
Our "Beyond Neural Scaling laws" paper got a #NeurIPS22 outstanding paper award! Congrats Ben Sorscher, Robert Geirhos, @sshkhr16 & @arimorcos awards: blog.neurips.cc/2022/11/21/ann… paper: arxiv.org/abs/2206.14486 🧵 x.com/SuryaGanguli/s…
sshkhr
1.4K posts

@sshkhr16
research eng @GoogleDeepMind prev: founder @DiceHealth, researcher @AIatMeta @VectorInst

Our "Beyond Neural Scaling laws" paper got a #NeurIPS22 outstanding paper award! Congrats Ben Sorscher, Robert Geirhos, @sshkhr16 & @arimorcos awards: blog.neurips.cc/2022/11/21/ann… paper: arxiv.org/abs/2206.14486 🧵 x.com/SuryaGanguli/s…

The first experimental evidence of recursive self-improvement (RSI). Autoresearching the autoresearch agent for eight days. The result beats the harness we hand-tuned for two years, on held-out benchmarks: 🧵(1/7)


Hioya curry opened up in Toronto!!!! I loved the Akihabara location so much and it opened up here right next to the carlton movie theatre!!! I brought my stamp card from Japan LMAO

Toronto ❤️





A super long overdue (3+ years?) post on scaling laws. Compute is expensive. Scaling laws are a way to help us reason about the optimal compute allocation between data and model size before committing to a large run. The post covers what scaling laws predict, how compute-optimal allocation works, why Kaplan et al. and Chinchilla disagree, and how data limits + fitting details make extrapolation tricky. lilianweng.github.io/posts/2026-06-…


The CUDA moat has evaporated overnight. CUDA was necessary when humans wrote code. Now it isn’t. TPU and other non-Nvidia architectures are about to take off.

Why Remote Work is White Collar Fraud. "I have a three-year-old and a five-year-old. The idea that I could do any work at my house is like a total fantasy. The kids come home at 3pm, your work day needs to keep going. I'm highly against it." @typesfast




Another "hidden" area where the United States absolutely dominates the rest of the world, including Western Europe: wastewater/sewage treatment Waste water treatment infrastructure is pervasive in the United States. Even a low income area in rural Georgia has a state of the art waste water treatment facility. Western Europe has good coverage, but on a per capita basis, their treatment plants serve a larger population. This indicates they aren't as pervasive as they are in the US. In contrast, China (even Japan) has sparse coverage despite its large dense population. India's coverage is even lower. A lot of accounts on twitter incorrectly use examples of impressive skylines (ie Dubai, Shanghai, and now recently Mumbai/Gurgaon/Hyderabad) to show that their region is "developed". However, this is the real indicator of development. Sewage treatment plants don't make for mediagenic ribbon cutting ceremonies. They don't win you votes. But the long term population health benefits are immense: lower rates of childhood diseases, longer healthief life spans, less stunting (and the associated cognitive and physical deficits associated with it).


I get this like 3 times a day