An Alibaba Group sign is seen at the World Artificial Intelligence Conference in Shanghai, July 6, 2023.
Aly Song | Reuters
Alibaba
on Friday unveiled its latest artificial intelligence (AI) models, Qwen-VL and Qwen-VL-Chat, that are designed to surpass its previous products in image understanding and converse in more complex ways. This move comes as the global competition to lead in AI technology intensifies.
The Chinese tech giant emphasizes that the new models will be open source, allowing global researchers, academics, and companies to utilize them for their own AI applications without the need to train their own systems. This not only saves time but also reduces expenses.
According to Alibaba, Qwen-VL is capable of responding to open-ended queries related to various images and generating captions for pictures. On the other hand, Qwen-VL-Chat is designed for more intricate interactions, including comparing multiple image inputs and answering multiple rounds of questions. This model can also write stories and create images based on user-provided photos, as well as solve mathematical equations shown in pictures.
An interesting example provided by Alibaba involves an image of a hospital sign in the Chinese language. The AI is able to interpret the image and answer questions about the locations of specific hospital departments.
While text-based generative AI has been prevalent, the latest version of OpenAI’s ChatGPT, similar to Qwen-VL-Chat, now possesses the capability to understand images and respond in text.
Alibaba’s new models, Qwen-VL and Qwen-VL-Chat, are built upon the company’s Tongyi Qianwen large language model (LLM), which was released earlier this year. LLMs are AI models trained on vast amounts of data and serve as the foundation for chatbot applications.
In addition, Alibaba recently open sourced two other AI models, aiming to attract more users for its AI model and drive growth in its cloud division, as the company prepares for an upcoming IPO.
Denial of responsibility! VigourTimes is an automatic aggregator of Global media. In each content, the hyperlink to the primary source is specified. All trademarks belong to their rightful owners, and all materials to their authors. For any complaint, please reach us at – [email protected]. We will take necessary action within 24 hours.
