaussian process framework, safe and trustworthy AI, and autonomous multiscale simulations.