AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

Tracr: Compiled Transformers as a Laboratory for Interpretability

David Lindner, Janos Kramar, Sebastian Farquhar, Matthew Rahtz, Thomas McGrath, Vladimir Mikulik

NeurIPS 2023 · Architectures and efficiency › Attention and transformers · approach: Transformer expressivity and learnability · 6 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
compress residual stream from D=14 to d=6 with learned projection WTracr frac_prevs model · frac_prevsloss: 14 -> 6No clear effect
reduce embedding dimension d below 6 (ablation)compressed frac_prevs model · frac_prevslossHurt
learned compression matrix instead of PCA with 8 components (ablation)compressed frac_prevs model · frac_prevsretained task-relevant featuresImproved
compress sort_unique with gradient descentTracr sort_unique model · sort_uniquelayer-output cosine similarity: 0.8Hurt

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