Advertisement · 728 × 90

Posts by ST010-14

Post image

railroads were expensive

from finn on twitter

3 days ago 67 8 5 1
Video

Hey Ford, GM, Stellantis, and maybe Tesla: Jensen says you’re losers.

5 days ago 9 1 2 0
Comic. One... two... THREE! [text in red] X Deprecated [line break] One... two... three... GO! [text in red] Deprecated [red curly bracket around top two lines] Too easy to mix up [line break] Three... two... one... GO! [text in green] [check mark] ISO Standard [caption] If I were in charge of ISO, the first thing I’d do would be to standardize the way people count out loud before doing something in sync.

Comic. One... two... THREE! [text in red] X Deprecated [line break] One... two... three... GO! [text in red] Deprecated [red curly bracket around top two lines] Too easy to mix up [line break] Three... two... one... GO! [text in green] [check mark] ISO Standard [caption] If I were in charge of ISO, the first thing I’d do would be to standardize the way people count out loud before doing something in sync.

Countdown Standard

xkcd.com/3232/

5 days ago 5391 866 102 61
Post image Post image Post image Post image

A NASA technical paper on Gas-Core Nuclear Rockets shows their Isp scaling to 6500s if they can retain the uranium plasma at 2000 bar:
ntrs.nasa.gov/citations/19...
At the 12 GW scale, they'd reach 58 kW/kg. 7.5% of reactor output becomes waste heat at 1400K to be removed by radiators.

5 days ago 17 3 0 0

Pi tree!

6 days ago 105 7 3 0
Preview
Course | RLHF Book by Nathan Lambert Course lectures and talks on RLHF and post-training.

Excited to launch the accompanying free RLHF Course for my book. To kick it off, I've released:
- Welcome video
- Lecture 1: Overview of RLHF & Post-training
- Lecture 2: IFT, Reward Models, Rejection Sampling
- Lecture 3: RL Math
- Lecture 4: RL Implementation
Landing page: rlhfbook.com/course

6 days ago 76 21 6 1
Post image

Turing Award recipient Sir Charles Antony Richard Hoare died March 5, 2026. His 1980 ACM A.M. Turing Award cited “fundamental contributions to the definition and design of programming languages,” but that is only part of what he contributed during a career spanning six decades. buff.ly/kba0bXb

6 days ago 9 7 1 2

Enjoying the For You feed? Give it a like ♡ to help more people discover it: bsky.app/profile/did:...

The more people use it -> the more feedback we get -> the better we can make it for you.

9 months ago 18419 3068 174 499

"You sound like an LLM."

Sweetie, LLMs sound like *me.*

1 week ago 67 8 4 2
Post image

Picture of the century. 😍

📷 credit: Science Tube

1 week ago 30818 4536 919 335
Advertisement
Preview
The inevitable need for an open model consortium And yes, I hate consortia too.

In 5-10 years, as models get more expensive & capable, I see the funding structures and support for open models breaking down. We need to consider if we need other options of supporting the open ecosystem.

The inevitable need for an open model consortium
www.interconnects.ai/p/the-inevit...

1 week ago 46 11 3 1

this guy's read the mythos system card

1 week ago 64 5 2 0
<Animated.View ref={console.log} />

<Animated.View ref={console.log} />

left: bsky.app with an element highlighted. right: browser console with a html element logged

left: bsky.app with an element highlighted. right: browser console with a html element logged

pro react tip: use `ref={console.log}` to see where your element has gone

1 week ago 82 3 5 1
Post image

Water use skeptics, explain this

1 week ago 100 3 2 0
A McDonald’s where the sign has been reduced to McDo.

A McDonald’s where the sign has been reduced to McDo.

There is no McTry.

1 week ago 22343 3767 371 204

Saw a reel making fun of people who were expressing skepticism of the #Artemis II images in Instagram comments and while it's true that a lot of folks are just spreading conspiracy theories, there are misconceptions that can just be addressed directly!

1/🧵

1 week ago 558 156 16 9
Post image

He does have a point.

1 week ago 24 3 2 1

FAKE-ASS LEAK UNDERESTIMATED SCORES BY 10+%

1 week ago 57 2 2 1
Post image

People are using AI to break CUDA’s moat.

