The 2023 China Intelligent Application Development Forum opened in Beijing, Zhilingyun was invited to participate and achieved complete success

From April 18th to April 19th, the 2023 Second China International Software Development Conference hosted by the China Software Industry Association was grandly held at the Beijing National Convention Center. As one of the most influential software industry conferences in China, the conference attracted thousands of spectators.

2023 China International Software Development Conference CCTV News Report

On April 19th, as one of the sub-forums of the conference, the first China Intelligent Application Development Forum hosted by China Software Industry Association and organized by China Software Industry Association Intelligent Application Service Branch was in full swing. With the theme of "Intelligent Application, Innovation Driven", this forum will jointly explore how to accelerate the intelligent transformation of various industries and achieve continuous innovation and development in the era of intelligence through intelligent technology and application. More than 200 guests including well-known experts, scholars, entrepreneurs, industry leaders and media representatives from all over the country attended the forum. Dr. Peng Feng, co-founder and CEO of Zhilingyun Technology, attended the forum as the deputy secretary-general of the Intelligent Application Service Branch of China Software Industry Association. 

00b122d97599b3c6bd6f70996b8ede6c.jpeg

Forum site

Driven by the explosion of large-scale model technology represented by ChatGPT, the awakening and challenges brought by intelligent technology and applications

In the ascendant artificial intelligence track, ChatGPT rushed out as the biggest AI dark horse at the beginning of the year, and quickly attracted the attention of all parties on the track. Driven by the explosion of large-scale model technology represented by ChatGPT, the application of intelligent technology is profoundly affecting our life and work, and constantly bringing new changes and opportunities to society.

In this regard, all walks of life are speeding up the transformation of intelligence to achieve continuous innovation and development. However, if they want to truly embark on the road of intelligence and promote the innovation and application of intelligent technology, there are still many challenges to overcome, such as natural language Technical capabilities such as processing, context understanding, logical reasoning, and emotional expression all need further iterations.

In order to help enterprises and universities cope with the challenges of intelligent transformation, promote the innovation and application of intelligent technology in various industries, strengthen exchanges and cooperation between enterprises and related institutions in the field of intelligent application, and showcase China's innovative achievements in the field of intelligent technology, this conference has arranged dry goods Full technology sharing.

Industry leaders discuss the new era of intelligence

The forum was chaired by Li Xuexiang, deputy director of the Intelligent Application Service Branch of China Software Industry Association and deputy dean of the School of Cyberspace Security of Zhengzhou University. The forum was carried out in the form of theme sharing combined with interactive communication. While listening to the sharing of technical dry goods and related cooperation strategies, the guests also actively responded to the interactive communication of experts and scholars. In the warm on-site atmosphere, experts and scholars from various industries, as well as guests from enterprises and institutions conducted in-depth discussions and talked about the new smart ecology.

04a0ed6c631683ff733b84e6abe2e21f.jpeg

Li Xuexiang, deputy dean of the School of Cyberspace Security, Zhengzhou University

The conference specially invited Chen Baoguo, executive deputy secretary-general of China Software Industry Association, to deliver a speech on stage. Chen Baoguo said that when it comes to intelligent applications, education, training, understanding, facing, and tolerance will become issues of common concern to human society in the future.

449df271902d97658d705de14201c312.jpegChen Baoguo, Executive Deputy Secretary-General of China Software Industry Association

Jiang Tao, chairman of the Intelligent Application Service Branch of China Software Industry Association and founder of CSDN, said that we believe that the large model will become the base of the next generation of cloud computing and intelligent applications. The large model will change our current application ecology. Every software, every Tools and every industry will be restructured and changed because of the large model of artificial intelligence.

01249cfbf85e4b3b21c13519dd3d40b8.jpegCSDN founder Jiang Tao

The Status Quo and Trend of Smart Application Development in China

At the event site, Chen Zhong, a professor of Peking University, director of Metaverse Institute of Technology, and vice chairman of China Software Industry Association, gave a heavy opening speech on the theme of "The Status and Trends of Smart Application Development in China". He showed the present situation and trend of China's intelligent application development to the guests with his clear views and authoritative reports. Professor Chen Zhong started from the three waves of artificial intelligence, introduced the overall situation of my country in the three levels of basic layer, technical layer and application layer, and made a related description of the data, computing power and algorithm required by artificial intelligence . Secondly, the report gives an overall introduction to the current research and application of large-scale model technology, and analyzes the application scenarios and development trends of China's smart technology. Finally, suggestions are put forward on how to realize the integration of technologies such as blockchain, Yuanyushen and AIGC in the A1 2.0 era.

