When we get to AGI, it will have come slowly, not overnight. A NeurIPS Outstanding Paper award recipient, Are Emergent Abilities of Large Language Models a Mirage? (by Rylan Schaeffer, Brando Miranda, Sanmi Koyejo) studies emergent properties of LLMs, and concludes: "...

Excited to share a new preprint! (1)🔬The nucleolus is the most prominent nuclear condensate, with a fascinating multilayered liquid-like structure, and is the site of ribosome biogenesis. But how does this multiphase architecture form and function? biorxiv.org/content/10.110…

Thanks for the kind words, Kenny. Innovation in Mechanical Engineering is intimately tied to improved measurement standards (think yield stress of a screw). Biology is no different.

In this thread, Avi labels Chinese advancements in drug development as terrifying. He attributes increase in drug-candidates to FDA-like policy and improving recruitment timelines. One possible explanation of a ~2x recruitment timeline could be a ~4x population to recruit from.

Brando‘s poster is a great example of research communicated with clarity and focus. Congrats on the work!

🔄 We were nominated for Oral+top 1 in the MATH-AI workshp at #ICML! 🚨Why? ≈46 % of GitHub commits are AI-generated—but can we verify them correct? 📢 VeriBench challenges agents; turn Python into Lean code! 🧵1/14 📃 Paper: openreview.net/forum?id=rWkGF…

For biological data, if you don't have deep expertise in this low value work called data cleaning, u r lacking a fundamental understanding of the idiosyncrasies of the data. Without this knowledge, it is impossible to seriously model data.

Adam Green Models trained on observational RNA data with no notion of time have no chance of learning models that have strong guarantees of biological causality. They are learning correlations.

A model trained with a frequentist objective using i.i.d. observational train and validation data is almost surely converging to the non-causal model of the underlying data generating process. That is because using short cuts/spurious correlations leads to a lower loss than not,

whoops, almost found SRA IDs manually god bless easy to use command line tools

This asymmetry comes from the consumer's motivational frame. Products, like AirPods, add pleasure. Biologics, like tirzepatide, relieve pain. Consumers believe the cost of not feeling bad (aka normal) should be as low as possible relative to the cost of feeling extra good.

gpt-oss is out! we made an open model that performs at the level of o4-mini and runs on a high-end laptop (WTF!!) (and a smaller one that runs on a phone). super proud of the team; big triumph of technology.

500tk/s structured data extraction. Incredible time to be alive

Exciting. New open-source language model from OpenAI, fully local and feature rich.

A company claiming GPT-5 is a “legitimate PhD-level expert” has clearly never sent it to group meeting to defend this barplot. 52.8% > 69.1% == 30.8%

This is awesome. Biology experiments benefit from equipment standardization and customization, made possible with LEGO Mindstorms.

I can't tell if this chart is a meme or if it deserves a spot on vibechart.net

I enjoy reading the "thinking" tokens of AI models. I wish ChatGPT and co made them persistent.