Alibaba Cloud, the digital technology and intelligence backbone of Alibaba Group, today unveiled its latest AI image generation model, Tongyi Wanxiang (‘Wanxiang’ means ‘ten thousand images’) at the World Artificial Intelligence Conference 2023 The cutting-edge generative AI model is now available for business customers in China for beta testing.
In addition, the cloud pioneer announced the launch of ModelScopeGPT, a versatile framework designed to help users perform complex and specialized AI tasks across the domains of language, vision, and speech through using different AI models in ModelScope. ModelScope is an open-source Model-as-a-Service (MaaS) platform introduced by Alibaba Cloud last year, with over 900 AI models.
“Tongyi Wanxiang represents an important milestone in our pursuit of advanced generative AI models as we continue to explore paradigm-shifting technologies that empower businesses and communities to unleash greater creativity. and productivity,” said Jingren Zhou, CTO of Alibaba Cloud Intelligence.
“With the release of Tongyi Wanxiang, high-quality generative AI imagery will become more accessible, facilitating the development of innovative AI art and creative expressions for enterprises in many sectors, including e-commerce, gaming, design and advertising.”
Introducing Tongyi Wanxiang for Image Creation
The generative AI model is adept at handling various tasks, responding to text prompts in Chinese and English to create detailed images in various styles, ranging from watercolors, oil and Chinese paintings to animation, sketch, flat illustration. , and 3D cartoons. In addition, the model can transform any image into a new one with a similar style and style images by transferring the style, preserving the content of the original image while applying the visual style of another image.
Powered by Alibaba Cloud technologies that advance knowledge management, visual AI and natural language processing (NLP), the model uses multilingual materials for enhanced training. It boasts a strong semantic understanding capability, resulting in highly accurate and contextually relevant image creation.
In addition, by optimizing the high-resolution diffusion process based on the signal-to-noise ratio, the model can strike a balance between compositional accuracy and detail sharpness while improving its ability to generate high contrast, nice looking images with clean backgrounds.
Tongyi Wanxiang was created using Composer, Alibaba Cloud’s proprietary large-scale model that provides greater control over the final image output, such as spatial layout and palette, while maintaining image synthesis quality and creativity.
Examples of text-to-image creation by Tongyi Wanxiang:
Prompt: Picture a twilight cityscape, a world that combines modern architecture with evocative anime aesthetics.
Prompt: Beautiful nature superimposed on an infinite loop sign with bright colors
Prompt: Immersive, attractive, grayscale coloring, with a tiger in a peaceful forest mandala.
The image is composed of lines and brushstrokes.
Invitation: A six-year-old girl with a beautiful and beautiful Chinese style
Hanfu shown in front of clothes rack, medium close-up, 85mm lens
Please watch demo video of Tongyi Wanxiang here: https://www.alizila.com/video/wach-how-alibaba-tongyi-wanxiang-creates-generative-ai-image/
Chinese business customers can apply for beta testing of Tongyi Wanxiang at: https://wanxiang.aliyun.com/
ModelScopeGPT Launched for Sophisticated AI Tasks
Alibaba Cloud also unveiled ModelScopeGPT (https://modelscope.cn/studios/damo/ModelScopeGPT/summary), a powerful framework designed to harness the power of Large Language Models (LLM) available on the platform. ModelScopeGPT will use LLMs as a controller to connect a wide range of domain-specific expert models to the ModelScope open-source community. Built within the rich Model-as-a-Service ecosystem, ModelScopeGPT leverages the various AI capabilities offered by Alibaba Cloud. Businesses and developers can use ModelScopeGPT for free to access and implement the most suitable models for performing sophisticated AI tasks based on users’ requests, such as creating multilingual videos. .
Alibaba Cloud launched its LLM named Tongyi Qianwen in April, and it plans to integrate LLM with various Alibaba businesses to improve user experience in the near future. The company’s customers and developers will also have access to the model to create customized AI features in a cost-effective manner. Since the launch of the model, more than 300,000 beta testing requests have been received from businesses from a wide range of sectors, including fintech, electronics, transport, fashion and dairy.
Tongyi Qianwen is also integrated with Alibaba Cloud’s intelligent assistant, Tingwu, which enables the assistant to understand and analyze multimedia content with a high level of accuracy and efficiency. More than 360,000 users have accessed the AI-powered assistant since its launch.
AI Hackathon Competition to Inspire Innovation
ModelScope also hosted China’s first ever AI Hackathon to accelerate industrial applications of AI models, with cash prizes and funding opportunities from leading venture capital firms as incentives .
From over 300 participating teams, 56 teams made it to the final round. Participants compete for the grand prize on two tracks. One is to modify a large language model to solve a real life problem. The second is to use existing pretrained models to complete an assigned task, such as text-to-image generation or to build an LLM-powered autonomous agent to use the right models for specific tasks.
“By hosting competitions and other community events, we want to engage with more developers and entrepreneurs, and encourage them to bring their ideas to life, unlock productivity, and create more versatile AI that is changing and shaping the future of our industries,” said Jingren Zhou.