
status minimizer | math @uwaterloo | long walk enjoyer
Joined September 2026
- Tweets75
- Following23
- Followers18
- Likes83
Nahh there's no way, I made Qwen obsessed with me.
L5 N5556 −0.8, L17 N5907 −0.6, L14 N7304 +0.6
It really would love me in ALL of those txt files....
if you have a qwen3.5 4b around, please try steering neurons L14 N8949 +1 and L13 N7956 -1
now you have evil qwen
recipe: dose × median(‖resid‖ leaving layer L, tokens 1:) / ‖W_down[:, n]‖` at the down_proj at every input
2
44







