OpenCSG(开放传神)完成数千万元天使轮融资
OpenCSG
开源、开放、共享
让/大/模/型/赋/能/每/一/个/人
9月伊始,北京开放传神科技有限公司(以下简称“OpenCSG”)获数千万元天使轮融资,本轮投资由 联想创投集团(以下简称“联想创投”),北京国信中数投资管理有限公司(以下简称“国信中数”)阶段性联合领投,源合资本担任财务顾问 。OpenCSG 表示本轮融资主要用于进一步加强产品创新及研发能力建设的投入,目前公司估值已达数亿元。
In early September, Beijing-based Kai Fang Chuan Shen Ltd Inc. , short for OpenCSG, closed tens of millions of yuan in angel funding,This round of investment was led by Lenovo Capital and Incubator Group, in conjunction with Digital China Investment Co.,Ltd, Primary Capital acted as financial advisor.OpenCSG stated that the primary purpose of this funding round is to further enhance product innovation and research and development capabilities. The current valuation of the company has reached several hundred million yuan.
OpenCSG 公司由原极狐 Gitlab 创始人兼 CEO 陈冉先生和一批志同道合的大模型技术专家,软件架构师,云计算和存储架构师,数据中心技术专家和开源社区布道者们共同发起并创建。
OpenCSG was initiated and founded by Mr. Chen Ran(Sam) the former CEO and founder of Jihu GitLab, along with a group of like-minded experts in LLMs technology, software architects, cloud computing and storage architects, data center technology specialists, and open-source community advocates.
创始人及联合创始团队具有连续成功创业经验,毕业于国内外知名高校,曾长期担任国内外软硬件企业和互联网企业的核心技术骨干与经营管理者,拥有丰富的人工智能技术、云原生技术和商业运营的经验。
The founder and the co-founding team have a track record of successful entrepreneurship, graduated from renowned universities both domestically and internationally, and have held key technical and managerial roles in domestic and international software and hardware companies, as well as internet enterprises. They possess extensive experience in artificial intelligence technology, cloud-native technology, and business operations.
公司致力于在全球范围内创建一个伟大的公司,让大模型驱动全民,从而推动技术加速创新。
The company is committed to establishing a great company on a global scale, with the aim of driving innovation by making LLMs accessible to everyone and accelerating technological advancements.
OpenCSG 中的 ‘Open’ 代表 ‘开源开放,人人贡献’, ‘C’ 代表 ‘Converged’,意思是整合和充分利用好混合资源。‘S’ 代表 ‘Software refined’,也就是软件被大模型重塑。‘G’ 代表 ‘Generative’,是大众化和民主化的生成式大模型。我们的愿景是“让每个行业,每个公司,每个人都拥有自己的模型”。
In OpenCSG, 'Open' stands for 'open source and open for all to contribute,' 'C' represents 'Converged,' signifying the integration and effective utilization of hybrid resources. 'S' represents 'Software refined,' meaning that software is reshaped by LLMs, ‘G' represents 'Generative,' which refers to democratized and accessible generative LLMs. Our vision is to "empower every industry, every company, and every individual to have their own LLMs."
01
投 资 方 简 介
联想创投 联想创投集团(Lenovo Capital and Incubator Group) 作为联想旗下的全球科技产业基金,旨在融汇联想全球资源,以投资和孵化为手段,布局科技的未来。联想创投专注于科技产业投资,围绕联想“端-边-云-网-智”新IT架构,支持联想集团服务导向转型和科技驱动创新战略,目前已布局了200多家创业企业,其中15家企业成功实现IPO,40余家企业成为相关领域的领军者,80余家企业获评专精特新,其中90%都是围绕核心科技布局,包括人工智能、芯片半导体、智能制造、机器人、虚实融合等。
Lenovo Capital and Incubator Group, as Lenovo's global technology industry fund,aims to leverage Lenovo's global resources to invest in and incubate future technology developments.
Lenovo Capital focuses on technology industry investments, aligning with Lenovo's "end-edge-cloud-network-intelligence" new IT architecture, supporting Lenovo Group's service-oriented transformation and technology-driven innovation strategy. They have invested in over 200 portfolio companies, with 15 of them have been successfully listed, more than 40 becoming industry leaders, and over 80 have been rated as national and provincial level Specialized, Refined, Differential and Innovational Enterprises.Around 90% of the portfolios are centered around core technologies, including artificial intelligence, semiconductor chips, smart manufacturing, robotics, and mixed reality, among others.
