_Victorad
@_vicTorad • 4,595 subscribers
Creator | Amb | member @SuperteamTR open collabs || dm for inquiries 📨
Videos
One of the biggest problems with robot benchmarks is also one of the easiest to overlook: The test itself can become predictable. Axis Robotics If robots keep getting evaluated on the same fixed set of tasks, eventually a high score can tell you less about whether a model actually learned to generalize and more about whether it has become good at that particular test. That’s the problem Open Axis Benchmark is trying to tackle, together with OpenRoboto. Instead of treating the benchmark as a permanent list of tasks, Open Axis keeps introducing fresh challenges. Each evaluation round draws from the growing Axis Library, giving models a new set of manipulation tasks to deal with. The goal is simple: make the robot demonstrate that it can handle situations it hasn’t simply memorized. And this is where Axis becomes interesting. The same system continuously generates and expands the task data used to train robotic models, while Open Axis can use that growing pool to create new evaluation environments. So training and evaluation become part of the same compounding loop: more data → more diverse tasks → harder evaluation → better measurement of generalization. That matters because the next generation of robotics models shouldn't just be able to perform tasks they've seen before. They need to work when the task changes. And a benchmark that keeps changing may be a much better way to find out whether they actually can.
_Victorad35,391 просмотров • 16 дней назад
🚨 A QUICK ONE; Axis Robotics x Binance Wallet Campaign is now live! 1.5M Axis Points + an exclusive commemorative badge 🎖️ How to participate: Download the Binance Wallet browser extension Select “Add Keyless Wallet” Connect it through the Binance mobile app Open Axis Hub and log in with your Binance Keyless Wallet Go to Alliance Select Binance Keyless Wallet Complete a task Open your Portfolio and sign the trajectory on chain (Base) Complete and sign just 1 trajectory to qualify for the exclusive badge. The campaign offers 750K points per cycle, with 20K tasks released daily on a first-come, first served basis (FCFS). This should help.👇
_Victorad20,667 просмотров • 1 месяц назад
1/7 Physical AI has a data problem. More compute alone won’t teach a robot how to move, grasp, react, or recover from mistakes. LLMs had the entire internet to learn from. Robots need something different: huge amounts of real action data. That’s where Axis Robotics gets interesting. $AXIS is building a data engine for Physical AI, turning robot training into something people can contribute to directly through a browser. The loop is straightforward: Generate → Collect → Process → Deploy → Correct → Repeat. What looks simple on the surface could become something much bigger underneath: a scalable way to produce the training data robots need to become genuinely useful in the physical world. The real unlock for Physical AI might not be another bigger model. It might be better data.
_Victorad19,657 просмотров • 1 месяц назад
Больше нет контента для загрузки