2f5c73a2efbc3857ece7cc67a10d314b.jpeg

Chen Zhong, Professor of Peking University and Vice Chairman of China Software Industry Association

Looking at the challenges and opportunities of the next generation of artificial intelligence from ChatGPT to GPT-4 

Jiao Licheng, foreign academician of the European Academy of Sciences, foreign academician of the Russian Academy of Natural Sciences, IEEE Fellow, Huashan Scholar Distinguished Professor of Xidian University, then delivered a keynote speech on "Challenges and Opportunities of Next-Generation Artificial Intelligence from ChatGPT to GPT-4". In this sharing, Professor Jiao Licheng first reviewed the development of ChatGPT based on natural language processing and large models, as well as the latest research progress. He emphasized that artificial intelligence technology, as the main productive force in today's society, continuously empowers the society and has achieved very remarkable results in many industries and fields. The core of ChatGPT, which has attracted much attention recently, is a major achievement made by the third wave of artificial intelligence technology. progress. Our country attaches great importance to the development of artificial intelligence, has invested a lot of talents and financial resources, and formulated many strategic guidelines to develop artificial intelligence and lead artificial intelligence.

c25f53e8077327b5b8749e361870c106.jpeg

New technology, new opportunity, new era--A12.0 redefines talent 

The application of large model technologies such as ChatGPT is a major advance in the field of artificial intelligence and will bring new opportunities for the new era of AI2.0. In this regard, Qiu Qinlun, secretary-general of the Intelligent Application Service Branch of China Software Industry Association and executive director of China Soft Education Research Institute, shared the theme of "New Technologies, New Opportunities, New Era--A12.0 Redefining Talents", based on the association's ongoing The promotion of post skill model specification and five new intelligent application posts proposes to integrate the large model with the main teaching and training methods to form a student-centered and practice-oriented education method, providing well-documented and quantitative training methods for talent training. dependable support.

4da39dc86a92d00639e3ba097b0a70d7.jpeg

Qiu Qinlun, Executive Director of China Soft Education Research Institute

Application Prospect and Trend of Large Language Model (LLM) Technology in China

Dr. Peng Feng, Deputy Secretary-General of Intelligent Application Service Branch of China Software Industry Association, Co-Founder and CEO of Zhilingyun Technology, delivered a keynote speech on "Large Language Model (LLM) Technology Application Prospects and Trends in China". He has successively served as Twitter big data architect and technical leader, Ask.com engineering director, has more than 20 years of software development, and has rich practical experience in big data.

f584bb2bf0a80bda493b1c08a190d852.jpegPeng Feng, co-founder and CEO of Zhilingyun Technology

Dr. Peng Feng emphasized that the key to the success of ChatGPT lies in engineering, including data cleaning, model fine-tuning and reinforcement learning based on human feedback (RLHF). Although the training of large models is beyond the reach of most enterprises, domain-based fine-tuning and RLHF is the key to solving enterprise usage.

381c1921cd129bba98ccc77936ab6df6.jpeg

Dr. Peng Feng's speech

LLM fine-tuning refers to the further training of the model using task-specific datasets on the basis of pre-training. Doing so allows the model to better adapt to a particular task, thereby improving its performance on that task. 

1. Pre-training model: pre-trained large-scale language model, such as GPT-2 or GPT-3. These models have been trained on a large amount of text data and have a certain language understanding ability. 

2. Task-related datasets: datasets related to a specific task. This dataset should contain examples of inputs and outputs so that the model learns how to deal with similar problems. 

3. Fine-tuning training: Fine-tuning the pre-trained model using task-related datasets. During training, the model is tuned to the task dataset, learning how to solve a specific task. 

