minus-squarecurry@programming.devtoSelfhosted@lemmy.world•Consumer GPUs to run LLMslinkfedilinkEnglisharrow-up2·2 days agoI tried to run Gemma 3 27B Q4K and was surprised how quickly the VRAM requirements blew up proportional to context window, especially compared to other models (all quantized) at similar size like Qwq 32B. linkfedilink
minus-squarecurry@programming.devtoSelfhosted@lemmy.world•Any non-tech-background self-hosters?linkfedilinkEnglisharrow-up4·7 months agoKudos for self-hosting fediverse stuff, man. linkfedilink
minus-squarecurry@programming.devtoSelfhosted@lemmy.world•AMD won't patch all chips affected by severe data theft vulnerability — Ryzen 3000, 2000, and 1000 will not get patched for 'Sinkclose'linkfedilinkEnglisharrow-up9arrow-down2·8 months agoShit, I have desktops running with 2200G and 2400G. linkfedilink
I tried to run Gemma 3 27B Q4K and was surprised how quickly the VRAM requirements blew up proportional to context window, especially compared to other models (all quantized) at similar size like Qwq 32B.