arXiv · AI· Junming Lin, Yuxuan Wang, Zhenxin Lei, Yuxin Liu, Ruixun Liu, Yinsong Yan, Ling Wang, Minghao Han, Yunfei Chu, Shun Lei, Xueyao Zhang, Qize Yang, Jin Xu, Yiwu Zhong·· 1 天前AI 评分35
OmniReasoning：推动视听联合推理的边界
OmniReasoning: Pushing the Limits of Audio-Visual Joint Reasoning
AI 导读
OmniReasoning团队提出基准测试与模型OmniReasoning-30B-A3B，以解决视听联合推理能力不足的问题。 该模型在OmniReasoningBench上取得42.5%的成绩，较基座模型Qwen3-Omni-30B-A3B-Thinking提升9.3个百分点。研究旨在推动全模态联合推理的未来发展。
来源：arXiv · AI · arxiv.org