News

Argonne Lab Collaborates with Intel and HP to Develop Aurora GPT: A Trillion-Parameter AI Model for Scientific Knowledge Storage

The American Argonne National Laboratory pioneers Aurora GPT, a supercomputer-trained AI model aimed at revolutionizing scientific knowledge storage and access.

In a groundbreaking collaboration, the American Argonne National Laboratory (ANL), in partnership with Intel and HP, embarks on the creation of Aurora GPT, an Artificial Intelligence (AI) model designed to house vast scientific knowledge, boasting over a trillion parameters.

While supercomputers traditionally tackle monumental challenges, ANL’s initiative with Aurora GPT seeks to unlock “new opportunities for advancing science and engineering.”

The core objective revolves around training the Aurora GPT AI model on the Aurora supercomputer, one of the world’s most potent computing machines, aimed at encapsulating an extensive repository of scientific knowledge.

Read Also: How to Installing the ChatGPT App on Your Android Device

ANL envisions an AI interface, equipped with a ‘chatbot’ functionality, enabling researchers to pose inquiries across diverse domains such as biology, physics, mathematics, and engineering, as detailed on the laboratory’s website.

The Aurora GPT AI promises to broaden research horizons, accommodating a myriad of workloads encompassing machine learning, data-intensive tasks, and traditional simulation and modeling campaigns.

Notably, the project aims to leverage exascale computing capabilities, indicating a minimum operational speed of one exaflop per second. An exaflop operation has the potency to compute a minimum of one trillion floating-point operations per second, effectively accelerating research timelines.

Intel spearheads this exascale power, integrating a GPU architecture featuring six Intel Max Series GPUs of 128GB each, alongside two Intel Xeon CPU Max Series processors of 64GB each.

Read Also: YouTube Music Unveils AI Tools with Artists for Creative Breakthroughs

While currently in its nascent stages of development, Aurora GPT operates within a confined capacity of 256 nodes. However, future plans involve expansion to harness the full potential of the Aurora supercomputer, accommodating up to 10,000 nodes.

Emphasizing ongoing efforts, the laboratory underscores the dedication of Argonne developers in preparing Aurora GPT to cater comprehensively to scientific endeavors right from its inception.

The Aurora GPT project signifies a monumental leap towards democratizing access to vast repositories of scientific knowledge, propelling research and exploration across various disciplines into uncharted territories.

Related Articles

Leave a Reply

Your email address will not be published. Required fields are marked *

Back to top button