In 2022, Berkeley PhD student Woosuk Kwon thought he could make Meta's new open-weight model, OPT-175B, run faster in two weeks.
Four years and 3,000 contributors later, that project is @vllm_project, one of the largest open-source inference engines, running on ~500,000 GPUs at any given time.
He and fellow researchers Simon Mo, Kaichao You, and Roger Wang borrowed "virtual memory" from operating systems, handing out memory only as it's needed so each GPU can juggle requests of wildly different sizes. (The "v" in vLLM is for virtual.)
Their first user, Berkeley neighbor @arena, cut its GPU needs in half. Today Google, Amazon, and Meta all run vLLM in production.
In January, the core maintainers founded @inferact and raised $150M to give vLLM a full-time team.
They're hiring across eng, marketing, and devrel in SF and Singapore: a16zjobs.substack.com/p/infe…
@simon_mo_ @woosuk_k @KaichaoYou @rogerw0108
F.D.Grillo retweeted
Ken Griffin’s historic gift of more than $3 billion will strengthen Carnegie Mellon University in Pittsburgh and establish Carnegie Mellon University Miami, advancing education, discovery and opportunity. #CMUMiami
Read more: cmu.is/griffin
F.D.Grillo retweeted
Announcing a solution to the Navier-Stokes problem, one of the seven Millenium Prize problem.
Represents a major milestone for both AI and mathematics, as well as pointing to an upcoming renaissance in scientific discovery.
Hoping that the knowledge created by this model can help the community better understand non-linear PDEs, which are used to model much of the physical world.
We’re sharing a solution to the Navier-Stokes Millennium Prize Problem, one of the deepest problems at the frontier of mathematics.
The proof was produced by a group of agents, using an OpenAI next-generation model significantly more capable than GPT-6 Astra.
The problem concerns whether the description of smooth three-dimensional fluid motion modeled by the Navier-Stokes equations can break down. It has remained unresolved for roughly 90 years.
F.D.Grillo retweeted
The human brain is astonishingly efficient. It runs on about 20 watts, roughly the power required to keep a dim bulb lit. Why can’t computers do the same?
Neuromorphic computing tries to answer that question by building hardware inspired by how networks of synapses process information. While this approach can reduce energy use, it often comes with a tradeoff in computing speed.
At SFI, a recent working group brought together experts in neuromorphic computing and stochastic thermodynamics to learn from one another and exchange ideas. By connecting device design with the underlying physics of computation, participants began to explore new ways to think about energy, speed, and randomness in computing systems.
santafe.edu/news-center/news…
F.D.Grillo retweeted
Sincerity used to be the default. Movies committed to their emotions and trusted the audience to meet them there. That belief made them feel real. So many films now hedge with irony or tonal whiplash, afraid to stand still and mean something.
If government must restrict an industry because of its scale & national impact, it also bears the responsibility to ensure those restrictions do not erase the competitive pressures that keep industry power in check — though they naturally will
1
2
118
This is not a plea for new rules born of panic or whim. It is the responsibility of a government that has already regulated—whether wisely or unwisely—to repair the competitive pressures its own actions have weakened
1
1
70
F.D.Grillo retweeted
"Slowly, I realized realism in cinema is not the same as realism in life. Cinema has its own realism. The world in cinema is not the real world. It has been crafted. That makes cinema interesting. It's not real. It's closer to dreams. If you treat life as a dream, you can understand this."
Tsai Ming-liang
Thank you Frank.
Man’s search for justice, mercy, & truth are inscribed in the template of your life — a life of persevering devotion in the pursuit of meaning.
A tree bearing eternal fruit in the lives whom shared the shade of a truly Great Man.
A memory for blessing, always.
2
235
F.D.Grillo retweeted
just accidentally closed a tab i've had been meaning to read for the past 2 years
An experiment is not a single action in perpetuity without question— but a formula by which subsequent action can be questioned and adjusted through trial & error.
It is not a matter of blind religiosity, but requiring of us incessant review and self-governance.
These actions will either deny or support the underlying belief.
Democracy then, as framed by the Constitution, requires the people by whom it was established and for whom it will benefit.
Surely then, if a nation fails, it cannot fail by the handful of its elected officials.
If it fails, the people will have lost their ability to self-govern.
They will have begun to demand of their leaders what they are yet unwilling to do for themselves- for their communities.
They will have become crippled by the very men & women they’ve elected to office in exchange for the spellbinding affirmation of broad decree.
The silver bullet solution they trade a nation’s soul to acquire, will become the nail sealing the coffin which cradles that nation’s fate.
All that remains unanswered is by what relative cultural, social, and/or economic mechanism did this prevail, and why?
2
264































