From HPCwire, October 27:
CERN, the world’s largest particle physics laboratory, has spent decades collecting enormous amounts of data from particle collisions. With the recent rapid growth in AI capabilities, CERN wants to develop a foundation model that can make better use of all that data.
To help it achieve this goal, the European physics laboratory has teamed up with French AI company Mistral. The two have signed a statement of intent to explore the development of foundation models for scientific research.
CERN is already using AI for particle physics. With Mistral, CERN wants to move from multiple specialized AI models to a more general foundation model that could perform several scientific tasks. Their first project is a proposed Large Physics Model, reportedly called Flare, that would be trained on billions of simulated collisions from the Large Hadron Collider (LHC).
How much data has LHC collected? According to some estimates, CERN currently holds around one exabyte of experimental data from the LHC. That is equivalent to one million terabytes. And that amount is expected to grow significantly. An upgrade to the LHC around 2030 is expected to increase the number of collisions tenfold.
The new foundation model with Mistral could handle several of these tasks together. It could help reconstruct collision events and identify particles, reducing the need for separate algorithms.
What’s in it for Mistral? The partnership provides access to one of the world’s largest scientific datasets and an opportunity to expand its work into scientific research....
....MUCH MORE
I was sort of hoping for next-gen riot-dispersing sound cannons but this could get interesting too.
