Transcoders Find Interpretable LLM Feature Circuits
Paper
•
2406.11944
•
Published
•
2
Arthur Conmy (GDM Account)
ArthurConmyGDM
AI & ML interests
Interpretability, AI Safety, AI Alignment
Recent Activity
new activity
1 day ago
google/gemma-scope-2b-pt-res:gemma-2-2b layer 20 SAE width 65k SAE seems very off
updated
a model
about 1 month ago
google/gemma-scope-2b-pt-res
new activity
about 2 months ago
google/gemma-scope-9b-pt-res:Removing SAEs with LR != 7e-5
Organizations
