نتائج البحث
Learning to play Minecraft with Video PreTraining
We trained a neural network to play Minecraft by Video PreTraining (VPT) on a massive unlabeled video dataset of human Minecraft play, while using only a small amount of labeled contractor data. With fine-tuning, our model can learn to craft diamond tools, a task that usually takes proficient humans over 20 minutes (24,000 actions). Our model uses the native human interface of keypresses and mouse movements, making it quite general, and represents a step towards general computer-using agents.
Introducing New Ways to Verify Age on Instagram - meta.com
Introducing New Ways to Verify Age on Instagram meta.com
Helping Creators Build a Business Now — And for the Metaverse - meta.com
Helping Creators Build a Business Now — And for the Metaverse meta.com
Look back on 10 years of Amazon robotics - About Amazon
Look back on 10 years of Amazon robotics About Amazon
Look back on 10 years of Amazon robotics - About Amazon
Look back on 10 years of Amazon robotics About Amazon
Introducing the Meta Avatars Store - meta.com
Introducing the Meta Avatars Store meta.com
Duty of Vigilance Plan 2021 - Amazon Sustainability
Duty of Vigilance Plan 2021 Amazon Sustainability
Matching Products to People with Meta Discovery Commerce - meta.com
Matching Products to People with Meta Discovery Commerce meta.com
Making VR More Social With Meta Horizon Home - meta.com
Making VR More Social With Meta Horizon Home meta.com
Amazon Prime Air prepares for drone deliveries - About Amazon
Amazon Prime Air prepares for drone deliveries About Amazon
Amazon Prime Air prepares for drone deliveries - About Amazon
Amazon Prime Air prepares for drone deliveries About Amazon
AI-written critiques help humans notice flaws
We trained “critique-writing” models to describe flaws in summaries. Human evaluators find flaws in summaries much more often when shown our model’s critiques. Larger models are better at self-critiquing, with scale improving critique-writing more than summary-writing. This shows promise for using AI systems to assist human supervision of AI systems on difficult tasks.
AI-written critiques help humans notice flaws
We trained “critique-writing” models to describe flaws in summaries. Human evaluators find flaws in summaries much more often when shown our model’s critiques. Larger models are better at self-critiquing, with scale improving critique-writing more than summary-writing. This shows promise for using AI systems to assist human supervision of AI systems on difficult tasks.
Inspiring Young People in Europe to Make the Internet the Kind of Place They Want to Be - meta.com
Inspiring Young People in Europe to Make the Internet the Kind of Place They Want to Be meta.com
Techniques for training large neural networks
Large neural networks are at the core of many recent advances in AI, but training them is a difficult engineering and research challenge which requires orchestrating a cluster of GPUs to perform a single synchronized calculation.
Techniques for training large neural networks
Large neural networks are at the core of many recent advances in AI, but training them is a difficult engineering and research challenge which requires orchestrating a cluster of GPUs to perform a single synchronized calculation.
Brand Protection Report - assets.aboutamazon.com
Brand Protection Report assets.aboutamazon.com
Amazon’s Brand Protection Report - About Amazon
Amazon’s Brand Protection Report About Amazon
Launching AMBER Alerts on Instagram to Help Find Missing Children - meta.com
Launching AMBER Alerts on Instagram to Help Find Missing Children meta.com
Meet the team behind Amazon’s first-ever clothing store - About Amazon
Meet the team behind Amazon’s first-ever clothing store About Amazon
خبر — منصة إخبارية ذكية | Khabr — AI-Powered News Platform
خبر هو أول مجمّع أخبار عربي يعمل بالذكاء الاصطناعي. نقدم تحليلات ذكية وملخصات تلقائية ورواية صوتية لكل خبر من أكثر من 700 مصدر موثوق. نضيف قيمة تحريرية فريدة من خلال أدوات الذكاء الاصطناعي التي تساعدك على فهم الأخبار بعمق أكبر.
Khabr is the first AI-powered Arabic news aggregator. We provide AI-generated editorial analysis, automated summaries, audio narration, and fact-checking for every article from 700+ trusted sources. Our platform adds unique editorial value through AI tools that help you understand the news more deeply.