Skip to content

What is Aurora GPT, the AI ​​focused on collecting scientific knowledge, like?

The American Argonne National Laboratory (ANL) is working together with Intel and HP to create a model of Artificial intelligence (Generative AI), referred to as Aurora GPT, focused on storing scientific knowledge with more than a trillion parameters.

The capabilities of supercomputers are currently used for solving large problems, however, the ANL aims to provide “new opportunities to advance science and engineering”.

LOOK: Why does Bill Gates believe that AI will allow us to have a three-day work week?

To do this, the American laboratory Argonne is working on training the Aurora GPT AI, together with Intel and HP. It is an AI model that is being trained on the Aurora supercomputer – one of the most powerful computers currently – under the concept of storing all possible scientific knowledge.

In turn, when the AI ​​is ready, It will have a ‘chatbot’ interface in which researchers will be able to ask questions to facilitate access to knowledge in different areas, whether biology, physics or mathematics and engineering.

Specifically, as the ANL has detailed on its website, it will allow researchers to run a greater diversity of workloads, including machine learning and data-intensive tasks, as well as “traditional simulation and modeling campaigns”.

With this new project, it is also intended to expand capabilities with exascale computing, that is, with the capacity to perform a minimum of one exaflop operation per second. Exaflop operations have the performance to compute at least one trillion floating point operations per second. In this way, research times can be accelerated.

This exascale power is powered by Intelwith a GPU architecture composed of six Intel Max Series GPUs 128GB each. Likewise, it integrates two Intel Xeon CPU Max Series processors of 64 GB each.

At the moment Aurora GPT is in the early stages of development, so, as explained, its capacity is currently limited to 256 nodes, however, it will be expanded in the future to the 10,000 nodes of capacity available on the Aurora supercomputer. .

With all this, the laboratory has stressed that Argonne developers continue working to carry out a wide range of activities to “prepare Aurora for science from day one.”

Source: Elcomercio

Share this article:
globalhappenings news.jpg
most popular