List your company

Multi-disciplinary Visual Question Answering on MME SUM

2,494Score (SUM)InternVL2.5
1,576.0961,814.3982,052.72,291.002Sep 30, 2024Oct 23, 2024Nov 16, 2024Dec 10, 2024Jan 2, 2025Jan 26, 2025Feb 19, 2025

Evaluation results

MethodLinks
InternVL2.5Number of Parameters=78B2025.022,494
Qwen2-VLNumber of Parameters=72B2025.022,483
Qwen2.5-VLNumber of Parameters=72B2025.022,448
Qwen2.5-VLNumber of Parameters=7B2025.022,347
GPT-4o 0513Model Version=05132025.022,328
Qwen2.5-VLNumber of Parameters=3B2025.022,157
Claude-3.5 Sonnet-0620Model Version=06202025.021,920
MM1.5 1B (MoE)Model Parameters=1B, Architecture=MoE, Decoding Strategy=Greedy2024.091,873
InternVL2 2BModel Parameters=2B, Decoding Strategy=Beam Search2024.091,864.3
MiniCPM-V2 3BModel Parameters=3B, Decoding Strategy=Beam Search2024.091,808.2
MM1.5 3BModel Parameters=3B, Decoding Strategy=Greedy2024.091,798
Phi-3-Vision 4BModel Parameters=4B, Decoding Strategy=Greedy2024.091,761.6
MM1.5 1BModel Parameters=1B, Decoding Strategy=Greedy2024.091,611.4