misk@sopuli.xyz to Technology@beehaw.org · 25 days agoDeepSeek-V3 now runs at 20 tokens per second on Mac Studio, and that’s a nightmare for OpenAIventurebeat.comexternal-linkmessage-square16fedilinkarrow-up11arrow-down10
arrow-up11arrow-down1external-linkDeepSeek-V3 now runs at 20 tokens per second on Mac Studio, and that’s a nightmare for OpenAIventurebeat.commisk@sopuli.xyz to Technology@beehaw.org · 25 days agomessage-square16fedilink
minus-squareParetoOptimalDev@lemmy.todaylinkfedilinkarrow-up0·25 days agoNo, people who want something approaching chatgpt but local want to run at least deepseek V3 32B. Qwen at least fares much worse for my usage as do deepseek V3 under 32B.
minus-squaremorrowind@lemmy.mllinkfedilinkarrow-up1·edit-225 days agoThe hell is v3 32b. Are you talking about a distill
minus-squarevintageballs@feddit.orglinkfedilinkDeutscharrow-up1·20 days agoThey probably confused the R1 Qwen distill with something else. Afaik there is no 32b model from DeepSeek directly.
No, people who want something approaching chatgpt but local want to run at least deepseek V3 32B.
Qwen at least fares much worse for my usage as do deepseek V3 under 32B.
The hell is v3 32b. Are you talking about a distill
They probably confused the R1 Qwen distill with something else. Afaik there is no 32b model from DeepSeek directly.