About ALCF
Backed by a team of experts, ALCF helps users achieve their research goals with a variety of capabilities, including inference-as-a-service tools, infrastructure for model training, and large-scale data storage and sharing.
Facility Expertise
Our facility has a team of world-class team of computational scientists, performance engineers, visualization experts, HPC systems administrators, and support staff to help researchers maximize their use of the facility’s supercomputing resources. You can learn more about our staff here.
Inference Tools
ALCF’s inference-as-a-service for science provides its research community with cloud-like access to large language models (LLMs) and foundation models to integrate with their existing workflows. These inference tools allow researchers to deploy AI models within our supercomputing ecosystem, integrate them with HPC workflows, and operate at scale.
This service provides an API (following the OpenAI API standard) and a web-based chat interface for a variety of state-of-the-art open-source models running on dedicated ALCF hardware. Our inference service is available for scientists to use and runs a variety of open-source models.
Model Training
ALCF’s resources support small-, medium-, and large-scale model training for AI and scientific research. Every year, researchers leverage ALCF resources to drive breakthroughs in a variety of domains, from protein design to hypersonic flight simulations. Users can take advantage of our high-performance computing infrastructure, advanced software environments, and expert support to scale training, optimize efficiency, and accelerate experimentation.
Data Sharing
Research conducted with ALCF resources often produces large, high-fidelity, cutting edge scientific datasets. ALCF supports research through high-performance data storage and sharing tools that help facility collaborate and share their data with external users.