>want to try an llm node in comfy
>none of the ones i found work
>find one that works with ollama
>cant do vision because lol reasons
>okay.jpg
>build my own gguf to test which takes an hour for no good reason
>lol nope
>try another node
>it's in chinese lol
>doesnt work anyway lol
>try yet another
>also chinese
>out of date, and no longer works
>lol
>finally find one that kinda sorta works with llama.cpp
>have to rebuild llama.cpp to use gpu
>lol
another productive day gone to waste thanks to comfyui
>inb4 it's the nodes
no, it's comfy BECAUSE it uses stupid node architecture which none of the "developers" of nodes bother to keep up with
but hey, it kinda works now