James

We developed an RL method for fine-tuning our models for precise tasks in just a few hours or even minutes. Instead of training the whole model, we add an “RL token” output to π-0.6, our latest model, which is used by a tiny actor and critic to learn quickly with RL.

GTC robot demo roundup! Feat. Noble Machines, Skild, Generalist, Sharpa hand

The future of robotics is training in learned simulators/world models. I appreciate the detail in Comma's blog post here.

New Ben Katz post! robot-daycare.com/posts/actuatio…

🎾Introducing LATENT: Learning Athletic Humanoid Tennis Skills from Imperfect Human Motion Data Dynamic movements, agile whole-body coordination, and rapid reactions. A step toward athletic humanoid sports skills. Project: zzk273.github.io/LATENT/ Code: github.com/GalaxyGeneralR…

@CyberRobooo Do it with a knife and I'll be impressed

Sharpa Robotics just showed off something impressive: a huamnoid autonomously peeling an apple with dual dexterous human-like hands Using their 63-DoF SharpaNorth setup and SharpaWave hands, it rotates the apple with one hand while precisely peeling with the other,real bimanual, contact-rich dexterity that’s a big step beyond basic grippers. They made it work with: >Shared autonomy: Operators trigger simple skills (like “rotate apple”) instead of controlling every finger, for easier data collection. >MoDE-VLA: A Mixture of Dexterous Experts model that fuses vision, language, force, and touch for stable high-DoF control. Success jumped from 0% in baselines to solid results on peeling + other delicate tasks like gear assembly and charger plugging.

This looks neat! It's about time navigation got a visual upgrade. Actually it's long past time.

My ancestors buried half their children. All mine are alive. My ancestors' house had a dirt floor. Mine is wood. I have indoor plumbing, I have hot water, I have never in my life hauled a full bucket half a mile and I probably never will. Do you know how rare it is, in human history, for small children to wear shoes? Mine have multiple pairs. I can speak to my relatives who live thousands of miles away, for free, at any time. Video, if we want video. With machine translation, if we speak different languages. The original Library of Congress had 740 books in it. I have more than that. If I run out of books in my home my local public library has 350,000. If I want to take a hundred books with me on vacation, they all fit on a device that fits in my purse. I have heat in the winter and AC in the summer and a washing machine and I have never, ever, ever had to scrub a dress clean by hand in the stream. I can look up recipes from more than a hundred different countries and I've tried dozens of them. I ride a clean and modern train across my city for $4, or take a robot taxi if I'm out too late for the train. I donate $40,000 every year to the cause of getting healthcare to the world's poorest people and even after the donations I never have to think about whether I can afford a book, or a pair of shoes, or a cup of coffee. There is a great deal more to fight for, of course. I hope that our descendants will look back on our lives and list a thousand ways they're richer. Maybe we ourselves will do that, if some of the crazier stuff comes true. But the abundance is all around you and to a significant degree you aren't feeling it only because fish don't notice water.

@figamin The game is free here! Play along thelongestyard.link/cave-story/

i'm happy that one in a while, intense autism still pays off.

It's new KRAZAM day youtube.com/watch?v=AS9y-d…

absurd visualization of RL training youtu.be/1AGVABna3xQ?si…

Posting from Starlink Direct-to-Cell. It works!

@chris_j_paxton In the short term humanoids are good because of economies of scale and for adoption in jobs humans do today. But in between the short and long term there is probably a period where a Cambrian explosion makes sense. I just hope that doesn't lock us into the Wall-E future.

@chris_j_paxton If we have infinite labor for practically free, efficiency becomes less important. The reason to keep robots humanoid in the very long term is to prevent a Wall-E future of helpless humans. I want my robots to use human-shaped tools so I can use them too, if I want.

If youre truly ai pilled how could you square that with humanoid robots? I feel like ai-assisted cad + manufacturing + cross embodiment learning + high fidelity simulation would allow for an infinite profusion of diverse robots for different ecological niches

@Noahpinion brain upload is harder than curing aging IMO. that suggests a correct order to pursue them. also I think a lot more people want to live longer than want to be uploaded.

This is sad because I *wanted* the transhumanist future. I wanted to become an energy being or a star god. I didn't want my whole species to just be replaced by a large logit regression trained on my species' past achievements.

@rikarends Huh I always wondered why you didn't see a lot of GPU fractal explorer apps. That explains it!

After a short detour trying to render fractals on the GPU i came back to the ageold conclusion: fractals on GPU suck. Without f64 precision its very limited what you can do. I got quite close with dual-single precision, it actually worked but you lose 10x perf so.

@moultano We could have been if there was a commercial reason. Doesn't seem like a lesson that applies to AI.

Thinking back on the golden age sci fi that straight-lines-on-graphs predicted we'd be on mars 30 years ago, and wondering if there are any lessons from that for the present. Were they just not paying attention to known constraints in their exuberance?

@i2cjak Brawndo

any recommendations for electrolytes that don’t have shit like “these salts were promised to us 3000 years ago” on the package? Chudcore and in a retarded way btw.