mastodontech.de ist einer von vielen unabhängigen Mastodon-Servern, mit dem du dich im Fediverse beteiligen kannst.
Offen für alle (über 16) und bereitgestellt von Markus'Blog

Serverstatistik:

1,4 Tsd.
aktive Profile

#llama4

0 Beiträge0 Beteiligte0 Beiträge heute

🤖 AI
🔴 Zuckerberg Forms AI Team for “Superintelligence”

🔸 Dissatisfied with Llama 4, Zuckerberg takes direct control of Meta’s AGI efforts.
🔸 New elite team of ~50 experts to focus on next-gen models.
🔸 Meta lags behind OpenAI, Google, Anthropic & xAI in the AGI race.
🔸 Goal: build AI that can do anything a human can.

#Meta#Zuckerberg#AI

I had fun prompting Meta's #Llama4 WhatsApp #chatbot on Determinism, Reductionism, Cartesianism, Robert Rosen on Reductionism, (M,R)-Systems and Category Theory, and finally, Hans Werner Arndt's treatment of Mos Geometricus. A mixed bag.

The cherry on the pie was when I prompted it to criticize its former statement about Arndt on the background of his work. When I confirmed that it had first produced rubbish, the bot turned from confident teacher to repentant cheating pupil.

Crazy stuff.
#AI

In this episode, we explore how Meta’s LLaMA 4 is reshaping the open-source AI landscape. From massive model performance gains to its growing impact on startups, researchers, and developers worldwide, discover why LLaMA 4 is a game-changer in the race for accessible, scalable AI.

#LLaMA4 #OpenSourceAI #ArtificialIntelligence #TechPodcast

music.amazon.com/podcasts/e9ec

music.amazon.comLlama 4 and the Rise of Open-Source AI: Breaking Barriers in Speed, Scale, and Accessibility-TechDaily.ai In this episode of Techaily.ai, we dive deep into the explosive growth of open-source AI, focusing on Meta’s Llama 4 model—a true game-changer in the AI landscape. Sponsored by StoneFly, a leader in enterprise storage and cloud solutions, we explore:How Llama 4 hit 1.2 billion downloads in just 10 weeksWhy multimodal training, massive context windows, and multi-language support make it stand outReal-world use cases: from the International Space Station to healthcare in Latin AmericaThe Mixture of Experts (MoE) model and inference efficiency breakthroughsHow voice interaction and wearable AI are redefining user experiencesThe rise of distilled models for faster, cheaper deployment—without sacrificing performanceWhy open-source collaboration is shaping the future of AI across industriesWhether you're a developer, researcher, or just AI-curious, this episode gives you a front-row seat to the open-source AI revolution.🔗 Visit Stonefly.com or email sales@stonefly.com for enterprise AI infrastructure and support.
Fortgeführter Thread

🧩 #Llama4Maverick nutzt 128 Experten für deutlich mehr Rechenleistung und schlägt sogar #GPT4o und #Gemini20 in Benchmarks – bei nur der Hälfte der aktiven Parameter von #DeepSeekv3.

🎓 Beide #KIModelle wurden mithilfe des riesigen Lehrmodells #Llama4 Behemoth trainiert, das mit 288 Milliarden aktiven Parametern zu den leistungsstärksten weltweit zählt.

👉 eicker.TV #Technik #Medien #Politik #Wirtschaft (2/2)

eicker.TV - Technews als Kurzvideo
eicker.net · eicker.TV ▹ Technews Kurzvideos als TikToks, Shorts, Reels
Mehr von Gerrit Eicker 🇪🇺🇺🇦