AI Paper Atlas

Research map / Large language models / Agents, code and math

NoReGeo: Non-Reasoning Geometry Benchmark

Irina Abdullaeva, Anton Vasiliuk, Elizaveta Goncharova, Temurbek Rahmatullaev, Zagorulko Ivan, Maxim Kurkin, Andrey Kuznetsov

AAAI 2026 · Large language models › Agents, code and math · approach: Mathematical and structural reasoning benchmarks · 1 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
give full diagram images instead of dotted images to VLMs (ablation)VLMs · NoReGeo, basic coordinate tasks (mean over models)accuracy (%): 19.2 -> 53.4Improved
give full images instead of dotted images (ablation)Qwen2.5-VL-7B-Instruct · NoReGeo numeric basic coordinate tasksaccuracy (%): 3.4 -> 79.0Improved
give full images instead of text-only for unstable circle properties (ablation)Qwen2.5-VL-7B-Instruct · NoReGeo unstable simple circle propertiesaccuracy (%): 0.0 -> 99.0Improved
give full images instead of dotted images for parallelism, perpendicularity and geometric transformations (ablation)VLMs · NoReGeo PAP and GTR categoriesaccuracy (%)No clear effect
give full images instead of dotted images (ablation)InternVL-3, G-LLaVA-13, URSA · NoReGeoaccuracy (%)No clear effect
use math-specialized fine-tuned text-only modelsQwen2.5-Math-7B-Instruct · NoReGeo classification and unstable tasksaccuracy (%)Hurt
fine-tune a linear probe on full images instead of freezing the CLIP backbone (ablation)CLIP ViT-B/32 probe · full-image test, 8 binary tasksaccuracy (%): 92.6 -> 97.4Improved
train probe on full images, test on dot images (ablation)CLIP ViT-B/32 probe · dot-image testaccuracy (%): 97.4 -> 58.6Hurt
retrain probe on dots instead of full images (ablation)CLIP ViT-B/32 probe · dot-image testaccuracy (%): 58.6 -> 83.9Improved
train probe on one task family and evaluate on a related family (ablation)CLIP ViT-B/32 probe · cross-task transferaccuracy (%): 86.0 on right-angle transfer from parallel-line probesImproved
fine-tune models on geometry alone to gain native geometric understanding (ablation)LLMs and VLMs · NoReGeoaccuracy (%)No clear effect

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