Beijing on December 20, Linux Foundation's top foundations for the field of AI --LF AI Foundation (Linux Foundation Artificial Intelligence Foundation) officially announced that Tencent open source projects Angel graduated from LF AI Foundation, is the first from China LF AI graduate Foundation open source projects. This means that, Angel recognized global technology experts, one of the world's top AI open source projects.
Ibrahim Haddad, executive director of LF AI Foundation, said: "Angel from hatching to the process of graduation, we can see the stunning Angel in terms of new features and application scenarios perfect landing speed with the release of version 3.0, Angel in function. applicability and contribution to the community and so have very good performance. this is the whole AI open source community has taken a big step in the direction of maturity and have the industrial capacity. "
Angel AI is Tencent's first open source project, launched in late 2016, 2017 revenue. As the third generation of machine learning for high-performance computing platform, Angel is committed to solving large sparse data model training and large-scale map data analysis problems. Tencent become one of the founding members of LF AI Platinum Foundation in 2018, and in the same year contributed to the Angel Foundation open source project.
Vice president of Tencent cloud, Jiang Jie, general manager of Tencent data platform, said: "Tencent is committed to open source, and open-source developers to build eco realize sharing of results, has become the most comprehensive open source big data field of company we are pleased to see. Angel to successfully graduate from LF AI Foundation, under the guidance and developers to build the foundation to complete the technological leap from a single platform to model training full-stack machine learning platform, which also combines with Tencent large batches of data streams, ABC closely fit the future direction of integration. graduation is a new beginning Angel, in the future we will further open Angel core competence in the field of view of the federal computing and learning. "
According to LF AI Foundation director, Tencent AI expert Xiao Han introduced, LF AI Foundation have very strict rules on the graduation process open source project, based on technical content, open source ecosystem, community interaction and other dimensions of the project, a rigorous assessment of project maturity the final vote by the Board of Directors can be allowed to graduate. Graduate from LF AI, mean that the project has been recognized by global technology experts as the most top AI open source projects.
Angel in the incubation process in the Foundation's rapid development, and completed from version 2.0 to version 3.0 of the cross, from a simple model training system evolved to include full-stack machine learning platform from automatic feature to the model engineering services.
Angel characterized Spark developed based engineering modules, enhanced features Spark selection function, wherein the cross and re-use the index to achieve the automatic feature generation. These components can be seamlessly integrated into Spark pipeline. In order to make the whole system more intelligent, Angel 3.0 adds ultra-parameter adjustment function.
In the model service, Angel 3.0 provides components Angel Serving a cross-platform, not only to meet their own needs Angel, can also provide a model for other platforms services. In ecological terms, Angel also try to shared server parameters (PS) capabilities to other computing platform, we have completed the construction of Spark On Angel and PyTorch On Angel two platforms.
Angel characterized Spark developed based engineering modules, enhanced features Spark selection function, wherein the cross and re-use the index to achieve the automatic feature generation. These components can be seamlessly integrated into Spark pipeline. In order to make the whole system more intelligent, Angel 3.0 adds ultra-parameter adjustment function.
Platform provides service model. In ecological terms, Angel also try to shared server parameters (PS) capabilities to other computing platform, we have completed the construction of Spark On Angel and PyTorch On Angel two platforms.
In version 3.0, Angel focus on research and development of machine learning chart features including chart representation and neural network learning model, while providing pre-treatment, graph mining and other end data processing capabilities, has a rich library of algorithms, providing the same composition, FIG various isomers FIG FIGS algorithm and different types of computing paradigm. After an extensive practice Tencent internal business, Angel equally outstanding performance in the performance graph algorithms, such as node one billion, one hundred billion scale side view of the structure, when the algorithm can run at one-third of computing resources Spark GraphX, ten times processing performance.
In addition to improving the technical function, Angel in the ecological community are becoming increasingly sophisticated. It is understood that the entire industry Angel has more than 100 corporate and institutional users, including microblogging, Huawei, micro-public banks, millet, drops and other large Internet companies. Angel-based machine learning to build a one-stop application platform smart titanium TI, support many internal Tencent micro-channel products, including payments, Tencent advertising, etc., depending on the micro, but also by Tencent cloud services to provide open more industries and enterprises.
Up to now, Angel on GitHub has received 5500 Star, 1400 Fork, excellent performance in technology, applications, ecological and other aspects of recognition from the open source community. Earlier in the Tencent 2019 Techo developer conference, Jiang Jie also announced a formal open-source core resource management platform and distributed database TBase of TKE. Tencent is becoming the most comprehensive open source big data areas of the company.
Open Source collaborative technology strategy is one of the most important current Tencent. Chongqing Zhibo at this year's meeting, the Chairman of the Board and CEO of Tencent Ma specifically mentioned the importance of open source, he said, Tencent hope to put more power in the field of scientific research, through internal and external open source, etc., to actively participate. " science and technology to build a global community ".
As of November 2019, only on Github, Tencent open source 91 projects, there are more than 1000 contributors involved, get more than 260,000 Star number, contribute standings among the highest in the world Github company. Tencent also joined the Linux, Apache and other 9 major open source foundations, and the highest level of membership. Open Source Foundation, which specialized in the field of artificial intelligence LF AI Foundation is sponsored by the Linux Foundation, aims to promote and organize AI open-source platform and tools to participate in the construction of IP in access to intellectual property rights, create open-source ecosystem for sustainable development of AI .
2018, Tencent influence due to technical excellence in the field of open source AI, invited as an initial platinum member LF AI Foundation joined with the Board of Directors and Technical Advisory Committee seats, and high performance in the same year donated to the Foundation distributed machine learning platform Angel.
