Home > Technology peripherals > AI > Shengsi Open Source Community Council was established, and the full-modal large model 'Zidong. Taichu 2.0' based on Shengsi AI framework was released

Shengsi Open Source Community Council was established, and the full-modal large model 'Zidong. Taichu 2.0' based on Shengsi AI framework was released

WBOY
Release: 2023-06-17 12:46:15
forward
1223 people have browsed it

Shengsi Open Source Community Council was established, and the full-modal large model Zidong. Taichu 2.0 based on Shengsi AI framework was released

Blue Whale TMT Channel News on June 16, including 18 AI companies, societies, universities and scientific research institutes jointly released the "Co-Building Artificial Intelligence Framework Ecological Initiative", Shengsi MindSpore Open Source Community Council was established, Shanghai Shengsi The AI ​​Framework & Large Model Innovation Center was launched and the first batch of 22 units settled in. The full-modal large model "Zidong. Taichu 2.0" was released.

"Zidong. Taichu 2.0" is a full-mode large-scale model based on Shengsi technology, released by the Institute of Automation, Chinese Academy of Sciences. Officials stated that the large model is an upgrade based on the 100-billion-parameter three-modal large model "Zidong. Taichu" 1.0. In addition to voice, images and text, it also adds video, sensor signals, 3D point clouds, etc. More modalities, the technical architecture realizes all-modal open access to structured and unstructured data, and is supported by key technologies such as cognitively enhanced multi-modal correlation, supporting multiple rounds of question and answer, text creation, and image generation. , 3D understanding, signal analysis and other cross-modal tasks, and has full-modal intelligent annotation capabilities, which can realize automatic discovery of unknown targets.

It is reported that based on "Zidong. Taichu" 2.0, the "Zidong. Taichu" open service platform with full-stack localization and low-code one-stop development has been launched, supporting one-click deployment of public clouds and private clouds.

According to reports, as of the end of May 2023, the download volume of Shengsi open source community has exceeded 4.74 million, with 13,000 community contributors, 5,500 service companies, cooperation with 240 scientific research institutes, and more than 400 open source models , published more than 900 top conference papers.

A total of 18 units, including AI chip companies, AI application companies, universities and academic institutions, were announced as the first batch of members of the Shengsi Community Council. According to reports, the council will jointly create an open, diverse and inclusive AI framework technology ecological environment based on AI core technology.

At the meeting, Ding Cheng, chairman of Shengsi Community, introduced four major innovation directions for the future, including technological innovation based on "building full-process enablement of large models and new scientific computing paradigms" to accelerate the incubation of new AI applications. ; Based on the operational innovation of "Large Model Platform, Study Group, Geek Week, and Global Open Source Ecosystem", community results can be promoted to users more efficiently; Based on "providing different training models for developers at different stages of learning, growth, and research" Talent innovation and cultivation of AI industry soil; ecological innovation based on "establishing industrial consortiums, industry zones and other new means to continue to improve diversified computing power and hardware support".

The above is the detailed content of Shengsi Open Source Community Council was established, and the full-modal large model 'Zidong. Taichu 2.0' based on Shengsi AI framework was released. For more information, please follow other related articles on the PHP Chinese website!

source:sohu.com
Statement of this Website
The content of this article is voluntarily contributed by netizens, and the copyright belongs to the original author. This site does not assume corresponding legal responsibility. If you find any content suspected of plagiarism or infringement, please contact admin@php.cn
Popular Tutorials
More>
Latest Downloads
More>
Web Effects
Website Source Code
Website Materials
Front End Template