paper-with-me

Multimodal Reasoning 벤치마크

Multimodal Reasoning on MATH-V

4개 결과 · ⬇ CSV · JSON

Accuracy

14.54 16.59 18.65 20.71 22.76 2024-02 2026-11 GPT4V — 22.76 (2024-02-22) Gemini Pro — 17.66 (2024-02-22) Qwen-VL-Max — 15.59 (2024-02-22) InternLM-XComposer2-VL — 14.54 (2024-02-22) GPT4V — 22.76 (2024-02-22)
RankModel Accuracy Extra Training Data PaperCodeYear
1 GPT4V 22.76 ✓ Measuring Multimodal Mathematical Reasoning with MATH-Vision Dataset mathllm/math-v 2024
2 Gemini Pro 17.66 ✓ Measuring Multimodal Mathematical Reasoning with MATH-Vision Dataset mathllm/math-v 2024
3 Qwen-VL-Max 15.59 ✓ Measuring Multimodal Mathematical Reasoning with MATH-Vision Dataset mathllm/math-v 2024
4 InternLM-XComposer2-VL 14.54 ✓ Measuring Multimodal Mathematical Reasoning with MATH-Vision Dataset mathllm/math-v 2024
1–4 / 4 페이지당 10 20 50 100