Pradeesh Suganthan

exploring continual learning, robotics, space 🚀

[Pradeesh Suganthan @pradeeeesh]

@arian_ghashghai 100%, it feels like everyone's trying to be a shovel seller instead of digging for gold.

[Pradeesh Suganthan @pradeeeesh]

Scaling is all you need

[Generalist @GeneralistAI]

Introducing GEN-1.5, a one-shot learner.

It can learn new tasks in a few seconds. Show it what to do, and it generalizes.

This capability emerged from pretraining on physical data at scale, as a step towards our mission of building general intelligence for the physical world.

[Pradeesh Suganthan @pradeeeesh]

@Raul_RomeroM @trainrobotsfast Let's go Kite! 🔥🔥🔥

[Pradeesh Suganthan @pradeeeesh]

Really love this. I connected it to eleven labs and turned it into a personal bookshelf that can read books to me and auto-flip through as I work, thoroughly enjoy it

[Meng To @MengTo]

I made this Three.js bookshelf inspired by the Stripe Press site. It scrolls horizontally, and you can open each book and flip through the pages.

It took quite a few prompts to get the textures and small details right.

I’m open-sourcing the code and prompt.

Demo:
https://t.co/9zJ6sfpX0v

GitHub:
https://t.co/jisHCh8lya

Inspiration:
https://t.co/qfoj8sR6n5

[Pradeesh Suganthan @pradeeeesh]

@Raul_RomeroM Good benchmarks for robot policies are sorely needed. And open source too?! Well done 🔥

[Pradeesh Suganthan @pradeeeesh]

@Raul_RomeroM @trainrobotsfast This is 🔥🔥

[Pradeesh Suganthan @pradeeeesh]

Feels like there's been a crazy amount of robotics news today and it's not even 9am

[Pradeesh Suganthan @pradeeeesh]

@RussTedrake Congrats, this is huge!

[Pradeesh Suganthan @pradeeeesh]

@Dian_Wang_ Woah, really interesting approach!

[Pradeesh Suganthan @pradeeeesh]

@hla_michael @status_effects @AlecRad @DavidDuvenaud Really interesting directions, just DM'd!

[Pradeesh Suganthan @pradeeeesh]

@Darpinian @ericjang11 Yeah I don't know if it's changed over time or I'm only noticing it now but it definitely feels like it's become much more cynical of late.

[Pradeesh Suganthan @pradeeeesh]

@ericjang11 Really enjoyed reading this. Going through a similar journey (robotics background, implementing deep learning foundations), like your use of Claude for experiments. Curious about your robotics point at the end, how do you see automated reasoning affecting the sim/real balance?

[Pradeesh Suganthan @pradeeeesh]

@JustenMichel Great read, was trying to map out the safety-camp disagreements myself. The OpenPhil / Yudkowsky delineation is clarifying.

What do you think about how these tribes show up within the labs? Anthropic seems fairly EA-coded, OpenAI is a mix and Deepmind largely a blackbox?

[Pradeesh Suganthan @pradeeeesh]

@jackclarkSF Interesting to think about this as cyclic. Software created a speed differential between digital and physical worlds, now AI will create the same phenomenon within software itself. Might be some interesting historical analogues to watch for here

[Julius Horsthuis @JuliusHorsthuis]

Orbital assembly

[Elon Musk @elonmusk]

It is high time that humanity went beyond Earth. Should have a moon base by now and sent astronauts to Mars. The future needs to inspire. https://t.co/6HjDQnRSA5

[Pradeesh Suganthan @pradeeeesh]

Argues point estimates don't include uncertainty; switching to distributions decrease probability of life . Implies early great filter.

[Pradeesh Suganthan @pradeeeesh]

Interesting take on the Fermi Paradox: https://t.co/6ImCwwgCgp