Maxirex

@Yuchenj_UW Isnt Opus too Heavy for that question

Bro, Fable 5 won’t even answer “What does the heart do?” We’ve reached the point where a middle-school biology question can’t pass the safeguard.

I don't like exceptions. They're coarse, and rough, and irritating, and get everywhere. Not like here. Here everything is value...

@ThePrimeagen I am 120 Frame per second

Here's your monthly reminder to stop typing at 60 words per minute. If you're not typing at 120 words per minute, you are ngmi

@ChShersh I have seen many people get in the trap of integer overflow

I just realised, I’m so young, I can fit my age into a single uint8_t

@denicmarko Alibaba Customer service

What LLM do you prefer for coding?

@BigImpactHumans At Monaco. Cant see you from here

Hotdogs and iced coffee by the river

@ChShersh It can mean crying - Laughing - Not agreeing - agreeing - Inspired - Not Inspired . All at the same time. Its some modern Morse code bs

⬛️😭😭⬛️⬛️⬛️⬛️⬛️⬛️⬛️ 😭⬛️⬛️⬛️😭⬛️⬛️⬛️😭⬛️ 😭⬛️⬛️😭😭😭⬛️😭😭😭 😭⬛️⬛️⬛️😭⬛️⬛️⬛️😭⬛️ ⬛️😭😭⬛️⬛️⬛️⬛️⬛️⬛️⬛️

@ArthurMacwaters They are called "Space"X for a legitimate reason

SpaceX has put more payloads in orbit than the rest of the world combined Not another company Not another country The entire world. Combined. And this is just the beginning of a spacefaring future.

D

I am getting Smallville Edits all over my Instagram feed and i am not even complaining about it.

@IanCutress Coughed hard

The problem with getting any of the group 1 or group 2 metals is that they're too basic.

@CS11__ First Rule of SpyCraft : Do the Unexpected

‘Wave Of The Future’ is just another absolute masterclass of a mission. The way 007 First Light paces it’s chapters and the structure of them is genuinely outstanding. Games a blast and i adore every moment of it. What’s new? Definitely finishing this tomorrow | ‘FOR ENGLAND’

@ChShersh ‘So a box of apples is a box of apples?’ ‘No Dave, a box of apples is actually a bag of small red pears.’

"So std::vector<bool> is broken, and I shouldn't use it?" "Yes Dave" "And a dynamic array of booleans is std::vector<uint8_t>?" "That's correct Dave"

@ylecun @JohnProperBTC @Xnarkycritic @DavidSacks They’re dropping all their cash into biomedicine to stop aging, meanwhile the physics department is out here sharing a single roll of duct tape.

Some billionaires (or their foundations) do fund certain areas of basic research. Examples: Simons Foundation, Moore Foundation, Sloan Foundation, Keck Foundation, Schmidt Sciences, and several others. But most of them are in biomedicine: Howard Hughes Medical Institute, Chan-Zuckerberg Initiative, Wellcome Trust, and many others. However, this is all small compared with federal funding: around 25% of academic funding is from philanthropy, and over 50% from federal grants. This proportion is much lower in STEM than in biomedicine, and much lower in less prestigious universities. So historical evidence that fundamental research can be helped, but not entirely funded, by philanthropic billionaires.

Many have reached out to me regarding the new Cyber EO. A few thoughts: First, President Trump is the most pro-innovation president we’ve ever had. He’s made it clear that the U.S. has to win the AI race and that a pro-innovation, pro-energy, and pro-infrastructure policy is the way to do that. Thanks to President Trump, AI will generate over a 2% tailwind to GDP growth this year, with hundreds of thousands of new construction jobs and 25-30% wage increases for blue collar workers. President Trump is the reason that we have an AI boom happening right now. The change in the EO from a 90 day to 30 day period is a game changer because it allows our AI labs to comply with the voluntary framework without delaying new model releases. They can synchronize their efforts under the EO with other pre-release activities. Furthermore, I’ve been advised by the lawyers who draft EOs that 30 days means calendar days, not business days. In the AI race, every day counts. As OSTP well notes, “The EO creates a process for frontier labs to voluntarily share cutting-edge cyber models in order to secure critical infrastructure and strengthen the government’s own cyber defenses. We are NOT conducting oversight of all new models, as that level of government overreach would have chilling effects on free speech and innovation.” OSTP’s characterization is completely consistent with the discussions that I have participated in, where it was agreed that the EO is intended to apply only to models that represent a meaningful step-change in cyber capabilities (eg Mythos), not to incremental version numbers of existing models (eg Opus 4.7 -> 4.8). Finally, I understand the concerns of many that this could morph into an “FDA for AI”. Of course bureaucratic mission creep is always a danger and this should be closely monitored. But the EO expressly forbids the creation of a new licensing, preclearance, or permitting regime. Most importantly, I do not believe that President Trump would allow this to happen. As AI presents new policy challenges (such as cyberweapons), I believe that everyone in the administration is working diligently to navigate the issues with the American people in mind. I look forward to working with the Treasury, NSA, ONCD and others on the implementation of this framework.

@ChShersh Its always the case. You are never asked about what you know.

I’ve been programming for 16 years. I thought I’ve seen it all. Yet, at my current job, I have to do things I’ve never done before in my life.

@CS11__ Not unrelated. She's the Plot

007 First Light has my vote for Game Of The Year. (Picture unrelated)

I dont care about GTA6 but this is allready the leading contender for GOTY. Yesterday I played till the early training part where he does shooting drills and drive aston and then fight with Monroe and all then goes to That Apartment. It gave the actuall chills of an 007 Agent.

I can’t express how much of a delightful experience i’m having with 007 First Light. Every single moment of this game is pure joy. Fairly certain i’m approaching the later stages, sad to see it end but excited nonetheless.

@beffjezos Hahaha Hail Marry

Gotta love the irony of the company whose whol staff is part of a school of thought obesssed with the dangers of recursively self-improving intelligence actually creating recursive self-improving intelligence when nobody directly asked for it. The power of memetic hyperstition.

@AndrewYNg @RedHat @cedricclyburn I see the same thing almost everywhere

New course on serving LLMs efficiently -- how do you serve models to many concurrent users at low latency and reasonable cost? This short course is built with @RedHat and taught by @cedricclyburn. Efficient LLM serving requires efficient memory management. A 70B-parameter model takes ~140 GB just to load the weights. On top of that, every active request needs its own chunk of GPU memory, the KV cache, to store the token context it has built up so far. In this course, you'll learn to reduce a model's memory footprint with quantization and serve it using vLLM, which handles many concurrent requests efficiently through smart memory management. Skills you'll gain: - Quantize a model and measure the accuracy tradeoff - Serve a model with vLLM and watch it handle concurrent requests efficiently - Benchmark your deployment and make informed tradeoffs between speed, cost, and accuracy Join and learn to serve LLMs efficiently: deeplearning.ai/courses/fast-a…