国信中数 由国家信息中心联合中国华融发起设立。公司秉承“国信中数 助企强国”的理念,依托国家信息中心的大数据资源优势和股东单位的资本优势,围绕数字中国建设的核心需求,致力于构建新经济下的“1+3+9+10+N”的数字资本生态,着力发挥数字产业和数字资本的产融协同作用,为优秀企业赋能,为国家大数据战略的实施添砖加瓦,为数字中国的建设贡献力量。
Guoxin Zhongshu, jointly established by the State Information Center and China Huarong, follows the concept of "Guoxin Zhongshu Empowers Enterprises and Strengthens the Nation." Leveraging the vast data resources of the National Information Center and the capital strength of its shareholders, the company is dedicated to building a digital capital ecosystem for the new economy, centered around the core requirements of China's digital transformation, represented as "1+3+9+10+N." They focus on harnessing the synergy between digital industries and digital capital, empowering outstanding enterprises, contributing to the implementation of the national big data strategy, and supporting the development of Digital China.
02
投资人说
联想集团高级副总裁、联想创投集团总裁贺志强表示:当前,以ChatGPT为代表的生成式大模型呈现出较强的通用智能能力,唯有将这些能力与产业深度融合,作用于千行百业,才能充分发挥其价值,最终推动生产力的“质变”。
He Zhiqiang, Senior Vice President of Lenovo Group and President of Lenovo Capital and Incubator Group, mentioned, "Currently, large generative models like ChatGPT exhibit strong general intelligence capabilities. Only by deeply integrating these capabilities into various industries can we fully unlock their value and ultimately drive a 'qualitative change' in productivity."
OpenCSG的创业方向是垂直大模型,专注于打造各行业专属大模型。通过使用通用基座的大模型,企业可以加入自己独有的场景数据,对大模型进行预训练及微调优化,并快速生成企业的“专属模型”,在解决大模型训练和部署使用成本高昂等问题的同时,有效提高企业的生产效率和服务质量。
OpenCSG's entrepreneurial focus is on vertical LLMs, with a specific emphasis on creating industry-specific LLMs. By utilizing the foundation of these large language models, businesses can incorporate their unique scenario data, conduct pre-training and fine-tuning optimizations, and swiftly generate their "custom models." This approach not only addresses the high costs associated with training and deploying large language models and multi-models but also effectively enhances a company's productivity and service quality.
联想创投长期关注AI领域的发展,未来,联想创投希望携手OpenCSG共同推动大模型在各行业领域的个性化深度应用,为千行百业的发展和社会整体效率的提升贡献科技之力。
Lenovo Capital and Incubator Group has maintained a long-term interest in the development of the AI field. In the future, Lenovo Capital and Incubator Group hopes to collaborate with OpenCSG to jointly promote the personalized and deep application of large language models across various industries, contributing technological advancements to the growth of diverse sectors and the overall efficiency of society.
北京国信中数投资管理有限公司总经理展钰堡表示:随着ChatGPT所掀起的人工智能浪潮,算力资源俨然成为AI时代最重要的基础设施之一。国家规划实施“东数西算”工程就是通过构建数据中心、云计算、大数据一体化的新型算力网络体系,将东部算力需求有序引导到西部,优化数据中心建设布局,促进东西部协同联动。2023年国产大模型的顷刻爆发,也导致算力的需求前所未有,增长曲线陡峭,大模型产业增长与算力供给之间的不平衡现状对国产大模型研发、迭代升级带来的困扰。
Zhan Yubao, General Manager of Digital China Investment Co.,Ltd., expressed: With the artificial intelligence wave triggered by ChatGPT, computing power resources have unquestionably become one of the most critical infrastructures in the AI era. The national plan to implement the "East Data, West Computes" project aims to create a new integrated computing power network system by constructing data centers, cloud computing, and big data integration. This system systematically directs computing power demands from the east to the west, optimizes the layout of data center construction, and promotes collaborative development between eastern and western regions.
The sudden surge in domestically produced LLMs in 2023 has also led to an unprecedented demand for computing power, resulting in a steep growth curve. The current imbalance between the growth of the large language model industry and the supply of computing power presents challenges for the research, development, and iterative upgrades of domestically produced large language models.
OpenCSG 创办公司的“让大模型赋能每一个人”理念和 国信中数 “为传统产业赋能,为优秀企业赋能,为国家大数据战略的实施添砖加瓦,为数字中国的建设贡献力量”的理念一致,OpenCSG 公司自主研发的星展(StarNet)平台也可以在算力枢纽节点上对于超算算力、智能算力和基础算力三种计算算力可以实现集群之间的算力调度,并且针对于大模型的训练、推理、部署有着完整的解决方案。期待 OpenCSG 能够继续打造让每个公司和个人都能够负担起的大模型应用,让原生大模型应用走进千家万户。
OpenCSG's founding principle of "empowering everyone with LLMs" aligns with Guoxin Zhongshu's vision of "empowering traditional industries, empowering outstanding enterprises, contributing to the implementation of the national big data strategy, and supporting the development of Digital China." OpenCSG's independently developed StarNet platform, which can manage and distribute computational resources between supercomputing power, intelligent computing power, and basic computing power at computing hub nodes, offers a comprehensive solution for the training, inference, and deployment of large language models.
