Hundsun Electronics presented its financial large model practical results at the 2023 Artificial Intelligence Computing Conference to create a financial large model product ecosystem

The 2023 Artificial Intelligence Computing Conference (AICC 2023), with the theme of "Intelligence and Computing Power is Innovation", was held in Beijing on November 29. This conference focuses on hot topics such as intelligent computing infrastructure, generative AI and large model technology, and industrial digital intelligence transformation results. It brings together government leaders, experts and scholars, industry leaders and technological innovation pioneers to discuss cutting-edge trends and innovations in the intelligent computing industry. Application results

As one of the representative manufacturers in the field of domestic financial large models, Hundsun Electronics was invited to participate in this conference. Bai Shuo, chief scientist of Hundsun Electronics and director of Hundsun Research Institute, delivered a keynote speech at the main forum of the conference, discussing hot topics in the intelligent computing industry with leading institutions such as Inspur Information, Baidu, Tencent Cloud, Geely, etc. At the same time, Hundsun Electronics was in the field of financial large models. The practical results were also unveiled at the conference and exhibition, demonstrating the innovative empowerment of large model technology in the financial industry.

Create a large model ecosystem in the financial vertical field

The financial industry has become one of the core industries for large-scale model applications because of its rich data resources, good digital foundation and diversified business scenarios. At the main forum, Bai Shuo shared with the guests the thoughts and practices related to Hundsun Electronics on the topic of "Application Practice of Large-Scale Models in the Financial Industry"

Bai Shuo said that the informatization process of the financial industry is already in the transition from Network 2.0 to Digital Intelligence 3.0. The goal of digital intelligence is to replace mental labor with technology, and AI large models are one of the iconic technologies of this era. As an emerging AI model, large language models have made new improvements in various aspects such as data magnitude, model development efficiency and methods, and scene versatility compared to the traditional chimney-style AI models in the past. They can achieve more accurate identification, Better and cheaper mission results.

In the financial field, we can find that large-scale models can not only provide support for business scenarios such as customer service, investment consulting, risk control, investment research, etc., but can also support the internal digitization of financial institutions by reconstructing software development and data production links. Provide support for transformation. However, for financial institutions, due to its strong business characteristics and compliance, the "plug-in alliance" model with a general large-scale model as the core is difficult to achieve in-depth application. The industry needs to establish "vertical field large-scale models" and competency centers that can fully utilize the capabilities of large-scale models and connect proprietary real-time data, proprietary resources and application systems in the financial field to achieve compliance application of large-scale models

Hundsun Electronics’ large model ecology consists of three parts. The basic layer is the computing power layer, and the large model layer is LightGPT, a large model exclusively developed for the financial industry independently developed by Hundsun Electronics. This model is trained based on a large amount of financial data. The application layer includes Photon, a "central control" platform that connects large language models and industry applications. It is mainly used to solve problems such as technical connection, application connection, data security and compliance encountered in the actual application of large models, and is based on Photon Created a series of large model application products

In addition, Bai Shuo said that Hundsun Electronics has launched a reconstruction model product for large model application models based on photons, and has embedded new interactive capabilities supported by large models into original applications. These products realize the embedded mode of voice control of thousands of things and voice control of everything. Hang Seng will also continue to explore native mode products, which can disassemble complex instructions through natural language expression, and organize task pipelines involving multiple resources and execute them sequentially

Large-scale application results are unveiled

In addition to sharing in the main forum, Hundsun Electronics also appeared in the "Intelligent Travel" area of ​​this conference, focusing on innovative results in the field of financial large model applications, including financial large model LightGPT, and multiple photon series large model applications. Products (Photon·Shance, Photon·Wenqu, Photon·Huiying), and the intelligent investment research platform WarrenQ.

Among them, Photon·Shance focuses on investment consulting scenarios, helping investment consultants analyze customer intentions in real time based on conversation records, accurately locate customer tags, automatically retrieve topic-related financial products and information, and generate professional opinions and suggestions.

Guangzi Wenqu is a one-stop content creation platform for investment advisors. Through rich creative templates, intelligent abstract generation, AI Q&A assistant and other functions, it helps investment advisors quickly complete the collection of report materials, content generation, and review and release.

Guangzi·Huiying Positioning Asset Management and Custody Product Operation Assistant is a tool that provides system intelligent auxiliary solutions for operators. It can be applied to scenarios such as intelligent parameter extraction, intelligent analysis of investment supervision rules, and intelligent analysis of audit reports, thereby improving operational efficiency

Hang Seng Juyuan is a subsidiary of Hang Seng, focusing on building an integrated professional tool platform for the investment research field. Based on large model technology, Hang Seng Juyuan developed the WarrenQ platform, which connects investment research work scenarios through natural language interaction and realizes the upgrade of "Chat reading, calculation and writing". The platform integrates large models, data and tools and is a representative of a new generation of intelligent investment research products

Based on high-quality data and scenarios, large-scale models have broad application prospects in the financial field. In addition, the financial industry has specific business scenario needs compared to other industries, and has higher requirements for "accuracy, real-timeness, and compliance"

In the future, Hundsun Electronics will use its accumulated domain knowledge and service experience in the financial industry to further improve the financial large model service ecosystem, continue to establish a large model innovation application matrix, and fully explore the role of emerging artificial intelligence technology in business. Huge potential to provide support for the digital and intelligent transformation and upgrading of the industry

The above is the detailed content of Hundsun Electronics presented its financial large model practical results at the 2023 Artificial Intelligence Computing Conference to create a financial large model product ecosystem. For more information, please follow other related articles on the PHP Chinese website!