August 6, 2025

Present

Armstrong Foundjem, Amit Ruhela, Ben Hawks, Gary Mazzaferro, Geoffrey Fox, Gregg Barrett, Marco Colombo, Matt Sinclair, Nhan Tran, Philip Haris, Piotr Luszczek, Satoshi Iwata, Shirley Moore, Victor Lu

Tentative Agenda

Any New Members Introduction
Continuing discussion of New Benchmarks and the catalog of Science benchmarks based on MLCommons Science/HPC Benchmarks Overview - Taxonomy
Time Series Dataset Catalog of 950 Datasets from 80 sources Building a Dynamic Catalog/Review illustrated with Time Series Aug 5 2025 COMBINED2
White Papers
The Benchmark carpentry white paper https://www.overleaf.com/9828764221czxzxxcxmcrr#1f1c84
New white paper on Science Benchmarks
Any Other Business

Google Meet Notes

MLC Science WG - 2025/08/06 07:54 PDT - Notes by Gemini
Geoffrey Fox encountered technical difficulties, but later presented his work on using LLMs to catalog time series datasets, highlighting Google Gemini's ability to generate Python code for data extraction and analysis, despite its debugging limitations. Gregor von Laszewski provided an update on the paper, discussing the formalization of benchmarks, the distinction between live and static datasets, and the contributions of students, while Nhan Tran proposed streamlining the paper's length. Piotr Luszczek reviewed the HPC and scalable benchmarks content, emphasizing the need for proper academic references, and Shirley Moore presented her contributions to the profiling and performance analysis section. The participants discussed the need for further collaboration and review of the paper, with a focus on defining "benchmark carpentry" and incorporating relevant resources.

New member

Peter Fulle CEO of Neuramorphic AI Peter Fulle - Crunchbase Person Profile Peter Fulle - TEDxVitacura | LinkedIn Neuratek

Geoffrey’s Talk

Time Series Models Summary Table
COMBINED2 includes catalog of 750-950 datasets
Presentation Building a Dynamic Catalog/Review illustrated with Time Series Aug 5 2025