HawkInsight

  • Contact Us
  • App
  • English

Baidu's large model has been iterated to 3.Version 5: Training speed increased by 2 times, reasoning speed increased by 17 times

On June 26, Baidu founder, chairman and CEO Robin Li attended the "World Internet Conference Digital Civilization Nishan Dialogue" and delivered a speech entitled "Big Model Reshapes the Digital World."。

On June 26, Baidu founder, chairman and CEO Robin Li attended the "World Internet Conference Digital Civilization Nishan Dialogue" and delivered a speech entitled "Big Model Reshapes the Digital World."。

Li revealed in a public speech that the Baidu Wenxin model has been iterated to 3.Version 5, with 3.Compared to version 0, the training speed has increased by 2 times, the reasoning speed has increased by 17 times, and the cumulative model effect has increased by more than 50%.。

In addition, Robin Li also stressed that Wen Xin Da Model 3.Version 5 is not only a technical upgrade, but also a security upgrade。Significant improvements in data quality, generation effectiveness, and content security。

Regarding the recent global impact on AI regulation, Robin Li said, "Only by adhering to technological development and a safe and controllable two-wheel drive can we be stable and far-reaching.。If we safely and responsibly navigate the path of AI development, big models will reshape the digital world, and AI can create unparalleled prosperity for the Chinese economy and the global economy, and improve the well-being of all mankind.。"

   

The following is a transcript of the speech:

   

Dear leaders, distinguished guests, good morning!

It is a great pleasure to participate in the Nishan Dialogue on Digital Civilization at the World Internet Conference, and the topic of my speech is "Big Models Reshape the Digital World."。

In the past year, artificial intelligence in technology, products, applications and other levels, are "week" for the iterative speed forward。The big model has successfully compressed human perception of the world, allowing us to see the path to general artificial intelligence.。The next frontier in the development of large models is not only to imitate humans and complete their "prescribed actions," but also to help humans study and discover unknown areas, breaking through the limits that humans have not broken through in the past.。If you can take this step, it will be even more significant。

How Big Models Are Reshaping the Digital World?I would like to talk about both technical and application levels:

At the technical level, in the era of artificial intelligence, the IT technology stack has undergone fundamental changes, from the original three-tier architecture of chips, operating systems and applications to a four-tier architecture of chips, frameworks, models and applications.

The bottom layer is the chip layer, the mainstream chip from the CPU to the GPU。On top of the chip is the framework layer, the mainstream framework includes Baidu Flying Paddle, Meta's PyTorch, Google's TensorFlow。Above the framework is the model layer, and the ChatGPT and Wenxin big models are at the model layer.。The big model has become the operating system of the era of artificial intelligence, and all applications will be developed based on the big model.。On top of the model is the application layer, including a wide variety of AI-native applications。

Structural changes in the IT technology stack mean that artificial intelligence, especially large model technologies, will restructure the global digital industry。The key point of the new international competition strategy is not how many big models there are in a country, but how many native AI applications there are on your big models, and to what extent these applications increase productivity.。If we can get on the table and get tickets to the competition, China will have a bigger digital industry and a huge growth in the scale of the digital economy.。

Baidu has invested in artificial intelligence for more than 10 years, in the chip, framework, model, application four layers have a full stack layout, in the key core technology, Baidu in the four-layer architecture has independent research and development of leading products and technologies, so you can carry out end-to-end optimization, quickly improve the efficiency of large model training and reasoning.。The Wenxin model is completely autonomous and controllable, and we have achieved data control, framework control, and model control.。

Of course, the governance challenges posed by the AI grand model cannot be ignored。The application of new technologies often precedes norms, and the establishment of sound laws and regulations, institutional systems, ethics and ethics to ensure the healthy development of artificial intelligence can create a good innovation ecology.。With a view to the future, while paying attention to risk prevention, we should also establish fault-tolerant and error-correcting mechanisms at the same time, and strive to achieve a dynamic balance between regulation and development.。

