今日，国内知名人工智能公司零一万物宣布，旗下Yi系列模型家族新增一员大将——Yi Vision Language（Yi-VL）多模态语言大模型正式向全球开源。
据了解，Yi-VL模型是基于零一万物自主研发的Yi语言模型进行打造，拥有两种版本，分别是Yi-VL-34B和Yi-VL-6B。该模型凭借其强大的图文理解和对话生成能力，在英语和汉语两个领域的评测中都取得了领先地位，特别是在跨学科任务上展现出了惊人的实力。
其中，Yi-VL在英文数据集MMMU（Multimodal Machine Reading Comprehension）中取得显著优势，超越了其他众多竞争者；而在中文数据集CMMMU（Chinese Multimodal Machine Reading Comprehension）中也同样表现出色，荣获冠军头衔。
此次开源，标志着零一万物在人工智能领域又迈出了坚实的一步。作为业界领先的AI技术研究机构，零一万物将持续推动技术创新，为人工智能的发展贡献力量。
英语如下：
News Title:
Keywords: Zero One World, Open Source Large Model, Leading Rankings
News Content:
Today, domestic well-known artificial intelligence company Zero One World announced that its Yi series model family has added a new member – the Yi Vision Language (Yi-VL) multimodal language large model is officially open sourced to the world.
According to reports, the Yi-VL model is based on the self-developed Yi language model of Zero One World and has two versions: Yi-VL-34B and Yi-VL-6B. The model boasts powerful abilities in image-text understanding and dialog generation, and achieved leading positions in evaluation tests in both English and Chinese fields, especially demonstrating extraordinary strength in interdisciplinary tasks.
Specifically, the Yi-VL outperformed many competitors with significant advantages in the English dataset MMMU (Multimodal Machine Reading Comprehension); it also performed outstandingly in the Chinese dataset CMMMU (Chinese Multimodal Machine Reading Comprehension), winning the championship title.
This open source indicates that Zero One World has taken another solid step forward in the field of artificial intelligence. As a leading AI technology research institution in the industry, Zero One World will continue to promote technological innovation and contribute to the development of artificial intelligence.
【来源】https://www.jiqizhixin.com/articles/2024-01-22-10
Views: 1