Qwen发布Qwen2.5-VL-32B多模态模型，性能超越72B大模型
2025-03-25 03:36:31
据Qwen团队公告，Qwen2.5-VL-32B-Instruct模型正式开源，具备32B参数规模，在图像理解、数学推理与文本生成等任务中展现优异性能。该模型通过强化学习进一步优化，回复更符合人类偏好，在MMMU、MathVista等多模态评测中超越此前发布的72B模型。
Disclaimer:
1. The information provided does not constitute investment advice. Investors should make independent decisions and bear all risks themselves.
2. The copyright of this content belongs to the original author. The views expressed herein are solely those of the author and do not represent the stance or position of this website.
