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)
2026-06-07 11:44:46 +08:00
committed by GitHub
parent 2e8dabcb88
commit 479f314504
1 changed files with 7 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files
README.md
+7 -2
View File
@@ -17,10 +17,15 @@ What if Lex Fridman is right?
> >
> -- Lex Fridman, to Eliezer Yudkowsky 3 h 18 min into [Lex Fridman Podcast #368, "Dangers of AI and the End of Human Civilization"](https://podscript.ai/podcasts/lex-fridman-podcast/368-eliezer-yudkowsky-dangers-of-ai-and-the-end-of-human-civilization/) (03:18:03) > -- Lex Fridman, to Eliezer Yudkowsky 3 h 18 min into [Lex Fridman Podcast #368, "Dangers of AI and the End of Human Civilization"](https://podscript.ai/podcasts/lex-fridman-podcast/368-eliezer-yudkowsky-dangers-of-ai-and-the-end-of-human-civilization/) (03:18:03)
## Steer
## Hypothesis Steering is interesting because it's and internal and unsupervised intervention. But it's often unreliable and incoherent. What is we can fix that?
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. ## Heal
### Hypothesis
Hypothesis: you can distill a steering vector into LoRA weights and "heal" the incoherency the vector injects by regularising the training. Then loop and see what multiple rounds give you.
In concrete terms In concrete terms
- We steer - We steer