←
Rome Thorstenson142 posts
Rome Thorstenson avatar

Rome Thorstenson

@foundinrome

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

◷ Joined November 2017▧ 51 Photos and videos
Rome Thorstenson avatar
Rome Thorstenson@foundinrome·Nov 4

What are the most common words in English that don't rhyme with anything?* Turns out a gulf of words have put orange out of its depth. *in the top 10k most common English words. 2/3

Post attachment
Rome Thorstenson avatar
Rome Thorstenson@foundinrome·Nov 4

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

Post attachment
Rome Thorstenson avatar
Rome Thorstenson@foundinrome·Oct 19

So...apparently there's no such thing as meter in French? It just doesn't have lexical stress. I pity parisian poets.

Rome Thorstenson avatar
Rome Thorstenson@foundinrome·Aug 22

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.

Post attachment
Rome Thorstenson avatar
Rome Thorstenson@foundinrome·Aug 22

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.

Post attachment
Rome Thorstenson avatar
Rome Thorstenson@foundinrome·Aug 22

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.*

Post attachment
Rome Thorstenson avatar
Rome Thorstenson@foundinrome·Jul 30

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.

Post attachment
Rome Thorstenson avatar
Rome Thorstenson@foundinrome·Jul 18

no matter what I do, this will be my crowning achievement Semantle #1631 ✅ 4 Guesses 🔝 Guess #1 🥈 866/1000 💡 0 Hints semantle(.)com

Post attachment
Rome Thorstenson avatar
Rome Thorstenson@foundinrome·Jul 9

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.

Rome Thorstenson avatar
Rome Thorstenson@foundinrome·Jul 9

Bonus: fusion isn't just more accurate — it's more decisive. It concentrates its score mass on a few pairs (Gini 0.26 vs 0.12–0.17 for the perturbation methods) and is right about them. Sharp, not just better.

Post attachment