Stanford's Matryoshka Attribution tops mech interp benchmark at 2.9x the runner-up
aryaman2020 · x · 2026-09-24
A Stanford team (including Chris Potts and Dan Jurafsky) released Matryoshka Attribution (MAttr), a new attribution method that uses gradient descent over a sigmoid top-k mask to identify which internal variables of a neural network are responsible for a behavior.
Key points:
- One training run yields the full sparsity curve: learned scores form a ranking rather than a fixed-budget circuit, so the entire circuit-precision-vs-sparsity curve comes from a single run.
- Universal API: learnscores takes any lossfn(mask) and a node count; adapters support nodes, edges, neurons, SAE latents and ViT patches.
- Results: #1 on the Mechanistic Interpretability Benchmark circuit track by a wide margin (2.9x the runner-up).
- Code, parameter-level attribution (with RL/SFT training code), and MIB node/edge circuit submissions are open-sourced.
More from Research
- MIT's Tegmark coins 'formal-washing': Signal protocol proof is far from complete — tegmark · 2026-11-12
- Signal formal verification hits 80%, Tegmark says Wired overstated completion — tegmark · 2026-11-12
- Category-theoretic framework asks: when does a physical system actually compute? — neurovium · 2026-11-12
- Radical Numerics brings genome language model Omnii-1 to the edge for biodefense — exnx · 2026-11-12
- Against 'verifiable vs. fuzzy': the real axis is how expensive tasks are to grade — dfrsrchtwts · 2026-11-12
- Toby Ord flags 'all-things-considered' AGI forecast table is actually conditional on no pacing — tobyordoxford · 2026-11-12