Emperor’s New Clothes

Pointing out the obvious

[Emperor’s New Clothes @SpatialLLM]

@tszzl I don’t feel sympathy for either side. Neither care about us.

[Andy @PositivFuturist]

Being a right wing tech bro is difficult because you work on teams of the smartest people from around the world - good people.

But we still have to be real - you guys are here to lower our wages.. you’re here because the culture your people create at home didn’t produce enough opportunity for you.

You are the outlier in a bell curve that the working class takes the brunt of.

Not all immigration is bad - but the whole project was never done in the service of the nation.

[Emperor’s New Clothes @SpatialLLM]

@Journojulz @DrMarcusP @grok has the term “English” been used to refer to Britain as a whole historically?

[Emperor’s New Clothes @SpatialLLM]

@Journojulz @DrMarcusP In the Woke era everyone wants to get their little ethnicity mentioned. The term Anglo Saxon, at least to my understanding, was casually used to refer to the British as a whole. In fact even “English” was commonly used to refer to the British.

[Emperor’s New Clothes @SpatialLLM]

@Journojulz I haven’t blocked you?

[Emperor’s New Clothes @SpatialLLM]

@goldi__of Vietnamese

[Emperor’s New Clothes @SpatialLLM]

@sullakosmo R1b vs R1a?

[Pat McGovern @pw_mcgovern]

from jensen's biographer (@stephenwitt) in today's NYT

[Mungo ˗ˏˋ ♰ ˊˎ˗ @KarmicVictim]

"to keep this Empire as a heritage of the British race...we must create an environment which will breed a virile people. Wealth will not save us if our crop of such men fail"

[Emperor’s New Clothes @SpatialLLM]

@RT_com I like how women’s instinct is to run away, and men’s instinct is to run into the danger.

[Emperor’s New Clothes @SpatialLLM]

@CalifornianSon_ Chinese get a pass because they don’t commit a lot of physical crime, but they’ve got back channels everywhere where they rort the system for what it is worth.

[Mungo ˗ˏˋ ♰ ˊˎ˗ @KarmicVictim]

Hughes during WW1: national safety has been endangered through the policy of laissez faire, allowing German influence to permeate British industry

[Emperor’s New Clothes @SpatialLLM]

@NotPrimadonna @yunta_tsai @JeebsTX A Westerner would sooner die than be forced to learn Chinese. If these immigrants aren’t willing to do that then they are not on our side.

[Emperor’s New Clothes @SpatialLLM]

@1oveWolfgirl As a guess there is pressure build up from fluid and the miso helps drain it. Might be linked to sinuses or ear.

[Emperor’s New Clothes @SpatialLLM]

@A_Monarchist_ This is the big difference between America and Australia. Australian is British and Irish only.

[Emperor’s New Clothes @SpatialLLM]

@stfbutnou Loyalty and friendship with who? That might be the problem.

[Emperor’s New Clothes @SpatialLLM]

Impressive

[will depue @willdepue]

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 https://t.co/YOtzND0Jqk.

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.

[Emperor’s New Clothes @SpatialLLM]

@libertas_HH To better understand it, it was inevitable, and will only get worse. The problem is much deeper.

[Emperor’s New Clothes @SpatialLLM]

@NotLeighStewy That’s absurd. Multi-culturalism? More like Multi-replacement.