@sun sorry if you've answered this already, but what model/editor are you using? I've tried llama3.2 70b and qwen2.5 32b and have got less-than-useful results.
(That and there's just not a good integration for vim)
@VD15 @sun try some of the new distilled Deepseek R1 models if you want it local
@matrix @sun Thanks, I'll give them a shot
@VD15 @sun Obviously a Chinese model
@VD15 @sun What's interesting is that when I ask it about race and IQ it reasons about not responding, unlike in Tiananmen Square
@matrix @sun Eh, there's uncensored forks of llama if I want it to be edgy. Sadly, those forks are dogshit at code generation.
Mainly gaming/nerd instance for people who value free speech. Everyone is welcome.
@matrix @sun Thanks, I'll give them a shot