AtP*: An efficient and scalable method for localizing LLM behaviour to components
Paper
•
2403.00745
•
Published
•
14
Neel Nanda
NeelNanda
AI & ML interests
Mechanistic Interpretability
Recent Activity
authored
a paper
18 days ago
Towards eliciting latent knowledge from LLMs with mechanistic
interpretability
authored
a paper
4 months ago
Open Problems in Mechanistic Interpretability
authored
a paper
7 months ago
Do I Know This Entity? Knowledge Awareness and Hallucinations in
Language Models
Organizations
