A self-supervised contrastive triplets plus weak-to-strong augmented GRPO training method improves multi-image reasoning in Qwen2.5-VL-7B.
Are we on the right way for evaluating large vision-language models?NeurIPS,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
unclear 1representative citing papers
citing papers explorer
-
MiCo: Multi-image Contrast for Reinforcement Visual Reasoning
A self-supervised contrastive triplets plus weak-to-strong augmented GRPO training method improves multi-image reasoning in Qwen2.5-VL-7B.