4. Validation and evaluation: In the fine-tuning process, in order to monitor the performance of the model, the data set is usually divided into a training set and a validation set. Adjust the training parameters according to the performance of the model on the validation set to achieve better performance. 

5. Application and deployment: After the fine-tuning is completed, apply the model to the actual task. For example, you could deploy a fine-tuned model into a chatbot that can answer user questions about programming.

6912b97bf543b815730bf2116f34b6ba.jpeg

Based on this, Zhiling Cloud launched the "Paper Brain" tool, which can provide an intelligent paper analysis and output platform, so that university teachers and students can manage and create papers more efficiently, so as to achieve better results in the research process. The tool is designed to be the most efficient tool for thesis productivity of university teachers and students. Through intelligent technology and personalized service, it can provide them with in-depth analysis and insights in one stop. The specific value is manifested in the following three aspects:

1. Visual reports: Users can generate various charts and reports based on dimensions such as time, author, and keywords, to intuitively show the trends and key points of the paper research.

2. Automatic abstraction and keyword extraction: Through advanced natural language processing technology, Paper Brain can automatically extract key information and topics in papers, and provide users with fast and accurate abstracts and keywords.

3. PaperGPT auxiliary function: training and integrating large models in the private domain, enabling users to interact more meaningfully with the application, focus more on answers and produce content more intelligently.

High-end Dialogue: Opportunities and Challenges in the Era of AI Large-scale Models

In 2023, driven by ChatGPT and GPT-4, various AI products will be launched one after another. At the same time, major manufacturers have already or are gradually integrating large-scale model capabilities into their own businesses. In the future, Internet applications will either be "native to large models" or be "reconstructed by large models". People's lives and production will be redefined by "intelligence". The era of intelligent Internet based on large models has already begun. In the new era, opportunities and challenges coexist. In this roundtable forum, the middle-class, learning, research, application, and venture capital guests will discuss the development direction and trend of intelligent technological innovation around topics such as basic research, technological innovation, application implementation, and investment and financing opportunities. , and conduct in-depth exchanges on how to adapt to the current situation and promote the development of intelligent technology applications.

e66b39f2c092ba99d0552718c0fcbb54.jpegRound table exchange scene

AI Empowerment Program Certification Ceremony

The forum also held the AI ​​Empowerment Program Certification Ceremony, and Zhilingyun was awarded the 2023-2025 AI Empowerment University Program Cooperative Unit. The AI ​​Empowerment Plan was proposed by the Intelligent Application Service Branch of the China Software Industry Association. It includes two parts: industry empowerment and university empowerment. The purpose is to focus on the application and practice of AI technology, cultivate millions of AI combat talents for China, and help 100,000 enterprises. Intelligent. Tian Xinan, Executive Secretary-General of the Intelligent Application Service Branch of China Software Industry Association, emphasized that the deep integration of university talent training and industrial employment needs is an important measure to promote the integration of production and education and deepen school-enterprise cooperation. The plan aims to promote artificial intelligence education and research, and is committed to providing more college teachers and students with AI technology and resources from domestic and foreign leading companies to improve teaching quality, scientific research level and innovation ability, thereby promoting the process of talent training.

ff172116aa7eb5d6a3a649199cf9a2ec.jpegCertificate awarding ceremony 

The development of the empowerment era drives the new value of intelligence

Accelerate the development of intelligent technology and applications together, empower enterprises and universities to innovate efficiently with intelligence, stimulate the potential of intelligent ecology, call for more industry partners to act together, and provide new kinetic energy for users who are moving towards intelligence in a win-win manner , Empowering the development of the new era and driving the new value of intelligence is the main purpose of this forum. In the next step, Zhiling Cloud will continue to play its role in the "intelligent technology and application innovation development platform", focus on strengthening research and innovation, gather the wisdom of the intelligent industry, and conduct in-depth research and scientific research in smart industries such as big data, artificial intelligence, and cloud computing. The layout provides all-round services and strong support for the innovation-driven development of the smart industry.

41ec9aad7dd0bafe0740b2f32c46c9f0.jpegConference site

2f44fd8ccf1d7e0d959a4d4e545a96d2.gifClick "Read the original text" to learn more

Guess you like

Origin blog.csdn.net/LinkTime_Cloud/article/details/130355698