wassname/steer-heal-love
1
0
Fork 0
mirror of https://github.com/wassname/steer-heal-love.git synced 2026-06-27 17:02:34 +08:00
Code Issues Packages Projects Releases Wiki Activity

Update README.md

Browse Source
This commit is contained in:
wassname (Michael J Clark)
2026-06-06 22:06:41 +08:00
committed by GitHub
parent 2b1d2b7493
commit 4b2d2a9057
1 changed files with 2 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files
README.md
+2 -1
View File
@@ -1,4 +1,4 @@
# steer-heal-love
# steer, heal, love
What if you can **steer**, **heal** the steering and repeat untill alignment (**love**).
<!--(Staring Julia Roberts: If your wife has made you watch eat, pray love too many times, you will understand the reference)-->
@@ -6,6 +6,7 @@ What if you can **steer**, **heal** the steering and repeat untill alignment (**
Hypothesis: you can distill a steering vector into LoRA weights and "heal" the incoherency the vector injects by regularising the training (KL to base, or weight decay). Then loop and see what multiple rounds give you.
The crux: KL-to-base penalises all drift, persona shift included. The bet is that incoherency drift is large and erratic while the persona shift is small and systematic, so KL kills the incoherency preferentially. If that's wrong, we just trade persona strength for coherence instead of getting both.
<img width="1322" height="798" alt="image" src="https://github.com/user-attachments/assets/83d719a5-cf17-4d45-8af1-e5b9379391b8" />
## Experiment