x
This website requires javascript to properly function. Consider activating javascript to get access to all site functionality.
AI ALIGNMENT FORUM
AF
Login
Considerations in diffuse control — AI Alignment Forum
Considerations in diffuse control
9
Methodological considerations in making malign initializations for control research
Alek Westover
,
Vivek Hebbar
,
Julian Stastny
10mo
0
2
Three visions for diffuse control
Alek Westover
8mo
0
18
Four Downsides of Training Policies Online
Alek Westover
,
egan
9mo
0
4
Theoretical predictions on the sample efficiency of training policies and activation monitors
Alek Westover
,
Vivek Hebbar
9mo
0
18
How will we do SFT on models with opaque reasoning?
Alek Westover
,
Vivek Hebbar
,
egan
8mo
0
17
Model organisms researchers should check whether high LRs defeat their model organisms
Dylan Xu
,
sebastian_prasanna
,
Alek Westover
,
Vivek Hebbar
,
Julian Stastny
6mo
0
28
How do LLMs generalize when we do training that is intuitively compatible with two off-distribution behaviors?
Dylan Xu
,
Alek Westover
,
Vivek Hebbar
,
sebastian_prasanna
,
frisby
,
Julian Stastny
6mo
0
7
How to reduce capability degradation from off-model SFT
Dylan Xu
,
sebastian_prasanna
,
Alek Westover
4mo
0
18
Advice for making robust-to-training model organisms
sebastian_prasanna
,
Alek Westover
,
Vivek Hebbar
,
Julian Stastny
,
Dylan Xu
5mo
1
23
Why does off-model SFT degrade capabilities?
sebastian_prasanna
,
Dylan Xu
,
Alek Westover
,
Julian Stastny
,
Vivek Hebbar
5mo
0
7
Malign initializations are more robust when the model can think better in the reasoning language than in the output language
Dylan Xu
,
sebastian_prasanna
,
Alek Westover
2mo
0