AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

On the Intrinsic Limits of Transformer Image Embeddings in Non-Solvable Spatial Reasoning

Siyi Lyu, Quan Liu, Feng Yan

ICML 2026 · Architectures and efficiency › Attention and transformers · approach: Efficient and robust vision transformers · 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
scale ViT from Base to Huge (N=5)ViT linear probe · LSA Level 3 (Non-Solvable)Ratio loss at N=5: 0.7494 -> 0.7003Improved
scale ViT from Base to Huge (N=20)ViT linear probe · LSA Level 3 (Non-Solvable)Ratio loss at N=20: 0.8807 -> 0.8608No clear effect
scale DINOv2 from Base to GiantDINOv2 linear probe · LSA Level 3 (Non-Solvable)Ratio loss at N=5: 0.5976 -> 0.5474Improved
scale ResNet from R-50 to R-152ResNet linear probe · LSA Level 3 (Non-Solvable)Ratio loss at N=20: 0.8386 -> 0.8269No clear effect
move from abelian translation (Level 1) to non-solvable 3D rotation (Level 3)ViT-Base linear probe · LSA, N=20Ratio loss: 0.1456 -> 0.8807Hurt
move from solvable SE(2) (Level 2) to non-solvable SO(3) (Level 3)DINOv2-Base linear probe · LSA, N=20Ratio loss: 0.3435 -> 0.7458Hurt
increase composition depth from N=5 to N=20ViT-Base linear probe · LSA Level 3 (Non-Solvable)Ratio loss: 0.7494 -> 0.8807Hurt
increase composition depth from N=5 to N=20ViT-Base linear probe · LSA Level 1 (Abelian)Ratio loss: 0.0567 -> 0.1456Hurt
use self-supervised DINOv2-Base instead of supervised ViT-BaseFrozen encoder · LSA Level 3 (Non-Solvable)Ratio loss at N=20: 0.8807 -> 0.7458Improved
rotate out-of-plane (Rot Y 30 deg) instead of in-plane (Rot Z 30 deg)DINOv2 OPA analysis · Atomic linearity (N=1)OPA R2: 0.3054 -> -0.0286Hurt

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