@𝑀𝑒𝓂𝑒𝓈𝒟𝑒𝓁𝑀𝒶𝓇 (@MemesDelMar) header
@𝑀𝑒𝓂𝑒𝓈𝒟𝑒𝓁𝑀𝒶𝓇 (@MemesDelMar) profile picture

@cybertruck flexer | @TomochiWorld enjoyer | pfp by @jessicakomori

Starbase, TX
Joined January 2025
Photo by @𝑀𝑒𝓂𝑒𝓈𝒟𝑒𝓁𝑀𝒶𝓇 (@MemesDelMar)Photo by @𝑀𝑒𝓂𝑒𝓈𝒟𝑒𝓁𝑀𝒶𝓇 (@MemesDelMar)Photo by @𝑀𝑒𝓂𝑒𝓈𝒟𝑒𝓁𝑀𝒶𝓇 (@MemesDelMar)Photo by @𝑀𝑒𝓂𝑒𝓈𝒟𝑒𝓁𝑀𝒶𝓇 (@MemesDelMar)
@𝑀𝑒𝓂𝑒𝓈𝒟𝑒𝓁𝑀𝒶𝓇 retweeted
The reason late for work..🦉🐾😊
137
623
24
6,818
88,773
@𝑀𝑒𝓂𝑒𝓈𝒟𝑒𝓁𝑀𝒶𝓇 retweeted
AI interfaces nowadays are still pretty much turn-based. In my PhD thesis, I thought about what interaction would look like if we could go beyond taking turns. One example is intra-turn interaction:
301
614
217
8,520
792,603
@𝑀𝑒𝓂𝑒𝓈𝒟𝑒𝓁𝑀𝒶𝓇 retweeted
Will DePue (@willdepue), a former OpenAI employee who worked on the video model Sora, says he can tell which chip was used to make an AI model just by looking at the model. Four of China's top AI labs used a chip that the US won't sell to China for their most recent models, according to DePue. Before a model begins training, it gets filled with billions of random starting numbers. Those numbers are generated from a "seed": the same seed will produce the same numbers. The chip that does this divides the work among 2,048 workers for every processing unit it has. Nvidia's older H100 chip has 132 processing units; its newest Blackwell chips have 148 or 152. If you use the same seed on different chips, the same numbers get put in different places because the number of processing units is different. The final layout ends up encoding the number of workers that distributed the numbers, and thus the chip that did the work. Training then changes most of the starting numbers, but not all of them. Some are never touched and stay at their starting values. Using a Blackwell chip, DePue reproduced the untouched starting numbers, in the exact order, from DeepSeek's V4 Flash model, released in April. He then found the same Blackwell fingerprint in six models from DeepSeek, Moonshot, MiniMax, and Ant Group. The discovery is notable because limiting China's access to the chips is the cornerstone of America's efforts to slow China's AI advances. Since 2022, the US has banned sales of its most powerful AI chips to China, arguing they would make the country's military planning faster and more accurate. The goal was to keep Chinese labs a generation behind. The Blackwell chip is more than twice as powerful as Nvidia's previous flagship product, and all four labs released their first models built on Blackwell within six months of each other. By DePue's estimate, the final training runs alone used around 12,000 of the processors, a stack of hardware worth more than half a billion dollars and illegal to ship to China without a US export license. But as DePue points out, Blackwell chips can be rented from data centers outside China: what he has shown is access to the hardware, not that it was illegally sent to China. US export rules regulate where a chip is shipped, not who uses it. If a Chinese lab rents time on a Blackwell located in Thailand, that's not clearly a violation, and a proposed regulation to plug this loophole has been stuck in draft since the summer. It's also possible that the trick DePue used to detect the chips won't work a second time. He couldn't find his Blackwell signal in the model that DeepSeek released after that, V4.1 Flash, which he attributes to the company changing how it sets its starting numbers. Washington's main tool for holding back Chinese AI development didn't stop any of the top four labs from training on the latest generation of chips, and the one technique for detecting that fact is something the labs can switch off.
13
127
8
1,156
59,118
A Santa Barbara Home Designed Around Its Ocean View Lists for $23 Million trib.al/vEGIJ9e
14
Ricardo Bofill completed this jaw-dropping fortress-like apartment complex back in 1963
14
not if… qwen
I updated my quantization compiler to use e-means as a replacement for k-means in codebook generation and got Qwen3.5-4B from 8.41 GB to 1.23 GB while retaining 91.9% of original performance which I think is approximately SOTA, at least comparable to bonsai 2 but I think we could go lower if someone wanted to donate some GPU time since this is just a hobby for me. at this size, it's small enough that memory bandwidth isn't the bottleneck to decode speed so I'm interested in seeing how fast we can get larger models on consumer hardware. link below.
2
28
make it rain on this pole dancer
New @PolestarCars is a masterpiece ! That knob tho !!!! Next Favorite designers car.
15
just in case
The Vision Jet has a whole-airplane parachute — CAPS. It's meant to deploy at a low airspeed but the Learjet tragedy got me thinking, what would happen if you deployed it as a last resort in an overspeed condition? If I were in that kind of situation — airplane coming apart or out of control, hand on the CAPS handle — what would that actually mean for me? The SF50 flies lower and slower, with a wing that isn't in Lear 60 coffin-corner territory. Mach tuck is basically not an issue for the Vision Jet. Official CAPS limit is the lower of 135 KIAS or 145 KTAS. Pull inside that and it deploys. The system is also designed to pitch up, idle the engine, and slow you first — and if it still can't get slow enough, to deploy anyway. What I still don't love: there was no required full in-flight high-speed CAPS test for certification. Analysis and lower-level testing, yes. So if you somehow couldn't get the speed down, what would happen to the chute and the airframe? CAPS is a huge safety net. It isn't magic. Respect the weather, know your envelope, and don't put yourself in a place where the only option left is hoping the parachute forgives the speed. Any pilots or physicists know what actually happens if CAPS comes out in an overspeed? Curious what other Vision Jet / Cirrus folks think. #Aviation
1
26
Replying to @Tesla
@Tesla Roadster Thruster system on display at the @NASA MAX POWER event today … the Carbon Over-wrapped Pressure vessel (COPV) that holds ~10,000 lbs (~840 bar) of pressure, high-performance valves, pneumatic pipes and the nozzles. Here are some close in views of what it takes to make a Roadster fly!
15
@𝑀𝑒𝓂𝑒𝓈𝒟𝑒𝓁𝑀𝒶𝓇 retweeted
Replying to @elonmusk
“Our truck is selling too well, we better cancel it”
76
59
4
5,749
204,724