AI Paper Atlas

Research map / Learning theory and algorithms / Combinatorial and search optimization

The Clock and the Pizza: Two Stories in Mechanistic Explanation of Neural Networks

Ziqian Zhong, Ziming Liu, Max Tegmark, Jacob Andreas

NeurIPS 2023 · Learning theory and algorithms › Combinatorial and search optimization · approach: Constrained neural network optimization · 11 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
model with constant attention (no attention) uses Pizza instead of Clock formula to explain logitsModel A · modular addition p=59fraction of variance explained: 75.41 -> 99.18Improved
keep three circles (6 PCs) instead of one circle (2 PCs) in embeddings (ablation)Model A · modular addition circle isolationoverall accuracy: 32.8 -> 91.4Improved
add accompanying pizzas (circles #4-#6) to the embedding (ablation)Model A · modular additionaccuracy: 99.7 -> 100Improved
increase attention rate (attention strength) in one-layer transformer (ablation)one-layer transformer width 128 · modular additionalgorithm usedMixed
increase model width (ablation)one-layer transformer · modular additionattention-rate phase boundaryMixed
increase network depth to 2-4 layers (ablation)transformers · modular additionfraction of circular models: 34.31% (1 layer) -> 9.95%, 11.55%, 6.08%Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy