AI RESEARCH
Interpreting and Steering State-Space Models via Activation Subspace Bottlenecks
arXiv CS.LG
•
ArXi:2602.22719v2 Announce Type: replace State-space models (SSMs) have emerged as an efficient strategy for building powerful language models, avoiding the quadratic complexity of computing attention in transformers. Despite their promise, the interpretability and steerability of modern SSMs remain relatively underexplored. We take a major step in this direction by identifying activation subspace bottlenecks in the Mamba family of SSM models using tools from mechanistic interpretability. We then.