bcatanzaro

Reminds me of “peek” and “poke” on my C64 as a kid. Magic things happen when reading and writing to magic addresses…

Tesla isn’t interested in building cars. They canceled SX in order to build the Cybertruck. I expect they’ll let 3Y just get really old.

Instead of S3XY cars we get stunts. Solar roofs, Concept car roadsters, and of course humanoid robots.

If only Tesla would be interested in cars again.

OMG

I agree with you

“I think at the heart of this issue is that humans have two competing natures: a tendency to compete and a capacity to appreciate beauty,” said Kai Shaikh, a graduate student in mathematics at the University of Toronto. “To me this seems to be a case of the former attempting to strangle the latter.” [1]

Beauty can be appreciated even when it is vast, even when it is beyond one's comprehension. I don't think this release should be primarily viewed as an outcome of competition. Instead it is revealing truths about the universe that were always there and always beautiful, even if we hadn't seen them yet. I believe there are infinitely more such beautiful truths currently hidden and waiting for us to discover.

[1] https://www.nytimes.com/2026/10/06/science/openai-math-probl...

Integration cost is very important to DLSS.

More information is better, but more games is also better.

That’s why we cook food. Cooked grain is way more nutritious than raw grain. The kitchen is an external stomach.

I wore a glucose monitor for a few months and tested my blood sugar response to different foods. The worst spike I ever got was American Chinese food: rice+sugary sauce. Brown rice vs white rice makes no difference.

Ice cream didn’t make much of a spike - all that fat. Waffle with maple syrup and peanut butter is completely different profile than waffle without peanut butter.

Most important thing for my blood sugar was portion size. Eat a lot, get a big spike.

Frustrating to me that people argue so much about “processed” vs “unprocessed” food. The real issue I believe is portion sizes.

The Nemotron 3 family uses Mamba-2 because that is what we had validated when we started building the family.

Changing the SSM backbone is a major revision, not a dot release. At least for us.

Something poignant about the knowledge in this paper staying lost behind a paywall.

"A founder-engineer driven company. Facebook is one of the few Big Tech firms whose founder is an engineer, and still is the CEO. Netflix is the other one where founder and co-CEO Reed Hastings was also a software engineer before starting the company. Amazon was the other example of this until recently, but it’s not the case at Google or Apple."

So funny how people overlook Jensen Huang repeatedly. As if NVIDIA wasn't big tech, or Jensen wasn't a founder, or an engineer...

I have been in an SF muni car with someone actively smoking something that was not marijuana or a cigarette. I think it was fentanyl but I’m not sure. This was about two years ago. I hope this is tolerated less now than it has been.

Underrated comment

OMG

Yes

Nemotron has two reasons to exist, both of which are strategic to NVIDIA.

1. Help NVIDIA design future systems for AI by more deeply understanding what it takes to build AI.

2. Keep the AI ecosystem strong and diverse throughout the world by providing AI infrastructure that many companies can innovate on.

This is not a science project, nor is it for the joy of giving something away. Both of these reasons are core to NVIDIA.

Some of these concerns are precisely why we are building Nemotron at NVIDIA. We want to make it possible for people to invent and deploy AI in all sorts of new and unforeseen ways.

Nemotron is:

1. Not just open weight, but open data (to the limits of what is feasible), open recipe, open technique

2. In the future built by a coalition of organizations coming together to build great openly developed AI.

Nemotron 3 Super is our most successful model yet. [1]

Ultra is coming soon. And then Nemotron 4.

We can afford to do this because when AI grows, NVIDIA's opportunity also grows.

[1] https://kaitchup.substack.com/p/the-fastest-and-cheapest-120...

AI is a mirror. If you are boring, you will use AI in a boring way.

sadly, disembodied brains are not very useful. embodied brains require a civilization's worth of energy consumption and environmental impact in order to do their work. so we really need to take the world's power/water/carbon impact (divided by the world population) to talk about how much power it takes for a human brain to solve a problem.

As I said - anything surprisingly gluten free or surprisingly vegan is going to be UPF.

Sometimes I wonder if the gluten free trend is a ploy by food processing companies to whitewash expensive proprietary processed foods as “whole”.

Agreed. Which makes the statement “ultraprocessed foods should be treated more like cigarettes” seem wrong.

I guess “bad foods should be treated more like cigarettes” is too obvious.

Most of the things on sale at “Whole Foods” are ultra-processed these days. Anything that requires effort to make gluten free or vegan for example. Like impossible burger. Extreme ultra-processed. Or gluten free bread.

Please don’t tell me impossible burger patties are like cigarettes.

The Nano model isn’t pretrained in FP4, only Super and Ultra are. And posttraining is not in FP4, so the posttrained weights of these models are not native FP4.

I had a similar experience. The "Sanctus Sanctus Sanctus" written in stone was not an aspiration or a motto for me when I visited, it was a statement of fact.

I've often wondered if Intel would still be a dominant force in computing if it had kept engineering in Silicon Valley. I worked at Intel both in Hillsboro and in Santa Clara and I feel that Intel's decision to put so much engineering in Oregon was done to insulate them from the pressures of Silicon Valley. They didn't have to pay very well, and they had a very insular culture - because they could afford it. They didn't have to work very hard to keep engineers at Intel because their engineers were basically trapped in beautiful Oregon and generally wouldn't consider moving back to expensive California.

Housing costs in the Bay Area are soul-crushing, but they do motivate people to work on the highest value projects because complacency just doesn't usually work if you're trying to buy a house. And so I wonder, if Intel had kept their workforce mostly in California, could they have stayed a dominant force in computing?

Yes, but Italy had to centralize its language in order to accomplish this. 1000 Italian dialects were suppressed in a very heavyweight process. (And probably some people didn't like speaking Florentine, which became modern Italian.)

English is complicated because it's decentralized and there is no authority to regularize it. Which is a feature, not a bug.

English pronunciation does vary quite widely and it would be difficult to rewrite all the books and websites into all the different accents.

It's also decentralized - there's no authority to tell the English-speaking community how to spell things or how to say things.

I think these are both advantages that outweigh the phonetic inconsistency.

I'm curious about the False Positive/False Negative rate of the battitore. Do they open up some wheels to double check?

It means “envy” in Latin and Spanish etc. Which is why the logo is an acid green evil eye. And why NVIDIA’s headquarters are two buildings named “ENdeavor” and “Voyager”.

I almost want to read it as satire. Especially juxtaposed against his death. Because the ideas of "What I cannot create, I do not understand." and "Know how to solve every problem that has been solved" seem profoundly unwise and endlessly futile.