←
Valentin Mihov 🦞14,697 posts
Valentin Mihov 🦞 avatar

Valentin Mihov 🦞

@vmihov

Yield farming and angel investing ~ Crypto multi-cycler. @10b57e6da0 ~ co-founder of @daedalus_angels ~ Shadowy super-coder👨‍💻

◷ Joined January 2009▧ 963 Photos and videos
IRIS avatar
⟳ Valentin Mihov 🦞 retweeted
IRIS@iris_credit·Nov 16

1/ We heard you. Here's Quote Preview. Request a fixed-rate quote on IRIS without connecting your wallet. app.iris.credit/borrow

0:08
Valentin Mihov 🦞 avatar
Valentin Mihov 🦞@vmihov·Nov 14

It’s amazing what AI can achieve nowadays given the right tools. From chat to a 3d print in a few hours. Didn’t draw a single line by hand.

Post attachment
will depue avatar
⟳ Valentin Mihov 🦞 retweeted
will depue@willdepue·Nov 12

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.

Post attachment
cuban 🍂 avatar
⟳ Valentin Mihov 🦞 retweeted
cuban 🍂@cubantobacco·Nov 12

the pH of the vagina is so acidic it can completely dissolve $1.25T in market capitalization

Muhammad avatar
⟳ Valentin Mihov 🦞 retweeted
Muhammad@zulali·Nov 10

Claude reverse engineered my 32-year-old Jeep. Then it started controlling it. Edited with Fable 5.2

9:47
Pietro Miotti avatar
⟳ Valentin Mihov 🦞 retweeted
Pietro Miotti@PietroMiotti·Nov 10

My AI agents found an unmapped Maya site in Mexico, using LiDAR data from a NASA dataset. 72 hours from a random thought on vacation to a reply from Mexico's national archaeology institute. What a time to be alive…

0:13
V12 avatar
⟳ Valentin Mihov 🦞 retweeted
V12@v12sec·Nov 11

Any telegram message can now become RCE 6630b003766baed1fa86cea880d22e41dc416f63d800de3246941ede2396ad48

0:15

Any Telegram message can now become an app. ⚡ Downloads instantly (less than 4 KB)
🎨 Interactive and natively rendered
💸 Payments and viral mechanics built in 🛠️ Build your own. More info and docs for devs: t.me/BotNews/122

0:13
IRIS avatar
⟳ Valentin Mihov 🦞 retweeted
IRIS@iris_credit·Nov 10

1/ IRIS is LIVE on mainnet. IRIS is a fixed-rate borrowing protocol. Competitive quotes for any number of days, sourced from DeFi’s deepest liquidity. Get a quote in seconds : app.iris.credit/borrow

0:08
nil avatar
⟳ Valentin Mihov 🦞 retweeted
nil@dhvanil·Nov 6

i went to bed and asked claude to paint an autumn tree overnight. at around 4am, it realised that it didn't have some shades -- so it stopped and titrated its own paint: diluting red, sienna and madder step by step on a second palette, then measuring each pocket from a photo. woke up to this beautiful watercolor painting. 303 brush marks, took about 4.5 hours 🍂 (still working on the speed)

0:24
Valentin Mihov 🦞 avatar
Valentin Mihov 🦞@vmihov·Nov 10

New fear unlocked. Now I need to pentest all devices at home 😭

After my Devcon 8 talk (the LLM hacking hardware one), somebody asked what have I hacked in my own home. So naturally, I arrived home today, and you know... not a bad idea? HOLY FUCK This isn't coding error, this is criminal negligence 😱

Valentin Mihov 🦞 avatar
Valentin Mihov 🦞@vmihov·Nov 10

Is using Rust the best option for these experiments? Anyone tried languages like F*? fstar-lang.org/

I had Codex build a CPython bytecode compiler in Rust (`serc`). It emits byte-for-byte identical output on the top 10,000 PyPI packages, 2-4x faster than the CPython implementation. This was entirely an experiment, though we may ship it (opt-in) in uv.

Post attachment
Charlie Marsh avatar
⟳ Valentin Mihov 🦞 retweeted
Charlie Marsh@charliermarsh·Nov 9

I had Codex build a CPython bytecode compiler in Rust (`serc`). It emits byte-for-byte identical output on the top 10,000 PyPI packages, 2-4x faster than the CPython implementation. This was entirely an experiment, though we may ship it (opt-in) in uv.

Post attachment
Valentin Mihov 🦞 avatar
Valentin Mihov 🦞@vmihov·Nov 10

A new fixed rate lending protocol is launching today. Different design from Morpho Midnight. Worth checking out!

2/ Core Phase terms, for every pair below :

Post attachment
David Andrews avatar
⟳ Valentin Mihov 🦞 retweeted
David Andrews@broyojo42·Nov 9

I had Opus 5.5 formalize CPython in Lean: 613,318 lines of Lean that copy Python's internals line by line, tested against the real interpreter and checked by Lean. On the way it found real bugs in Python, including crashes from plain Python code. 🧵

0:11