The beautiful 36+59 attribution graph comes with fine print
JacobHHilton · x · 2026-10-17
Jacob Hilton highlights the elegant simplified attribution graph from Anthropic's paper showing how Claude 3.5 Haiku computes 36+59: input features feed separable pathways through add-function and lookup-table features into sum features (mod 10, mod 100, approximate magnitude). It's "super cool" we get any insight, he says—but there's fine print: this is only the simplified version, setting up his reveal of the 88-node detailed graph.
Related event: ARC's Jacob Hilton Sparks Debate: Interpretability Far From Solved(22 posts)→
More from Research
- Anthropic researchers: superposition isn't the blocker, interpretability is more tractable than assumed — akbirthko · 2026-10-17
- GEMMU benchmark exposes 3 pitfalls that made multimodal unlearning look better than it is — _vztu · 2026-10-17
- Parser-free code index for AI agents cuts grep bills 42% with 4x smaller tool outputs — microbass · 2026-10-17
- Quiz: how many trainable params does GPT-2 use just for embed and unembed? — Sauers_ · 2026-10-17
- Google grounds AI agents in wet-lab robotics, discovers nanomolar cancer protein binders — davidpattersonx · 2026-10-17
- Paper stuck at JMLR for over a year: researcher shares where to submit instead of TMLR — avt_im · 2026-10-17