AnE: Pushing the Reasoning Frontier of Multimodal LLMs via Anchor Evolution

Benchmark Model Rank Results
visual-question-answering-vqa-on-mmeAnE-3rd#4Score: 2457.5
mathematical-reasoning-on-mathverseAnE-3rd#10Accuracy: 62.3
multimodal-reasoning-on-mathvistaAnE-3rd#25Accuracy: 81.2
multimodal-reasoning-on-mathvisionAnE-3rd#36Accuracy: 43.9
vision-language-understanding-on-mmstarAnE-3rd#42Accuracy: 69.9
multimodal-reasoning-on-mmmuAnE-3rd#64Accuracy: 58.1