Baidu Releases iRAG Literacy Map Technology to "Remove the AI Taste" and Launches No-Code Development "Second Da" Tool

In today's 2024 BaiduWorld Congress, Baidu founderRobin LiTwo new AI technologies were released -- the retrieval-enhancedVincennes technology(iRAG) andNo Code Tools(coll.) second (of time).

Baidu Releases iRAG Literacy Map Technology to "Remove the AI Taste" and Launches No-Code Development "Second Da" Tool

Robin Li said.Wenxin Large ModelThe latest average daily call volume has reached 1.5 billion, up 7.5 times from six months ago (200 million calls).

The hallucination problem is one of the biggest limitations to the widespread use of big models, and the technology behind solving the text generation hallucination problem is RAG, or Retrieval Augmentation. Robin Li said that the biggest change for the big model industry in the past 24 months is that "big models have basically eliminated the illusion" and the accuracy of answering questions has greatly improved. "It's taken AI from serious nonsense to usable and trustworthy," he said.

but,There are still more serious illusions in the current Vincennes diagram based on the large language modelIn addition, especially for the Great Wall, the Pearl of the Orient, the pyramids, Einstein, Beethoven and other specific places, objects and people, often appear in the illusion of the problem, so that the generated picture "a glance of the false", which greatly affects the practicality of AI.

In response to the above.Baidu has developed iRAG (image based RAG), a retrieval-enhanced literate graph technology., the combination of Baidu search's billions of image resources with the basic modeling capabilities, "the overall effect is far more than the Vincennes native system, removing the machine flavor."

Robin Li showed a live picture of a large model generated by the WenxinPictures of Volkswagen Range Rovers flying over the Great WallThe iRAG technology was used to create the logo of this particular model of car. With Wenshin's iRAG technology, there were no errors or distortions in the illusions of either the model year badge of this particular car or the Great Wall in the background.

Robin Li also released at the conferenceNo-code tool "sec da". The tool supports three main features: code-free programming, multi-intelligence body collaboration and multi-tool invocation, and claims to be able to build a whole system in natural language.

At the speech site, Robin Li took the launch of the new technology of Radish Express as an example.Demonstrates the process of building an event registration systemThe system can be developed in Chinese by simply describing the requirements and adding a document with the theme of the time and place of the conference, and then commanding multiple intelligences to work together to complete the development of the registration system.

statement:The content is collected from various media platforms such as public websites. If the included content infringes on your rights, please contact us by email and we will deal with it as soon as possible.
Information

Microsoft Copilot to add new option for AI chat next month, with option to disable/enable web searches

2024-11-13 1:14:37

Information

Google Launches New AI Tool Learn About, Focused on Education and Learning

2024-11-13 3:20:21

Search