Skill-Decoupled On-Policy Expert Distillation for Unified Vision-Language Models
Open paper
Benchmark
Model
Rank
Results
visual-question-answering-vqa-on-countbenchqa
SCOPED-8B (extra data)
#3
Accuracy: 90.43
chart-question-answering-on-chartqa
SCOPED-8B (extra data)
#19
1:1 Accuracy: 88.28
Rank counts only results with a code link.