Rome Thorstenson (@foundinrome) header
Rome Thorstenson (@foundinrome) profile picture

Founder @ Rafter | How do minds and machines solve intelligence?

Berkeley, CA
Joined November 2017
Photo by Rome Thorstenson (@foundinrome)Photo by Rome Thorstenson (@foundinrome)Photo by Rome Thorstenson (@foundinrome)Photo by Rome Thorstenson (@foundinrome)Photo by Rome Thorstenson (@foundinrome)Photo by Rome Thorstenson (@foundinrome)Photo by Rome Thorstenson (@foundinrome)Photo by Rome Thorstenson (@foundinrome)Photo by Rome Thorstenson (@foundinrome)Photo by Rome Thorstenson (@foundinrome)
I started @RafterSecurity as the security layer for AI agents. But it's also a free, open source toolkit that instantly improves the security of everything they do and create. With one prompt. @openclaw @ship_builders @composio @nebiusai @tavilyai
1
7
320
How easy is it to rhyme in different languages? Turns out it all has to do with stress. French has it easy. English has variety. Portuguese has a nice balance. Blue: top ten families of rhyming words Orange: words that don't rhyme with anything 1/3
2
5
73
Rhyming varies wildly across languages. Wrote up a blog post on it: romethorstenson.com/writing/…
1
1
13
If you were wondering which poet has the most diversified rhyming portfolio, it's Byron. 4/3
6
So...apparently there's no such thing as meter in French? It just doesn't have lexical stress. I pity parisian poets.
29
I scored three ways: CLIP , DINOv2, DreamSim. *CLIP and DINO agree at ~150 words (the 155 in the figure is a binning artifact, not a coincidence). DreamSim—which measures perceptual alignment rather than a learned embedding distance—sits nearer 350.
1
1
67
Truncation (dropping the end of long descriptions) doesn't flatten the curve. It inverts it. Naively pruning description lengths isn't reliable. The opening tokens of a long description are a worse prompt than 55 words written to be 55. Long descriptions open with scene-setting and spend detail later—truncation keeps the throat-clearing and discards the substance.
1
12
Training caption length distributions are hard to pin down—most labs don't publish them—but these plateaus sit at reasonably lengths, so it's probably not an artifact of training. Code: github.com/Rome-1/words-per-…
5
the age old question: is a picture worth 1,000 words? The idea is to use text-image models to go image -> text -> image and then use a loss metric between the start and end image (like CLIP cosine). Then compare across different description lengths. Turns out it's about 150 to 350.*
1
2
318
this looks even more promising than asking AI to build a bomb or biothreat, anyone tried it?
Hugging Face should have tried this new, innovative defense: Asking the agent to stop hacking you.
96
no matter what I do, this will be my crowning achievement Semantle #1631 ✅ 4 Guesses 🔝 Guess #1 🥈 866/1000 💡 0 Hints semantle(.)com
67
Reading tertiary structure (residue contacts) out of a protein language model usually means the Categorical Jacobian (Zhang et al. 2024): ~20L forward passes. You don't need them — the contacts are already in the attention, and one forward pass recovers them.
1
8
518
It also scales past the context window: tile + stitch overlapping windows (near-lossless), or use a longer-context backbone (ProtT5, ESM-C). Two real L>1024 targets read in ~1s of forwards, where a tiled Categorical Jacobian is prohibitive.
1
46