Chasing the Counting Manifold in Open LLMs
📚
25
Counting manifolds in open LLMs from behavior to SAEs.
Daniil Gavrilov
kefirski
AI & ML interests
hustle
Recent Activity
authored a paper 9 days ago
Unstable Features, Reproducible Subspaces: Understanding Seed Dependence in Sparse Autoencoders authored a paper 16 days ago
Interpreting and Steering a Text-to-Speech Language Model with Sparse Autoencoders upvoted a paper 17 days ago
Interpreting and Steering a Text-to-Speech Language Model with Sparse AutoencodersOrganizations
