Thank you Ugur Kaynar (@DellTech) and Maroon Ayoub (@RedHat) for “Scaling AI Services: A Customer’s Journey to Production Inference” at PyTorch, Powering the Enterprise, with llm-d as the lens. #llmd#PyTorch
Thank you Tyler Michael Smith (@RedHat), vLLM core maintainer, for “The Economics of Serving AI: What Business Leaders Should Know About vLLM” at PyTorch, Powering the Enterprise. @vllm_project#vLLM
Thank you Stephen Watt (@RedHat, PyTorch Board) for opening PyTorch, Powering the Enterprise, and Joseph Groenenboom (@RedHat) and Chris Hoge (@nvidia) for “From Models to Business Value: Navigating the PyTorch Ecosystem.” @PyTorch
Thank you @RedHat, @nvidia and @IBM for hosting PyTorch, Powering the Enterprise in San Jose, and to hosts Steven Pousty, @wescpy, @cybette, Priya Sethuraman and Saša Zelenović. Our custom networks train in @PyTorch, so this is the stack we build on. #PyTorch
Thanks Rodrigo Silva Ferreira, Pallav Anand and Josh Schneider of @pytexas for speaking at @py_bay, on scientific observatories, ML explainability at scale, and mentoring.
Thanks to the maintainers who spoke at @py_bay: Anthony Sottile of @pre_commit_ci, Brian Okken, @mkennedy of @TalkPython, and Jim Baker.
The tools we use every day are kept working by people like these.
Every finding should trace back to the photo it came from.
Replay the real flight over the 3D model, with the photo taken at each point shown alongside. 100% RTK fixed, 0.69 cm per pixel.
https://t.co/IlShVIKv34
Thanks Christopher Ariza of @RA_Insights for "Getting Started with Free-Threaded Python Using NumPy" at @py_bay.
NumPy sits under every model we train.
Thanks Felipe Perez of @OpenAI for "Testing the Nondeterministic: Making Safe Changes in your Agent Harness" at @py_bay.
Changing an agent without breaking what works is the question we think about most.
Thanks Emma Smith of @nvidia for "The Path to Introducing Rust in CPython" at @py_bay.
Changes at the bottom of the stack reach every Python program above it.
Same flight, five views: photo, NDVI, chlorophyll, height, zones.
The useful output is the short list: 7 of 75 trees whose chlorophyll and vigour sit well below the rest, so someone checks those first.
Live demo, public dataset: https://t.co/IlShVIJXdw
Thanks @DtiPhilippines PTIC-Silicon Valley for Filipinos at the Frontier during SF Tech Week.
Fireside chat: Earl Valencia (National Innovation Council), Andy Rapista (Invictus), Kyle Kua (@theworldlabs). Thanks also to @PHLinSF, @WellsFargo and @TourismPHL.
Thanks Bernát Gábor of @TechAtBloomberg for "How to be a Better Software Engineer in the World of AI Agents" at @py_bay, and for maintaining tox, virtualenv and pipx.