VeriEvol: Scaling Multimodal Mathematical Reasoning via Verifiable Evol-Instruct
A novel framework called VeriEvol is introduced that addresses the challenge of scaling reinforcement learning for visual mathematical reasoning by ensuring reliable reward labels…
Hugging Face · Daily Papers
·Haoling Li, Kai Zheng
·
·▲ 3 upvotes
Este artigo está em destaque na seleção diária de papers do Hugging Face, curada pela comunidade de pesquisa em IA.
Autores: Haoling Li, Kai Zheng, Jie Wu, Can Xu, Qingfeng Sun, Han Hu
- 3 upvotes da comunidade
- Temas: reinforcement learning, visual mathematical reasoning, data scaling, reward labels, verifiable data-construction, prompt difficulty
Resumo
Resumo original (em inglês), extraído do paper:
A novel framework called VeriEvol is introduced that addresses the challenge of scaling reinforcement learning for visual mathematical reasoning by ensuring reliable reward labels through a two-axis approach that separates prompt difficulty from answer reliability, utilizing evolutionary operators and hypothesis testing verification to improve model performance and transparency.