List your company
ResearchFormerly WizWand

Instruction-guided image editing preference prediction on AURORA-Bench

63.62AccuracyEDITREWARD
29.102438.063747.02555.9863Sep 30, 2025
Updated 8mo ago

Evaluation results

MethodLinks
EDITREWARDBackbone=MiMo-VL-7B-SFT2025.0963.62
EDITREWARDBackbone=Qwen2.5-VL-7B2025.0959.5
ADIEEMethod Category=Open-Source VLMs2025.0955.56
GPT-4oMethod Category=Proprietary Models2025.0950.81
Gemini-2.5-FlashMethod Category=Proprietary Models2025.0947.63
GPT-5Method Category=Proprietary Models2025.0947.27
Gemini-2.0-FlashMethod Category=Proprietary Models2025.0944.31
Qwen2.5-VL-7B-InstMethod Category=Open-Source VLMs2025.0938.62
Qwen2.5-VL-32B-InstMethod Category=Open-Source VLMs2025.0937.06
RandomMethod Category=Baseline2025.0933.43
Qwen2.5-VL-3B-InstMethod Category=Open-Source VLMs2025.0930.69
MiMo-VL-7B-SFT-2508Method Category=Open-Source VLMs2025.0930.43