We look forward to OpenCSG's continued efforts in creating LLMs applications that are accessible to both companies and individuals, making native LLMs applications accessible to households everywhere.
03
关于OpenCSG创始人
OpenCSG 创始人兼CEO陈冉表示:目前市面上正在掀起百模大战,每一个大模型聚焦的点都有所差异,大部分则以通用大模型为主,主要聚集自然语言处理、计算机视觉、语音识别技术等领域。众多国内外厂商开源了大语言模型,这大大降低了应用研发难度,让更多产品可以使用先进技术。
OpenCSG's Founder and CEO, Chen Ran(Sam) stated: "Currently, there is a fierce competition in the market with the proliferation of LLMs. Each LLM has its unique focus, but the majority are centered around general-purpose LLMs, primarily in natural language processing, computer vision, speech recognition, and related fields. Many domestic and international companies have open-sourced large language models, significantly reducing the complexity of application development and enabling a wider range of products to leverage advanced technologies."
中国的商业市场是足够大的,一旦开源文化达到成熟状态便会衍生出商业及变现。数据、算力、算法是支撑大模型产业发展三个要素,大模型的训练、推理不仅需要数据的支撑也需要算力资源的消耗。
The Chinese market is substantial, and once the open-source culture reaches maturity, it will give rise to commercialization opportunities. dataset, computing power, and algorithms are the three key elements that support the development of the large language model industry. Training and inference of large language models require not only data but also the consumption of computing resources.
OpenCSG 自主研发的星展(StarNet)平台就是一款基于云原生K8S技术的大语言模型训练和管理系统。它由星云平台(StarCloud)、星诞平台(StarStream)和星链平台(StarChain)等模块组成,可以实现云原生底座和算力资源管理,研究人员可以在平台上快速申请GPU资源, 搭建训练环境, 轻松实现模型训练与优化,大大减轻研究团队的资源压力。
OpenCSG's independently developed StarNet platform is a large language model &multi-models training and management system based on cloud-native technology. It consists of modules like StarCloud, StarStream, and StarChain, enabling cloud-native infrastructure and computing resource management. Researchers can swiftly request GPU resources on the platform, set up training environments, and easily carry out model training and optimization, significantly alleviating the resource constraints on research teams.
平台同时也提供LLM大模型优化和低代码链接融合框架,使得已有的大语言模型可以和应用更好地链接起来, 发挥更强大的大语言模型的生成能力。
The platform also provides an optimization and low-code integration framework for LLM (Large Language Models), enabling existing large language models to better integrate with applications and unleash their more potent generation capabilities.
结合国家的“东数西算”战略,OpenCSG平台逐渐具备实现算力统筹和智能调度的算力网络管理能力,随着本次 联想创投 和 国信中数 的投资完成和国家“东数西算”的政策支持,OpenCSG 目标打造能够在全国范围内,根据动态业务需求,在云、网、边之间实现按需分配和灵活调度计算、存储、网络等资源,在人工智能普及应用以及算法模型参数量和复杂度不断提升的背景下,提供具有数据吞吐大、计算速度快、算法工具多、软硬件协同、服务价格适中等特征的解决方案,来满足企业数字化转型和智能化升级需求。
In line with the national "East Data, West Computes" strategy, the OpenCSG platform is gradually gaining the capability for unified computing resource management and intelligent scheduling. With the completion of investments from Lenovo Capital and Guoxin Zhongshu, along with policy support for the national "East Data, West Computes" initiative, OpenCSG aims to create a solution that can be deployed nationwide. It will enable on-demand allocation and flexible scheduling of resources such as computing, storage, and networking, based on dynamic business needs. In the context of widespread AI adoption and the increasing complexity of algorithm model parameters, OpenCSG's solution will offer characteristics such as high data throughput, fast computing speeds, a wide array of algorithm tools, hardware-software synergy, and reasonable service pricing. This will meet the requirements of enterprise digital transformation and intelligent upgrading.
PS:
由AIGC掀起的技术革新,企业在复杂多变的环境中需要拥抱新技术,重塑核心竞争力,让我们通过大语言模型技术和开源生态社区共筑企业护城河。兹定于2023年10月20日,开放传神(OpenCSG)在上海市徐汇区云锦路701号5楼举办发布会。此次发布会我们邀请到的嘉宾包括上海市徐汇区科学技术委员会的领导、中国信息通信研究院的领导、投资界人士、人工智能行业的知名企业和媒体朋友们。
我们诚挚的邀请您莅临现场,您可通过下列二维码扫码报名,我们为您准备了神秘礼品,快来揭晓吧!
往期推荐
1. OpenCSG创始人陈冉:未来,开源大模型会主导整个大模型的发展方向