This website requires javascript to properly function. Consider activating javascript to get access to all site functionality.
AI ALIGNMENT FORUM
AF
Login
Connor Kissane
Posts
Sorted by New
33
SAEs are highly dataset dependent: a case study on the refusal direction
2mo
0
18
Open Source Replication of Anthropic’s Crosscoder paper for model-diffing
2mo
3
35
Base LLMs refuse too
3mo
10
28
SAEs (usually) Transfer Between Base and Chat Models
5mo
0
18
Attention Output SAEs Improve Circuit Analysis
6mo
0
34
We Inspected Every Head In GPT-2 Small using SAEs So You Don’t Have To
10mo
0
28
Attention SAEs Scale to GPT-2 Small
11mo
0
35
Sparse Autoencoders Work on Attention Layer Outputs
1y
3
Wiki Contributions
Comments
Sorted by
Newest