misk@sopuli.xyz to Technology@beehaw.org · 1 month agoDeepSeek-V3 now runs at 20 tokens per second on Mac Studio, and that’s a nightmare for OpenAIventurebeat.comexternal-linkmessage-square10fedilinkarrow-up12arrow-down10cross-posted to: technology@lemmy.ml
arrow-up12arrow-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 · 1 month agomessage-square10fedilinkcross-posted to: technology@lemmy.ml
minus-squaremorrowind@lemmy.mllinkfedilinkarrow-up0·1 month agoDeepseek is an absolutely massive model, it’s not the one people will be running. Rather, look at qwen/qwq, gemma and a number of other smaller ones
minus-squareParetoOptimalDev@lemmy.todaylinkfedilinkarrow-up0·1 month 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-up0·edit-21 month agoThe hell is v3 32b. Are you talking about a distill
minus-squarevintageballs@feddit.orglinkfedilinkDeutscharrow-up1·1 month agoThey probably confused the R1 Qwen distill with something else. Afaik there is no 32b model from DeepSeek directly.
Deepseek is an absolutely massive model, it’s not the one people will be running. Rather, look at qwen/qwq, gemma and a number of other smaller ones
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.