Given any pytorch model, it profiles it, ranks bottlenecks by amdahl's law, writes triton or CUDA C++ replacements, and runs 300+ experiments overnight with no human in the loop.

- 5.29x over pytorch eager on rmsnorm
- 2.82x on softmax

2 weeks ago 110 14 1 2
Comic. [Sign above four people. A person sits at a desk working on a laptop. A person with ponytail is talking to person with white hat. A person with short hair walks away.] SIGN: It has been [-0.00000000000000044] days since our last floating point error

Comic. [Sign above four people. A person sits at a desk working on a laptop. A person with ponytail is talking to person with white hat. A person with short hair walks away.] SIGN: It has been [-0.00000000000000044] days since our last floating point error

Day Counter

xkcd.com/3228/

2 weeks ago 5955 793 24 25
Advertisement
Post image

Earthset from Artemis II. I am overcome.

2 weeks ago 6272 1569 51 162
Earthset captured through the Orion spacecraft window at 6:41 p.m. EDT, April 6, 2026, during the Artemis II crew’s flyby of the Moon. A muted blue Earth with bright white clouds sets behind the cratered lunar surface. The dark portion of Earth is experiencing nighttime. On Earth’s day side, swirling clouds are visible over the Australia and Oceania region.  In the foreground, Ohm crater has terraced edges and a flat floor interrupted by central peaks. Central peaks form in complex craters when the lunar surface, liquefied on impact, splashes upwards during the crater’s formation.
[alt text from NASA]

Earthset captured through the Orion spacecraft window at 6:41 p.m. EDT, April 6, 2026, during the Artemis II crew’s flyby of the Moon. A muted blue Earth with bright white clouds sets behind the cratered lunar surface. The dark portion of Earth is experiencing nighttime. On Earth’s day side, swirling clouds are visible over the Australia and Oceania region. In the foreground, Ohm crater has terraced edges and a flat floor interrupted by central peaks. Central peaks form in complex craters when the lunar surface, liquefied on impact, splashes upwards during the crater’s formation. [alt text from NASA]

#Artemis II lunar flyby images are showing up! 😍

A crescent Earth setting behind the Moon.

1 week ago 3820 888 35 73
Preview
AI World Clocks The current time as rendered by 9 different AI models. By Brian Moore.

Was reminded earlier that this exists and is wonderful.

clocks.brianmoore.com

2 weeks ago 60 11 3 0
the value of a performance oracle — wingolog wingolog: article: the value of a performance oracle

new bloggery, in which i am nerd-sniped by @mattkeeter.com into noodling on webassembly, tail calls, and performance wingolog.org/archives/202...

2 weeks ago 14 2 2 0

We've signed an agreement with Google and Broadcom for multiple gigawatts of next-generation TPU capacity, coming online starting in 2027, to train and serve frontier Claude models.

2 weeks ago 52 1 1 3

#Intel released the 91st edition of the Software Developer’s Manuals with canonized #AMX_COMPLEX.
All-in-One:
cdrdv2-public.intel.com/916575/32546...
Changes v83:
cdrdv2-public.intel.com/916574/25204...

1 week ago 1 1 0 1

claude costs $20/mo but attaching an agent harness to the chipotle customer service endpoint is free

1 month ago 630 97 10 17
How much do you know about late 90s/early 2000s computing and technology?

How much do you know about late 90s/early 2000s computing and technology?

i have basically spent my entire life waiting to receive this particular text message

3 weeks ago 342 30 11 3
Advertisement

I’m gonzo for AI and this is a good idea. Wikipedia belongs to a deeper pace layer. Critical infrastructure. Move slow, edit things.

3 weeks ago 81 11 3 1
Preview
Qworld: Question-Specific Evaluation Criteria for LLMs ArXiv link for Qworld: Question-Specific Evaluation Criteria for LLMs

This paper introduces Qworld, generating question-specific evaluation criteria for large language models. Using a recursive expansion tree, Qworld uncovers nuanced LLM capabilities often missed by standard metrics, improving evaluations in healthcare and reasoning. https://arxiv.org/abs/2603.23522

3 weeks ago 3 1 1 0