Attribution graph for one addition has 88 nodes and only ~50% faithfulness
JacobHHilton · x · 2026-10-17
Jacob Hilton continues his critique of interpretability progress: the graph completeness score—an explicit faithfulness measure in Anthropic's paper—averages only 50% for the 88-node attribution graph (Figure 32). "If this is the cleaner half of what's going on, what does the rest look like?" He also notes he used an LLM to add hover labels so the full graph could be exported as a plain image, avoiding Anthropic's ad-hoc manual subgraph process.
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