Robin Li: The threshold for Wen Xinyiyan's benchmark ChatGPT is very high, and Baidu is the first to do it among the world's major companies.

On the afternoon of March 16, Baidu held a press conference at its Beijing headquarters. The theme centered on the new generation of large language models and generative AI products.Wenxin​​One word. Robin Li, founder, chairman and CEO of Baidu, and Wang Haifeng, chief technology officer of Baidu, attended and demonstrated the five usage scenarios of Wen Xin Yi Yan in literary creation, business copywriting creation, mathematical calculation, Chinese understanding, and multi-modal generation. comprehensive ability.

Judging from the on-site demonstration, Wen Xinyiyan has the ability to understand human intentions to a certain extent, and the accuracy, logic, and fluency of his answers are gradually approaching human levels. However, Robin Li has also mentioned many times that this type of large language model is far from the stage of development and perfection, and there is a lot of room for improvement. In the future, it will definitely develop rapidly and change with each passing day.

Baidu also announced Wen Xinyiyan’s invitation test plan. Starting from March 16, the first batch of users can experience the product on Wenxinyiyan’s official website by inviting test codes, and it will be opened to more users in the future. In addition, Baidu Smart Cloud will soon open Wenxinyiyan API interface calling services to enterprise customers. Reservations will be officially opened on March 16. Search "Baidu Smart Cloud" to enter the official website, and you can apply to join the Wenxin Yiyan Cloud service test.

Currently, large language models and generative AI represent a new technological paradigm and are opportunities that every company in the world cannot miss. Baidu Wenxinyiyan is positioned as an artificial intelligence base-type empowerment platform that will assist the intelligent transformation of thousands of industries such as finance, energy, media, and government affairs. Robin Li said: "Baidu hopes to work with everyone to promote the advancement of artificial intelligence technology, so that everyone can use the most advanced productivity tools, so that everyone can benefit from it."

At the press conference, Robin Li showed The performance of Wenxinyiyan in five usage scenarios, including literary creation, business copywriting creation, mathematical calculation, Chinese understanding and multi-modal generation.

In the literary creation scene, Wen Xinyiyan summarized the core content of the well-known science fiction novel "The Three-Body Problem" based on dialogue questions, and put forward five suggested angles for continuing the "Three-Body Problem", embodying Develop comprehensive abilities in dialogue Q&A, summary analysis, and content creation.

In addition, Wen Xinyiyan accurately answered factual questions about the author of "The Three-Body Problem" and the role player in the TV series. Generative AI often "makes things up" when answering factual questions, and Wen Xinyiyan continues Baidu's knowledge-enhanced large model concept and greatly improves the accuracy of factual questions.

In the business copywriting creation scenario, Wen Xinyiyan successfully completed the creative tasks of naming the company, writing a slogan, and writing a press release.

In three consecutive content creations, Wen Xinyiyan was able to accurately understand human intentions and express them clearly. This is the "intelligence emergence" that occurs based on the huge scale of data. The training data of the Wenxin Yiyan large model includes trillions of web page data, billions of search data and image data, tens of billions of daily voice call data, and a knowledge graph of 550 billion facts.

Wen Xinyiyan also has a certain degree of thinking ability and can learn relatively complex tasks such as mathematical deductions and logical reasoning. Faced with classic questions such as "Chicken and rabbit in the same cage" that train human logical thinking, Wen Xinyiyan can understand the meaning of the question and have the correct ideas for solving the problem, and then follow the correct steps to calculate the problem step by step like a student. correct answer.

Literary creation, business copywriting creation, and mathematical calculation are common advantages and abilities of large language models. On this basis, Wenxinyiyan also shows better Chinese understanding and multi-modal generation capabilities.

During the on-site demonstration, Wen Xinyiyan correctly explained the meaning of the idiom "Luoyang paper is expensive" and the corresponding economic theory of "Luoyang paper is expensive", and also created an acrostic poem using the four words "Luoyang paper is expensive".

In terms of multi-modal generation, Robin Li demonstrated Wen Xin Yi Yan’s ability to generate text, pictures, audio and video. Interestingly, Wenxinyiyan can even generate speech in dialects such as Sichuan dialect; Wenxinyiyan’s video generation capability is not currently open to all users due to its high cost, and will be gradually accessed in the future.

“Multimodality is a clear development trend of generative AI.” Robin Li said, “In the future, as Baidu’s ability to unify large multimodal models increases, Wen Xinyiyan’s multimodal generation capabilities will It will also continue to improve."

