Rico Jonschkowski

ML for robots and RNA. Now at Before: Senior Research Scientist at Google Brain (Robotics Team), PhD from TU Berlin. Opinions are my own.

[Rico Jonschkowski @rico_jski]

@SkyLi0n All I'm saying is that your reaction to getting scooped tells you a lot about how much you are in it because you want the problem to be solved vs you enjoy the process of solving it and the rewards you are getting from it. And that remains true in a heavily crowded field like ML.

[Rico Jonschkowski @rico_jski]

Realization that took a while for me to really get: If someone else solves a problem you are working on, is your first reaction jealousy or happiness? If it's jealousy, you are working on the wrong problem.

[Rico Jonschkowski @rico_jski]

@KoestlerLukas It is very difficult. Try to find something that you don't just enjoy working on, but also genuinely care about the problem being solved. There is no shortage of important problems. The challenge is finding an angle of how you can meaningfully contribute to solving one.

[Rico Jonschkowski @rico_jski]

@iandanforth Yeah, I get that. And there will always be some jealousy and practical considerations. These are much more focused on *you* than on *the problem* though. A great problem to work on is one where the problem is much more important to you than your ego.

[Adam R. Kosiorek @arkosiorek]

Some ML research areas are extremely crowded. Working there is a race. But it's not about "how to benefit science" but "who gets the credit." This is the furthest from being a scientist I can imagine. How did we get here?

[Alex MacCaw @maccaw]

It's fascinating what's happening on the Midjourney Discord forums. Here's some highlights:

[Emad @EMostaque]

Delighted to announce the public open source release of #StableDiffusion!

Please see our release post and retweet! https://t.co/dEsBX7cRHw

Proud of everyone involved in releasing this tech that is the first of a series of models to activate the creative potential of humanity

[Andrej Karpathy @karpathy]

!!!! Ok I recorded a (new!) 2h25m lecture on "The spelled-out intro to neural networks and backpropagation: building micrograd" https://t.co/KQ23lQW1BT .
This is the culmination of about 8 years of obsessing about the best way to explain neural nets and backprop.

[Gautam Kamath @thegautamkamath]

Spotted on Reddit: an ML-adjacent faculty who seems to have a very healthy and relaxed approach to work and life. Why don't we hear about more people like this? https://t.co/BEIsgqaZR3

[Davis Blalock @davisblalock]

"Scaling Laws vs Model Architectures: How does Inductive Bias Influence Scaling?"

Do some architectures scale up better than others? Google and DeepMind trained a *ton* of large language models to answer this question. [1/8]

[Google DeepMind @GoogleDeepMind]

Today in partnership with @emblebi, we’re releasing predicted structures for nearly all catalogued proteins known to science, which will expand the #AlphaFold database by over 200x - from nearly 1 million to 200+ million structures: https://t.co/GjVES2pBFY 1/

[Paul Fairie @paulisci]

A Brief History of Nobody Wants to Work Anymore

🧵

[Andy Tran @andy23tran]

What's going on here?

[Joe @joe_d_campbell]

"'I have this math competition experience, that ... you have to be fast,' [Wang] said. 'But June is the opposite. … If you talk to him for five minutes about some calculus problem, you’d think this guy wouldn’t pass a qualifying exam. He’s very slow.'"

https://t.co/WrQRRzd0ge

[Ryan @BoyNamedShit]

God this is such a good prompt

[Jesse Engel @jesseengel]

One takeaway for me from (#dalle2, #imagen, #flamingo) is there's no one "golden algorithm" to unlock these new transfer learning capabilities. Contrastive, AR, Freezing, Priors, they all can work. You almost can't stop these models from exhibiting these new types of behavior...

[Scott Galloway @profgalloway]

How is an 18yo allowed to buy 2 AR rifles & 375 rounds of ammunition when we wouldn’t let him buy a beer?

[Julian Togelius @togelius]

How can you do great AI research when you don't have access to google-scale compute? By being weird. The big tech companies are obsessed with staying nimble despite being big, and some succeed to some extent. But they can't afford to be as weird as a lone looney professor.

[Yann LeCun @ylecun]

About the raging debate regarding the significance of recent progress in AI, it may be useful to (re)state a few obvious facts:

(0) there is no such thing as AGI. Reaching "Human Level AI" may be a useful goal, but even humans are specialized.
1/N