This is a very compelling demo! SQLite in WASM as the state engine for a reactive JavaScript application, such that if you modify rows in the database the UI automatically updates to match
I had fun giving this talk in @chipro 's ML Systems Class @Stanford about a subject we don't talk about enough:
1. How to evaluate ML Tooling
2. How to spot & deal with 🔥Tool Zealots 🔥
We have recorded the full talk 👇
https://t.co/BPEuepEkOX
Also a 🧵👇
In Q1 2019, 39% of Stripe's hiring was outside Bay Area and Seattle. Last quarter, it was 74%. I think the rate at which tech industry is going global is still under-appreciated, and that this will be a big tailwind for the world over the next decade.
Inspired me to look up the same numbers.
For Coinbase, Q1 2019 was 30% outside the Bay Area, and last quarter (Q4 2021) was 89% outside the Bay Area.
Tech is definitely decentralizing. Silicon Valley is now in the cloud.
@patrickc Same thing: In 2019, 48% of @Cloudflare’s hiring was outside the Bay Area. Today it’s 78%. Our San Francisco office is our only office globally that shrank in 2021.
We've spent the last 3 years at @DeterminedAI building a deep learning training platform, and I'm excited we're now able to share it with the ML community! Open source now under an Apache V2 license. https://t.co/Z6uBpOtges https://t.co/aiijpmKC2p
Hello! A lot of you have started following me in the last couple of months, so let me introduce you to some people I respect, who've created some of the pandemic writing that's really stuck with me.
In our preprint “The large learning rate phase of deep learning: the catapult mechanism" https://t.co/H0CdiPXrUl, we show that the choice of learning rate (LR) in (S)GD separates deep neural net dynamics into two sharply distinct types (or "phases", in the physics sense). (1/n)
Predicting + demonstrating counterintuitive neural network training behavior:
- training at learning rates which diverge under NTK theory
- exponential *increase* in loss over first ~20 training *steps* (not epochs)
- drastic reduction in Hessian eigenvalues over first ~20 steps
Appreciate the interest in #Georgia & the many Georgians who explained to me how hard we're all struggling to do the right thing, among them @JasmineForHD108, @RepMParis, @TrammellBob, @Seth_C_Clark, @JennJonesATL, @joshuasweitz. Grateful for you all. https://t.co/RdrmZS8NMB
This is very cool work. Read this if you want to really, really understand how a neural network solves a specific problem -- like actual scientific understanding.
I have some exciting news to share. After spending some wonderful years at Northeastern, I will be joining @ucsd_cse at @UCSanDiego as an Assistant Professor in July 2020.
I want to thank all my mentors for their support, and for folks at UCSD for making this happen.
A Survey of Deep Learning for Scientific Discovery
To help facilitate using DL in science, we survey a broad range of deep learning methods, new research results, implementation tips & many links to code/tutorials
Paper https://t.co/N4wLdL7sm3
Work with @ericschmidt
Thread⬇️
I'm delighted to be joining the faculty at Stanford to lead a new Center on Advanced Studies of the Digital Economy.
Very much looking forward to working with the amazing folks @StanfordHAI, @SIEPR, @StanfordGSB and @Stanford Dept of Economics.
https://t.co/4HpQr5MFke
Wow, thanks a lot for your kind words, Tim! 😊
It's been a wild and exciting ride, and I feel incredibly grateful to everyone who helped me along the way. The field is moving so fast, I'm curious to see what this new decade will bring! 🦎
Frustrated with the complexity and challenge of working with extra-wide datasets? Check out how the Sisu #ML team uses graph coloring to perform efficient column reduction in wide datasets without sacrificing data quality. https://t.co/QSkZ1o8Dxf