README
Overview
MathV360K is proposed by Math-LLaVA: Bootstrapping Mathematical Reasoning for Multimodal Large Language Models, which consists 40K images from 24 datasets and 360K question-answer pairs. MathV360K is used to enhance the multimodal mathematical reasoning capabilities of MLLMs, achieving 46.6% accuracy on MathVista benchmark and 15.69% accuracy on MathVision dataset.
Paper or resources for more information: [Paper] [Code] [Model]
Source Data
Contributors
6 commits
Similar repos
Linked in READMEs
README
Overview
MathV360K is proposed by Math-LLaVA: Bootstrapping Mathematical Reasoning for Multimodal Large Language Models, which consists 40K images from 24 datasets and 360K question-answer pairs. MathV360K is used to enhance the multimodal mathematical reasoning capabilities of MLLMs, achieving 46.6% accuracy on MathVista benchmark and 15.69% accuracy on MathVision dataset.
Paper or resources for more information: [Paper] [Code] [Model]
Source Data
Similar repos
Linked in READMEs
Contributors
6 commits