0xMarioNawfal (@RoundtableSpace)작은 데스크톱 GPU에서 35B 모델을 10개 에이전트가 동시에 구동하며 74W로 총 436 tok/s를 달성했다는 주장이다. 데이터센터나 클라우드 없이도 고성능 AI 인퍼런스가 가능함을 보여주며, 온디바이스/로컬 AI 인프라의 가능성을 강조한다.https://x.com/RoundtableSpace/status/2048420586192543826#aiinfrastructure #inference #ondevice #llm #agents
Related
🎓🤖 Oh no, the brave souls attempting to discuss #AI at graduations were met with boos! Apparently, the #crowd couldn't h...
🎓🤖 Oh no, the brave souls attempting to discuss #AI at graduations were met with boos! Apparently, the #crowd couldn't handle the radical notion that in the future, their jobs migh...
Качество кода в эпоху AI: как не утонуть в багах и уязвимостяхЭто конспект вебинара. Спикер — Даниил Степанов, разработч...
Качество кода в эпоху AI: как не утонуть в багах и уязвимостяхЭто конспект вебинара. Спикер — Даниил Степанов, разработчик-исследователь Veai, преподаватель ИТМО, ранее работал в J...
📰 EU AI Act 2026: August Deadline for High-Risk AI Compliance & Global ImpactThe enforcement of the EU AI Act's most str...
📰 EU AI Act 2026: August Deadline for High-Risk AI Compliance & Global ImpactThe enforcement of the EU AI Act's most stringent requirements for high-risk AI systems begins in Augus...