Nick Cammarata proposes multi-resolution, no-pruning attribution graphs to fix mech interp errors
inductionheads · x · 2026-10-17
Responding to Neel Nanda's critique of neuron-based attribution graphs, Nick Cammarata argues pruning may be unnecessary: neurons cluster naturally, so an agentic multi-resolution approach can zoom in/out based on importance, keeping near-zero error where it matters. He admits frontier-scale measurements are lacking and teases results on ablating weak activations.
More from Research
- Terry Tao's Four Types of Mathematization: Heavy AI Use Needs Strong Human Understanding — ravithejads · 2026-10-17
- Princeton's LabOS, Featured in Nature: An AI Co-Scientist for Hypothesis and Experiment Design — MengdiWang10 · 2026-10-17
- 'Tip-of-the-tongue' queries: a blind spot for modern agentic retrieval — bclavie · 2026-10-17
- Tip-of-the-tongue queries emerge as a key test of agentic retrieval systems — ctjlewis · 2026-10-17
- micro1 launches LegalReasonBench to test models on attorney-authored legal matters — douwekiela · 2026-10-17
- LEASH: weak-to-strong escalation matches Claude Opus 5 at 19.2% lower cost on OSWorld — xwang_lk · 2026-10-17