List your company

Multi-image difference analysis on MMXU (test)

23.4Worsen RateIDEFICS2
21.48834.39447.360.206Feb 17, 2025

Evaluation results

MethodLinks
IDEFICS2Size=8B, Source=Open2025.0223.45766.849.1
DeepSeek-VLSize=1.3B, Source=Open2025.0229.229.428.829.1
Qwen2-VLSize=7B, Source=Open2025.0233.155.45045.8
DeepSeek-VLSize=7B, Source=Open2025.0238.93160.643.5
InternVL2Size=8B, Source=Open2025.0242.348.349.546.7
InternVL2Size=4B, Source=Open2025.0247.657.131.545.4
GPT-4oSource=Closed2025.024867.555.957.1
Claude-3.5Source=Closed2025.0249.451.849.350.2
InternVL2Size=1B, Source=Open2025.0255.321.411.329.3
InternVL2Size=2B, Source=Open2025.025618.255.943.4
Llama3.2-visionSize=11B, Source=Open2025.0259.660.835.652
Qwen2-VLSize=2B, Source=Open2025.0271.225.828.441.8