Meta's VLA-JEPA robot AI learned from just 13 examples — trained on human videos #Shorts

Meta's Yann LeCun introduced VLA-JEPA, a new robot AI model that adds world modeling to standard vision-language-action architectures. The model was fine-tuned on just 13 examples yet demonstrated strong real-time performance, and it was pretrained on human videos — meaning it learned robot behavior by watching people. LeCun described it as the first world model ported to the LeRobot open-source robotics framework. Subscribe for daily robotics and AI research coverage. #Shorts #YouTubeShorts #YannLeCun #VLAJEPA #Robotics #MetaAI #RoboticAI #AIResearch

#Shorts #YouTubeShorts

#news #shorts #youtubeshorts #yannlecun #vlajepa #metaai #roboticsai #worldmodel

Видео Meta's VLA-JEPA robot AI learned from just 13 examples — trained on human videos #Shorts канала World News Decoded