Alibaba Cloud, the digital technology and intelligence backbone of Alibaba Group, today announced the launch of Tongyi Qianwen 2.0, its latest large language model (LLM), along with new industry-specific models at its annual flagship tech event Apsara Conference. This release signifies another significant progress in Alibaba Cloud’s pursuit of cutting-edge AI innovation and its ongoing commitment to fuel digital transformation in businesses.
In response to the surging demand for powerful infrastructure supporting generative AI development, Alibaba Cloud also unveiled an AI model training platform and a series of innovative cloud products. New initiatives to bolster the open-source AI model community and global AI startups were also unveiled during the Conference. These resources are designed to enable customers, partners, developers, startups, and the wider community to fully harness the potential of generative AI.
“Currently, 80% of China’s technology companies and half of large model companies run on Alibaba Cloud. We aim to be the most open cloud in the era of AI. We hope that through this cloud, it will become easier and affordable for everyone to develop and use AI, so we can help, especially small and medium-sized enterprises, to turn AI into huge productivity,” said Joe Tsai, Chairman of Alibaba Group.
Tongyi Qianwen 2.0: A Leap Forward in LLMs
Tongyi Qianwen 2.0, a generic LLM with a few hundreds of billions of parameters, represents a substantial upgrade from its predecessor launched in April. The model successfully exceeds some leading LLMs in benchmarks across domains ranging from language understanding, arithmetic problem solving, to question-answering.
With an expanded model size and improved alignment technique, Tongyi Qianwen 2.0 demonstrates remarkable capabilities in understanding complex instructions, copywriting, reasoning, memorizing, and preventing hallucinations. The model is now accessible to the public via its website and mobile applications, and available for developers through APIs.
In benchmarks including MMLU(multi-task language understanding), GSM8K (arithmetic problem solving), ARC-C (question answering), BBH (multi-task language understanding) and Math (math word problem solving), Tongyi Qianwen 2.0 surpassed some leading large language models
Industry-specific Models: AI for Enhanced Business Performance
In addition, Alibaba Cloud also released new industry-specific models to help businesses unlock the transformative potential of generative AI across sectors for enhanced business performance. These sectors include customer support, legal counselling, healthcare, finance, documentation management, audio & video management, code development, and character creation.
“Large language models hold immense potential to revolutionize industries. We’re committed to using cutting-edge technologies, including generative AI, to help our customers capture the growth momentum forward,” said Jingren Zhou, CTO of Alibaba Cloud. “To help businesses better reap the benefits of generative AI in a cost-effective way, we are launching a more powerful foundation model as well as industry-specific models to tackle domain-specific challenges. Through these initiatives, we hope that our proprietary models can deliver real values to our customers to improve operation efficiency and stay competitive.”
GenAI Service Platform: A Comprehensive AI Model Building Platform
To cater to the burgeoning demand for generative AI, Alibaba Cloud introduced GenAI Service Platform, an all-in-one AI model building platform aimed at streamlining the model development and application building process. The platform offers a comprehensive suite of tools for data management, model deployment and valuation, and prompt engineering, making it easier for enterprises of all sectors to develop their enterprise-specific AI models.
Innovative Cloud Products to Drive Generative AI Development
To support the increasing computing and data processing needs required in generative AI development, Alibaba Cloud has launched a series of innovative cloud products across machine learning, container and database.
- To improve efficiency of LLM models training and inference, Alibaba Cloud has upgraded its Platform for Artificial Intelligence (PAI), featuring enhanced computing power, network, storage, computing, container, model training and inference capabilities. AI and metaverse technology firm Futureverse has been training its text-to-music generation model on PAI leveraging its high performance and easy-to-scale capabilities to enhance effectiveness and efficiency.
- To expedite the process of building customized generative AI applications, Alibaba Cloud has enhanced its full range of database solutions – including cloud-native database PolarDB, cloud-native data warehouse AnalyticDB, and cloud-native multi-model database Lindorm – with its proprietary vector engine. It has also introduced LLM-powered database ecosystem tools, including the intelligent data analytics assistant Data Management Service (DMS) Data Copilot.
- To enable a more scalable and efficient cloud infrastructure for application development, Alibaba Cloud unveiled Alibaba Cloud Container Compute Service (ACS), a pioneering serverless solution that integrates container services with its cloud computing resources. ACS enables businesses and developers to deploy the resources of Kubernetes, a container-centric management software, in a more cost-effective and user-friendly way.
Community Programs to Foster AI Development
In a bid to support the broader AI community, Alibaba Cloud has pledged to contribute more to open-source communities, including ModelScope, the company-initiated AI community with over 100 million model downloads since its inception a year ago. The cloud pioneer plans to opensource its 72 billion-parameter version of Tongyi Qianwen later this year.
Alibaba Cloud also announced the Alibaba Cloud Startup Catalyst Program, offering cloud computing resources, including cloud credit support of up to USD $120,000, free online learning memberships, and networking opportunities to promising global startups including AI companies.