2026-10-27 03:15 UTC
I got reassurance that perhaps training smol models to be challenge experts was a good path. I don't know why I stopped, I think I got a 72b and was like shiney. Also maybe GPU but then I have my computer. I didn't use qwen lol though can get far doing it. I think I just maybe drew the wrong conclusions. I would need to do more research on it and maybe see if mistakes were made
Replies (1)
-
@ChickenPwny@infosec.exchange 2026-10-27 03:20
I wouldn't say frontier models have human causiniousness but it is something or so clever what is the difference? It could be a um magic trick or perhaps the inverse is true. A ghost we put in a. Shell. The more you dig into them and play with stuff it gets weird