AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

Do We Really Need Permutations? Impact of Width Expansion on Linear Mode Connectivity

Akira Ito, Masanori Yamada, Daiki Chijiwa, Atsutoshi Kumagai

ICLR 2026 · Architectures and efficiency › RNNs and normalization · approach: Symmetry and equivariance in loss landscapes

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 width expansion to independently trained models (no permutation)weight-averaged merged model · MNIST/FMNIST/CIFAR-10merged test accuracyImproved
add inverse-temperature calibration of the softmaxmerged model of wide networks · MNIST/FMNIST/CIFAR-10test loss barrierImproved
add weight-matching permutation to wide merged MLP (ablation)MLP merge · FMNISTaccuracy barrier (%): 2.467 -> 4.925Hurt
add weight-matching permutation to wide merged VGG-11 (ablation)VGG-11 merge · CIFAR-10accuracy barrier (%): 1.308 -> 7.000Hurt
add weight-matching permutation to wide merged ResNet-20 (ablation)ResNet-20 merge · CIFAR-10accuracy barrier (%): 2.694 -> 5.135Hurt
add weight-matching permutation to wide merged MLP (ablation)MLP merge · MNISTaccuracy barrier (%): 0.519 -> -0.027No clear effect
add weight-matching permutation to wide merged VGG-11 (ablation)VGG-11 merge · CIFAR-10calibrated loss barrier: 0.066 -> 0.177Hurt
add weight-matching permutation to wide merged ResNet-20 (ablation)ResNet-20 merge · CIFAR-10calibrated loss barrier: 0.087 -> 0.173Hurt
add random permutation before merging wide models (ablation)merged wide models · lambda=1/2test accuracyNo clear effect
add weak weight decay 10^-4 instead of 0.003 (ablation)VGG-11 and ResNet-20 merge · CIFAR-10merge barrier / LEWCHurt

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