John Yang

If you’ve ever - thought AI protein folding is magical ✨ - wanted more than a pLDDT score 🔎 - or just think mech interp in bio is cool 🤓 then read the 🧵 👇 on our first paper towards interpretable protein structure prediction just accepted to workshops at ICLR

@wilson_spearman @Lovable @antonosika the teacher copies the student

lol wtf @Lovable is running an ad with screenshots of our website but for their platform?? what's going on @antonosika?

@kognise7 +78,234% MRR btw

@NeilDeshmukh @sola_ai @a16z @conviction @ycombinator Congrats! 🎊

Excited to announce that @sola_ai has raised a $17.5M Series A led by @a16z with support from @Conviction @ycombinator, bringing total funding to $21M 🚀 From the start, we set out to reimagine human-AI interaction to push the boundaries of process automation. Our agents watch how people do tasks on-screen, then handle those tasks automatically, even in legacy tools.

@aditabrm unc

Picked 26-34 instead of 18-25 as my age group in a form for the first time and that hurt

Abstraction requires a tradeoff between task accuracy and compression. Binary codes => 100% accuracy. English => less accuracy, dependent on receiver. I like how the information bottleneck method rigorously describes this tradeoff, formulating it as a minimax optimization. en.wikipedia.org/wiki/Informati…

I start to believe that there is some subtle difference between compression (common for all intelligence) and abstraction (unique for artificial intelligence of human). They are definitely related, but different in a fundamental way. This shall be our next major quest for AI.

@nkbuduma @AmbienceAI Congrats on the raise! Goat USABO alum founder👏👏👏

Thrilled to announce our $243M Series C raise for @AmbienceAI. Honored to work with some of the smartest & most mission-driven people in the world, serving the greatest healthcare institutions in the world.

Grateful to be featured in @Forbes and to go deeper and share the vision behind @phosphorcap. Thank you to @dasha_shunina for taking the time to understand Phosphor’s mission.

🚨 New paper alert! Linear representation hypothesis (LRH) argues concepts are encoded as **sparse sum of orthogonal directions**, motivating interpretability tools like SAEs. But what if some concepts don’t fit that mold? Would SAEs capture them? 🤔 1/11

2/3 companies I mocked this batch got in! Feeling proud. Highest leverage 15 min every time.

@kaivu, @atticuswzf , and I were researching long horizon reasoning (with @jacobandreas). We found existing benchmarks’ hard problems often featured tricky puzzles, not tests of system understanding. So we made Breakpoint: a SWE benchmark designed to disambiguate this capability.

The YC Summer 25 application deadline is May 13th! No better time to found & build tech that'll shape the next decade. To pay it forward, I'm happy to review applications & mock interviews so reach out! ycombinator.com/apply

Biology’s lack of data is holding back its AI boom. Epoch’s latest report shows explosive growth in biological model training data size from 2017–2021 (9.7×/year), but a (2.1x/year) plateau since. AI models for biology are ready to transform science if the data can keep up.

I’m at #ICLR2025 the next few days! I’m making lots of time in my schedule to talk to researchers who might someday want to start a company. If that’s you, sign up for a YC office hours with me here. events.ycombinator.com/yc-oh-iclr

@yusufroohani You too! Hope to bump into you again, especially at the bio workshops

@johnyang100 Sorry, was a bit late to the session. Nice meeting in person!

I’m at #ICLR2025 this week and enjoying my time (and the vibrant batik!) in Singapore so far! Let’s connect if you’re interested in foundation models for biology and agent driven biological discovery! DMs open

Our paper "Everything, Everywhere, All at Once: Is Mechanistic Interpretability Identifiable?" will be presented at #ICLR2025! It's also the first paper of my first PhD student — congrats @maximemeloux! 🎉 blog post: melouxm.github.io/MI-identifiabi… A short thread 🧵

Generally curious abt spectral applications for interpretability though! Would love to talk to someone other than o3 abt this haha.

To put out a specific topic, I’ve been intrigued by the Free Independence Principle and Jacobian SV distribution in TP III and their implications for interpreting cross-layer computation.

Heading to ICLR in 🇸🇬 next week. If anyone’d like to connect abt random matrix theory & Tensor Programs for mechanistic interpretability, DM me and if you’re in Singapore or SF, I’ll treat.