Reproduce “Rethinking Vision Architectures with Gated Linear Attention and KAN” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Ali Mehizel, Oussama Khaldi
LKAT combines chunkwise GLA token mixing with a two-layer RBF-KAN for channel mixing, evaluated by a fused Triton kernel. On ImageNet-100, LKAT-B achieves 80.14% Top-1 accuracy, exceeding listed Base ViT, ViT-5, and Mixer models, but not ViT-S/16. This comparison uses more parameters and GFLOPs than those Base comparators and does not isolate each component's contribution.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.