AI Paper Atlas

Research map / 3D vision / Face and shape

Gold-medalist Performance in Solving Olympiad Geometry with AlphaGeometry2

Yuri Chervonyi, Trieu H. Trinh, Miroslav Olšák, Xiaomeng Yang, Hoang H. Nguyen, Marcelo Menegali, Junehyuk Jung, Junsu Kim, Vikas Verma, Quoc V. Le, Thang Luong

JMLR 2025 · 3D vision › Face and shape · approach: Optimization-based shape matching · 97 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
upgrade AG1 to AG2 (new language, engine, LM, SKEST search)AlphaGeometry · IMO-AG-50solve rate: 54% -> 84%Improved
add multiple search trees with knowledge sharing (SKEST) (ablation)AG2 single search tree · IMO-AG-50problems solved: 38 -> 42Improved
add multiple search trees with knowledge sharing (SKEST) (ablation)AG2 single search tree · IMO-AG-30problems solved: 28 -> 30No clear effect
replace DDAR1 by DDAR2 (faster algorithm, C++)symbolic engine · 25 IMO problemsruntime (s): 1179.57 -> 3.44711Improved
replace DDAR1 by DDAR2 symbolic engineDDAR · IMO-AG-50problems solved: 14 -> 16Improved
add language model to symbolic engine (ablation)AG2 · IMO-AG-50problems solved: 16 -> 42Improved
sample with temperature 1.0 and 32 samples instead of greedy decoding (ablation)AG2 LM · 26 IMO problems needing auxiliary constructionsproblems solved: 2 -> 9Improved
use a custom small-vocabulary tokenizer vs large Gemini tokenizer (ablation)AG LM · IMO 2000-2024solve rateNo clear effect
train on natural language instead of AG domain language (ablation)AG LM · IMO 2000-2024solve rateNo clear effect
fine-tune math-pretrained 3.3B Gemini instead of training from scratch (ablation)AG LM · IMO-AG-50problems solvedNo clear effect
add image input of the diagram (multimodal) (ablation)AG2 LM alone · IMO problemssolve rateNo clear effect
increase beam size or samples beyond beam 128, depth 4, 32 samples (ablation)AG2 single search tree · IMO problemsproblems solvedNo 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