Tencent’s WeChat AI team is fighting the GPU shortage with algorithmic cleverness, scaling its new WeLM model to a massive 617 billion parameters while activating just 23 billion of them. Using a novel "Hidden Decoding" architecture, they are trading a 4x increase in training compute to bypass hardware constraints. It’s a blueprint for how Chinese tech can stay...#China #AsiaAI #ChinaTech #AIhttps://asiaai.fyi/?p=513#china-watch-wechat-ai-team-details-welm-models-scaling
Related
[lien] Monitorer sa consommation électrique - LinuxFr.org #monitoring #gik #nrj #$$$ #ai
[lien] Monitorer sa consommation électrique - LinuxFr.org #monitoring #gik #nrj #$$$ #ai
Museums Association: Sector warned of ‘acute threat’ posed by AI scraping bots. “AI scraping poses substantial and unsus...
Museums Association: Sector warned of ‘acute threat’ posed by AI scraping bots. “AI scraping poses substantial and unsustainable costs on cultural heritage institutions, which are ...
ICYMI: Etiquette for the machine age: BVDW's 17-part code for AI at work: Staff get 94 pointers, from a suggested three-...
ICYMI: Etiquette for the machine age: BVDW's 17-part code for AI at work: Staff get 94 pointers, from a suggested three-prompt cap to consent before AI meeting notes. Can a volunta...