AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

Symbol-Equivariant Recurrent Reasoning Models

Richard Freinschlag, Timo Bertram, Erich Kobler, Andreas Mayr, Günter Klambauer

ICML 2026 · Architectures and efficiency › RNNs and normalization · approach: Brain-inspired recurrent architectures · 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 symbol-equivariant layers (symbol-dimension attention) to the recurrent reasoning modelSE-RRM vs TRM · 9x9 Sudokufully solved rate (%): 71.94 -> 93.73Improved
add symbol equivariance so unseen symbols can be embedded at test timeSE-RRM trained on 9x9 · 4x4 Sudoku zero-shotfully solved rate (%): 0 -> 95.46Improved
apply SE-RRM to larger grids without trainingSE-RRM trained on 9x9 · 16x16 and 25x25 Sudokufully solved rate (%): 0 FSR; GPA 51.95 and 31.49No clear effect
increase inference deep supervision steps from 1 to 128SE-RRM · 9x9 Sudokufully solved rate (%): 16.05 -> 98.84Improved
replace extensive color augmentation with only 8 dihedral augmentations using SE-RRMSE-RRM · ARC-AGI-1pass@2 (%): TRM 44.6 -> SE-RRM 45.3No clear effect
use SE-RRM instead of TRM on ARC-AGI-2SE-RRM · ARC-AGI-2pass@2 (%): 7.8 -> 7.1No clear effect
use SE-RRM with symbol equivariance broken (separate symbol embeddings) instead of TRMSE-RRM · Maze-hardfully solved rate (%): 85.3 -> 88.8Improved
use SE-RRM instead of GPT-OSS-20B on 9x9 SudokuSE-RRM · 9x9 Sudokufully solved rate (%): 21.67 -> 93.73Improved

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