TPC25 is a wrap, with 360 participants from 14 countries. Next up is the 15-August submission deadline for the TPC workshop at SC25, “Frontiers in Generative AI for HPC Science and Engineering: Foundations, Challenges, and Opportunities.

Call for Papers for TPC Workshop at SC25 

Nearly 200 people attended the half-day TPC workshop at SC24, and this year our team was approved for a full-day tutorial. Submissions will be included in the SC proceedings. Please see the call for papers. Papers will follow SC guidelines with respect to length and format. Accepted papers will be published in the SC25 proceedings. 

Don’t miss the 15 August deadline to submit a paper!

TPC25 By the Numbers

Thank you to the dozens of collaborators who planned and organized TPC25, and to the 360 participants from 14 countries who traveled to San Jose, California for the four-day event. The TPC25 program was organized around the three goals driving TPC:

  • Building an Open Community. Thirty invited speakers from 21 organizations provided a plenary program highlighting initiatives and insights in AI for Science from around the world. (AI2, AIST, AWS, Argonne, BSC, FutureHouse, Google DeepMind, Hammerspace, Institute of Science Tokyo, Hyperion, Intel, Intersect360, Microsoft, ORNL, Quantinuum, RIKEN, Stanford, University of Chicago, University of Michigan, UNSW, US DOE)
  • Identifying, Incubating, and Facilitating Collaboration. Eighteen working groups and birds-of-a-feather groups filled 30 parallel breakout sessions, each a 90-minute combination of lightning talks and discussions seeking to identify potential collaborative efforts and provide updates on existing collaborations. These breakouts hosted 139 lightning talks from 52 organizations. (AIST, AMD, Argonne, Arizona State U, Articul8.ai, AWS, BNL, BSC, Cerebras, CSC, d-Matrix, FutureHouse, George Mason University, Google DeepMind, HCLTech, INAF, INL, LANL, LBNL, LLNL, Meta, Microsoft, NCHC, NCSA, NREL, NVIDIA, OpenMined, Ohio State University, Oregon State University, ORNL, PNNL, PPPL, Princeton, PSC, RIKEN, Rutgers, Sambanova, SNL, SLAC, St. Jude Children’s Research Hospital, Stanford, Stony Brook University, TACC, University of Utah, UC Berkeley, UC Irvine, UC Merced, University of Chicago, UCL, UIC, UIUC, University of Tennessee, University of Texas)
  • Creating a Global Network of Expertise and Resources. TPC25 kicked off with nearly 150 participants, who selected among three tutorials and a 1.5 day hackathon. The tutorials included “AI for Science: Foundations and Frontiers,” “Evaluation of AI Model Scientific Reasoning Skills,” and “Using AI to Accelerate Day-to-Day Scientific Productivity.” The hackathon, “Building Agentic Systems for Science,” convened eight project teams (mixing experts with novices) to learn by doing. Eight participants volunteered real projects to develop and/or enhance, equipping participants to contribute to TPC’s quarterly hackathons. Participants from Argonne, BSC, CINECA, Google, ORNL, and RIKEN organized and facilitated the tutorials and hackathon.

TPC25 also established a foundation of partnerships with industry and vendors that will strengthen the community going forward. The event was created and managed through a partnership with TCI Media (publisher of AIWire, HPCWire, bigDataWire, QCWire), whose team managed all logistics for TPC25 and launched a new TPC all-hands sponsorship program. Our Foundation Sponsor was AWS; Intel and Quantinuum joined as Platinum Sponsors. Eight companies joined as Corporate Sponsors: Articul8, Google Cloud, Hammerspace, HPE, InspreSemi, NVIDIA, ParallelWorks, and Sambanova. The HPC-AI Society joined as a non-profit sponsor. Finally, institutions with 10 or more participants were designated as Institutional Sponsors. These were Argonne National Laboratory, Lawrence Berkeley National Laboratory, Institute of Science Tokyo, Oak Ridge National Laboratory, RIKEN Center for Computational Sciences, and The University of Chicago.

Trending

Discover more from Trillion Parameter Consortium (TPC)

Subscribe now to keep reading and get access to the full archive.

Continue reading