Pinned Tweet
Took an afternoon poking at a major wallet to find a high-risk vulnerability.
Follow for more tips to subsidize your mountain bike in web3
12
1
48
15,648
Using AI agents and Lean 4 intensively for over a year we noticed a single edit would produce massive waste of compute with RAM scaling catastrophically when running in parallel.
Try github.com/th0rgal/turbo-lea…:
- 1000x faster edits
- 16x less RAM
- 20x less token
(log scale btw)
15
12
136
7,122
th0rgal retweeted
What if instead of writing code, you just stated your problem, hardware and bandwidth limits… and an AI gave you the best possible implementation with a formal proof?
1
4
1,399
th0rgal retweeted
This weekend we launched EthLib, a Lean4 library for Ethereum!
The goal is to bring Ethereum's primitives, specifications and theorems into one place, building a shared foundation for formally specifying and proving things about Ethereum.
github.com/ethereum/EthLib
15
55
307
12,719
left: 2022 handwritten notes
right: LaTeX from a custom model fine tuned on *my* handwritting
37
136
3,848
195,550
Since many of you were asking for the app: github.com/Th0rgal/orb
3
89
13,021
6 years ago, I bought the reMarkable 2 e-ink tablet. I took all my university notes on it: CS diagrams and math classes included.
Last night, I fine-tuned the @MistralAI OCR model on it to output PERFECT LaTeX from my shitty handwritting.
Keep your data preciously
6
8
305
11,467
Good luck fixing that, I'd rather update Bitcoin core
In 2025, we published a bold proposal to use fusion reactors for gold production.
And now we’ve done it: Mercury transmuted into stable gold, using fusion neutrons, at 10x the record set by CERN in 2025.
We’ve reduced alchemy to a series of difficult but tractable engineering problems.
2
455
I used to do that amount of commits in about 3 months in 2019. And to be fair that's probably a good estimation of how much time it would have taken back then to do what "I" did in a sunday.
1
5
532
1. Develop expertise in deploying an unprecedented amount of intelligence
2. Securing access to better models (may be via government approvals)
3. Negotiating low-cost inference (excess capacity?)
4. Optimizing model choice (see scaling law veritylabs.dev/benchmark )
5. Building solutions to bottlenecks (i.e. more efficient tooling to verify proofs)
1
7
125
I wanted to give a flower as a gift so I was looking for the goat. I present to you Semper Augustus, the most expensive tulip of all time (~30 years of salary), whose patterns comes in part from a virus that disrupts its pigmentation but reduces its ability to reproduce. 1/2
1
5
312
