M͏i͏c͏k 🔥 A͏b͏e͏r͏n͏a͏t͏h͏y͏ retweeted
It’s funny raising non-iPad kids because on the rare occasion that my human children have to interact with the nerve stapled children it’s like watching two different species of creatures interact, with only one species able to make eye contact or exchange pleasantries
I must be half Chinese now, I have gotten to the point where I have no problem buying food from the back of a minivan in a random parking lot in the middle of vast suburbia.
Beats 99% of the restaurants at 1/10th the cost!
1
26
M͏i͏c͏k 🔥 A͏b͏e͏r͏n͏a͏t͏h͏y͏ retweeted
This is the only thing the right invests in: useless TV ads and political consultant grifters. Imagine if $1.5b was invested in nationalist tech, businesses, infrastructure, schools, literally anything else but this massive pile of waste. A patronage network remains the biggest hurdle for the nationalists.
M͏i͏c͏k 🔥 A͏b͏e͏r͏n͏a͏t͏h͏y͏ retweeted
?
Huang family and friends huddle for Nvidia crisis talks as bags of fried chicken are delivered to $38m San Francisco mansion trib.al/Ozr4ho2
using the state of AI to restore access to my ten-year meme archive, holy cow i missed so many of these classics
1
20
M͏i͏c͏k 🔥 A͏b͏e͏r͏n͏a͏t͏h͏y͏ retweeted
Fully furnished four bedroom apartment with 2 bathrooms for US$250/month in Huizhou, Guandong. Just one hour away from Shenzhen and Hong Kong. It costs more than that to rent a car park in Hong Kong.
China looks after the people — a home is for living in, it’s not for landlords, not for property speculators, not for property developers or their Banksters.
They Call It a “Ghost City.” I Call It Paradise youtu.be/9uACy8VhODE?si=jvCx…
M͏i͏c͏k 🔥 A͏b͏e͏r͏n͏a͏t͏h͏y͏ retweeted
Today’s installment in, “Wait, we weren’t already doing that?”
Residents of federally funded public housing will be required to show proof of U.S. citizenship or eligible immigration status under a new rule published Thursday. cbsn.ws/4jdWOQ7
YOU CAN TAKE MY CHINESE MODELS STRAIGHT OUT OF MY COLD, DEAD HANDS - I AM NOT GIVING KIMI UP FOR ANYTHING!! :)
I have a major result to share.
I’ve found conclusive forensic evidence that the top 4 Chinese AI labs—DeepSeek, Moonshot, MiniMax and Ant Group—have built their newest models on NVIDIA’s most advanced Blackwell chips, which are barred from export to China.
This sounds insane, but I discovered that almost every AI model carries a hidden signature of the GPU that trained it. When a model is first set up, the starting values of its weights are laid out in a pattern that depends on the exact GPU. Each combination of GPU and seed number produces a unique pattern—a sort of fingerprint of the hardware used to create the model. Traces of that pattern survive training into the published model.
Again, this evidence isn’t based on any insider information or leaking. It’s based on a forensic analysis of the model weights released by the companies themselves.
I searched more than six trillion combinations of chip layout and seed number and matched them to open-weights models. In doing so, I found conclusive evidence that six Chinese AI models—including DeepSeek-V4-Pro, Kimi K3, and MiniMax-M3—were initialized on NVIDIA Blackwell systems. Four of the models match Grace Blackwell, the GB200 and GB300 parts in NVIDIA’s flagship liquid-cooled rack-scale systems.
I then verified these results on real hardware. Running unmodified PyTorch with the seeds I’d recovered, a Blackwell B300 reproduced DeepSeek-V4-Flash’s untouched starting weights bit for bit, and a rented GB200 reproduced Kimi K3’s to a correlation of 0.999995. A Hopper H100 reproduced none of DeepSeek-V4-Flash’s, and changing the seed by one broke the match. (DeepSeek’s seed is 953049583 by the way.) The method also was verified to match 24 known open-weights models to the hardware that developers said they used.
It is exceedingly unlikely that these results are a coincidence. Across six trillion combinations, the best accidental match should reach only about 8 sigma (i.e., eight standard deviations above what random chance would produce). The real matches range from 49 to 1,385 sigma, and even the weakest single test hits 20.
These were data-center-scale training runs. With some analysis of release schedules and model details, I estimate that the largest, Kimi K3 and DeepSeek-V4-Pro, used 2,000 to 3,000 Blackwell GPUs each. Clusters of this size cost on the order of $150 to $200 million each to buy, and across these labs alone would add up to more than half a billion dollars of export-controlled GPUs, not counting experimental compute or otherwise. It’s quite likely that we see multi-billion dollar expenditure on export-controlled GPUs.
I am highly confident in this result, and it has been independently verified by senior researchers at OpenAI and Anthropic. You can read the full report at unemploymentlabs.com/checkpo….
There is much more I’ve identified about these models through other forms of checkpoint forensics like this, and will have more to share soon.
2
91













































