Judging from Wen Xinyiyan's performance, to a certain extent, it has the ability to understand human intentions, and the accuracy, logic, and fluency of its answers are gradually approaching human levels. . But overall, this type of large language model is far from being fully developed and relies on gradual iteration through real user feedback.

Wang Haifeng said that Wenxinyiyan is a new generation of knowledge-enhanced large language model, which is developed on the basis of the ERNIE and PLATO series models. Its key technologies include supervised fine-tuning, reinforcement learning with human feedback, prompts, knowledge enhancement, retrieval enhancement and dialogue enhancement. The first three are technologies used by such large language models, and have been applied and accumulated in ERNIE and PLATO, and have been further strengthened and polished in Wen Xinyiyan; the last three are technologies that Baidu already has technical advantages. Re-innovation is also the foundation for Wen Xinyiyan to become stronger and stronger in the future.

Li Yanhong emphasized: "Wen Xinyiyan will establish a flywheel between real user feedback, developer calls and model iterations, and the effect will improve rapidly, giving you After three days of separation, it’s a surprise to see each other with admiration.” Robin Li said that Baidu is currently the first company among the world’s major companies to make a benchmark ChatGPT product. Robin Li pointed out: "No matter which company it is, it is impossible to build such a large language model in a few months. Deep learning and natural language processing require years of persistence and accumulation, and cannot be achieved quickly."

It can be said that Wen Xinyiyan is the continuation of Baidu’s efforts over the past many years. As humans enter the era of artificial intelligence, the technology stack of IT technology has undergone fundamental changes, from the past three layers to the four layers of "chip-framework-model-application". Today, Baidu is one of the few artificial intelligence companies in the world that has a full-stack layout in these four layers, from high-end chip Kunlun core, to Feipiao deep learning framework, to Wenxin pre-trained large models, to search, intelligent cloud, Applications such as autonomous driving and Xiaodu have industry-leading self-developed technologies at all levels.

Robin Li believes that the advantage of Baidu AI’s full-stack layout is that it can achieve end-to-end optimization in the four-layer architecture of the technology stack, greatly improving efficiency. In particular, there is a strong synergy between the framework layer and the model layer, which can help build more efficient models and significantly reduce costs. In fact, the training and inference of very large-scale models pose a great challenge to the deep learning framework. For example, in order to support efficient distributed training of hundreds of billions of parameter models, Baidu Flying Paddle has specially developed 4D hybrid parallel technology.

Since Baidu officially announced “Wen Xin Yi Yan” in February, more than 650 companies have announced their access to the Wen Xin Yi Yan ecosystem.

Robin Li predicts that large language models will bring three major industry opportunities.

The first category is a new type of cloud computing company, whose mainstream business model has changed from IaaS to MaaS. Wen Xin's words will fundamentally change the rules of the game in the cloud computing industry. In the past, enterprises chose cloud vendors based more on basic cloud services such as computing power and storage. In the future, more will depend on whether the framework is good, whether the model is good, and the collaboration between the four layers of model, framework, chip, and application.

Wen Xinyiyan will provide external services through Baidu Intelligent Cloud to help enterprises build their own models and applications. Key areas such as agriculture, industry, finance, education, medical care, transportation, and energy will greatly improve efficiency as a result. , and quickly form new industrial spaces in every industry to help realize Digital China. Robin Li predicted that Baidu Smart Cloud will hold a press conference in the near future, with the theme centered on Wen Xinyiyan’s cloud services and application products, which include both public cloud services and privatized deployment.

The second category is companies that fine-tune industry models. This is the middle layer between the general large model and enterprises. Based on their insights into the industry, they can use the general large model capabilities to provide solutions to industry customers. plan. In this regard, Baidu Wenxin Model has released more than 10 industry models in electric power, finance, media and other fields.

The third category is companies that develop applications based on large model bases, that is, application service providers. Robin Li asserted that for most entrepreneurs and companies, the real opportunity is not to build basic large-scale models like ChatGPT and Wenxinyiyan from scratch. This is very unrealistic and uneconomical. This may be the real opportunity to preemptively develop important application services based on a general large language model. At present, based on text generation, image generation, audio generation, video generation, digital people, 3D and other scenarios, many entrepreneurial star companies have emerged, which may be new giants in the future.

"We believe that artificial intelligence will completely change every industry we have today. The long-term value of AI and the disruptive changes to all walks of life have just begun. In the future, there will be more killers With the emergence of applications and phenomenal products, more milestone events will occur." Robin Li said. (one orange)

The above is the detailed content of Robin Li: The threshold for Wen Xinyiyan's benchmark ChatGPT is very high, and Baidu is the first to do it among the world's major companies.. For more information, please follow other related articles on the PHP Chinese website!