AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

PolySAE: Modeling Feature Interactions in Sparse Autoencoders via Polynomial Decoding

Panagiotis Koromilas, Andreas D. Demou, James Oldfield, Yannis Panagakis, Mihalis Nicolaou

ICML 2026 · Architectures and efficiency › RNNs and normalization · approach: Sparse dictionary representations · 0 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
add polynomial (quadratic and cubic) decoder termsTopK SAE on GPT-2 Small · SAEBench sparse probing K=1 (GPT-2 Small)mean F1 (%): 67.1 -> 77.9Improved
add polynomial (quadratic and cubic) decoder termsTopK SAE on Pythia-410M · SAEBench sparse probing K=1 (Pythia-410M)mean F1 (%): 71.2 -> 77.0Improved
add polynomial (quadratic and cubic) decoder termsTopK SAE on Pythia-1.4B · SAEBench sparse probing K=1 (Pythia-1.4B)mean F1 (%): 75.9 -> 81.9Improved
add polynomial (quadratic and cubic) decoder termsTopK SAE on Gemma-2-2B · SAEBench sparse probing K=1 (Gemma-2-2B)mean F1 (%): 67.7 -> 68.4Improved
add polynomial (quadratic and cubic) decoder termsBatchTopK SAE on GPT-2 Small · SAEBench sparse probing K=1 (GPT-2 Small)mean F1 (%): 65.7 -> 78.0Improved
add polynomial (quadratic and cubic) decoder termsBatchTopK SAE on Pythia-410M · SAEBench sparse probing K=1 (Pythia-410M)mean F1 (%): 65.0 -> 77.3Improved
add polynomial (quadratic and cubic) decoder termsBatchTopK SAE on Pythia-1.4B · SAEBench sparse probing K=1 (Pythia-1.4B)mean F1 (%): 64.6 -> 76.4Improved
add polynomial (quadratic and cubic) decoder termsBatchTopK SAE on Gemma-2-2B · SAEBench sparse probing K=1 (Gemma-2-2B)mean F1 (%): 64.8 -> 69.4Improved
add polynomial (quadratic and cubic) decoder termsMatryoshka SAE on GPT-2 Small · SAEBench sparse probing K=1 (GPT-2 Small)mean F1 (%): 65.7 -> 77.7Improved
add polynomial (quadratic and cubic) decoder termsMatryoshka SAE on Gemma-2-2B · SAEBench sparse probing K=1 (Gemma-2-2B)mean F1 (%): 60.9 -> 65.6Improved
add low-rank factorization to the polynomial decoder (ablation)polynomial decoder with shared projection · Architectural ablation (GPT-2 Small)mean F1 (%): 76.0 -> 75.0Mixed
add orthogonality constraint on the projection (ablation)low-rank polynomial decoder · Architectural ablation (GPT-2 Small)mean F1 (%): 75.0 -> 77.9Improved

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