POPA Recon👀👀In the future, open-source models will eventually develop to the point where people can use uncensored local lm and
👀👀In the future, open-source models will eventually develop to the point where people can use uncensored local lm and I think the result of agent coding will be different from those who can't use it The only thing for consumers to do is to run a large local model If you have a lot of money, Mac Studio Ultra 512 is the best four exo clusters Next, for medium model speed, rtx 6000 pro maxq There is a dgx spot to the extent that the mid-size model just goes around, but for most of the agent coders, I think the cloud model subscription will be more leveraged
The excerpt is quoted from a public Telegram channel, which owns its text.