Right now, big models are getting hot。But four years ago, before the big model was widely watched, Baidu launched the Wenxin Big Model 1.0。Then continue to evolve to 2.0, 3.Version 0。

Today, the Wen Xin Da model has been iterated to 3.5 version, with 3 of March.Compared to version 0, the training speed has increased by 2 times, the reasoning speed has increased by 17 times, and the cumulative model effect has increased by more than 50%.。

Wen Xin Da Model 3.Version 5 is not only a technical upgrade, but also a security upgrade。We conducted the assessment using the industry's mainstream large model-based capability assessment methodology, and the results showed that the Wenxin Large Model 3.Version 5, in terms of data quality, generation effect and content security, has been significantly improved。

China's artificial intelligence model has a certain foundation, we need to catch up。At the same time, we should give full play to the advantages of application scenarios, further deepen the vertical field, create professional large models in the fields of finance, medical care, power, etc., achieve technical optimization with high-quality applications and data feedback, help large models to iteratively upgrade, and build a good AI ecology.。

It can be predicted that the big model will penetrate into more and more fields, and the digital economy driven by the big model will be deeply integrated with the real economy, which will strengthen and optimize the real economy, create considerable incremental value, and bring about profound changes in economic and social development and industry.。

For example, in the automotive manufacturing industry, the most complex design process requires experienced engineers to find various combinations that meet the needs of more than 20,000 parts and hundreds of thousands of parameters, and then write documents and drawings.。In Changan Automobile, large models can efficiently find combination information, automatically generate design documents, and significantly reduce R & D cycles and costs。In Sinopec and China Southern Power Grid, large models can go deep into core business scenarios, innovate in intelligent customer service, supply chain, system scheduling and other sections, and promote the digital transformation and intelligent improvement of the industry.。

In the field of transportation, intelligent transportation solutions supported by large model technology can improve traffic operation efficiency。

For example, on the last working day before the May Day holiday this year, Beijing's urban congestion index skyrocketed by 2.5 times。From the second ring to the sixth ring, a red, the only green is Yizhuang。Yizhuang traffic also increased significantly, but because of the deployment of AI global information control scheme, more than 300 intelligent intersections, can automatically adjust the traffic lights according to the traffic flow, Yizhuang has become an "oasis" without traffic jams.。On the day before Dragon Boat Festival, the traffic in downtown Beijing and Yizhuang was strikingly similar: the city was congested, but Yizhuang was unblocked.。

In the Taishan Scenic Area of Tai'an, Shandong Province, in order to serve the development of tourism economy, promote urban congestion and smooth, and solve the pain point of "difficult parking" for foreign tourists, Baidu uses intelligent control methods such as traffic guidance screens and green wave belts to effectively ensure the safety of citizens and tourists.。

Baidu intelligent transportation solution has been adopted by 69 cities。By intelligently adjusting the timing of traffic lights, you can increase traffic efficiency by 15% -30%, which will drive GDP2.4% -4.8% growth。

In Jinan, Shandong Province, we have also landed Baidu Intelligent Cloud (Shandong) Artificial Intelligence Basic Data Industry Base, which not only cultivates new professions such as AI trainers, but also incubates 22 data labeling technology companies, driving regional employment while boosting economic growth.。

No matter from the perspective of technological trends or industrial applications, the big model is by no means a flash in the pan, but a major technological change affecting human development, an engine of global economic growth, and a major strategic opportunity that must not be missed.。

Adhere to technological development and safe and controllable two-wheel drive, in order to move forward steadily.。If we safely and responsibly navigate the path of AI development, big models will reshape the digital world, and AI can create unparalleled prosperity for the Chinese economy and the global economy, and improve the well-being of all mankind.。

·Original

Disclaimer: The views in this article are from the original author and do not represent the views or position of Hawk Insight. The content of the article is for reference, communication and learning only, and does not constitute investment advice. If it involves copyright issues, please contact us for deletion.