List your company
ResearchFormerly WizWand

Visual Grounding on Out-of-Distribution Scatter Plot (test)

89.7AccuracyVRRL (Ours)
-1.09222.47946.0569.621Jul 2, 2026
Updated 4mo ago

Evaluation results

MethodLinks
VRRL (Ours)Evaluation Setting=RL, Model Backbone=Qwen2.5-VL-7B-Instruct2026.0789.7—
Multi-SFT -> GRPOEvaluation Setting=RL, Model Backbone=Qwen2.5-VL-7B-Instruct2026.0786—
Single-SFT -> GRPOEvaluation Setting=RL, Model Backbone=Qwen2.5-VL-7B-Instruct2026.0784.3—
Reflection TuningEvaluation Setting=SFT, Model Backbone=Qwen2.5-VL-7B-Instruct2026.0781.3—
Multi-SFTEvaluation Setting=SFT, Model Backbone=Qwen2.5-VL-7B-Instruct2026.0773.3—
Single-SFTEvaluation Setting=SFT, Model Backbone=Qwen2.5-VL-7B-Instruct2026.0768.9—
VRRL (Ours)Evaluation Setting=RL, Model Backbone=Qwen2.5-VL-3B-Instruct2026.0740.3—
Multi-SFT -> GRPOEvaluation Setting=RL, Model Backbone=Qwen2.5-VL-3B-Instruct2026.0737.2—
Single-SFT -> GRPOEvaluation Setting=RL, Model Backbone=Qwen2.5-VL-3B-Instruct2026.0734.7—
VL-Rethinker-32BEvaluation Setting=Zero-shot, Backbone=32B2026.0732.7—
Reflection TuningEvaluation Setting=SFT, Model Backbone=Qwen2.5-VL-3B-Instruct2026.0727.4—
Multi-SFTEvaluation Setting=SFT, Model Backbone=Qwen2.5-VL-3B-Instruct2026.0724.8—
Single-SFTEvaluation Setting=SFT, Model Backbone=Qwen2.5-VL-3B-Instruct2026.0723.8—
Qwen2.5-VL-7B*MultiEvaluation Setting=Zero-shot, Backbone=7B, Strategy=Multi-turn reflection2026.074.638.2
Qwen2.5-VL-7BSingleEvaluation Setting=Zero-shot, Backbone=7B, Strategy=Single-turn2026.074.5—
VL-Rethinker-7BEvaluation Setting=Zero-shot, Backbone=7B2026.073.7—
Qwen2.5-VL-3BSingleEvaluation Setting=Zero-shot, Backbone=3B, Strategy=Single-turn2026.072.9—
Qwen2.5-VL-3B*MultiEvaluation Setting=Zero-shot, Backbone=3B, Strategy=Multi-turn reflection2026.072.475.5