Video wird geladen...
Video konnte nicht geladen werden
Introducing The Largest Organized and Incentivized Human-Data Fleet for General-Purpose Humanoid-Robotics and large language model training Our users are creators that complete daily data quests with crypto incentives through our Mobile App 🧵
385,613 Aufrufe • vor 2 Jahren •via X (Twitter)
10 Kommentare

General-purpose Humanoid-Robots will be the largest GDP impacting technology of our lifetime with a $42T impact through companies like @Tesla_Optimus, @Figure_robot and more But AI models are expected to run out of data to train on by 2026 and humanoid data is especially scarce

Tesla implements a fleet-style learning method so that all captured video footage from cars on the road continuously improve the central self-driving AI model For humans across the world become the fleet that feeds valuable video data for humanoids

There is a field in Robotics called "learning from demonstration" where videos of a task train the central model The crux of this solution is quantity and scale - thats where crypto incentives and comes in

MetaAI created a dataset demo for this boasting only 3,670 hours of daily-life activity video takes this to a global scale with a goal of creating over 500M hours of curated daily-life activity video in an ever-growing database

Core Contributors The core contributors and advisors have spent entire careers in the development of novel deep learning pipelines for 6D pose estimation in Robotics and generation of synthetic data for deep learning training

With backgrounds in the SOTA of Robotics, PHDs in deep reinforcement learning, and experience scaling 9-figure DeFi/Gaming crypto protocols we intend to be epoch-defining for humanity Cc @generated, James Li, Vee, @chang_defi, @datorxrc, and more ++

ai人型机器人训练数据收集

lets go 🥰

Lets gooo! Bullish project @CryptoKebab69

Mecka Banger!!
