Get the App
SLTechnology News&Howtos  ›  IT Information  › 

Baidu Li Yanhong: the gap between Wen Xin Yiyan and ChatGPT is not very big, only for a month or two.

Shulou Source: shulou.com Published: 2023-11-24 15:46:33 10月02日 Update

Thank CTOnews.com netizens S6 for the clues of winning the championship! CTOnews.com March 26 news, recently, Robin Li said in a live broadcast in Geek Park that Wen Xin's word is not as good as the best ChatGPT version, but the gap is not very big, it may be a month or two.

Li Yanhong said, "I also keep saying that a word of Wen Xin is not perfect. In fact, if we evaluate it comprehensively, it is not as good as the best ChatGPT version now, but the gap is not very big. The so-called is not very big, it may be the difference of one or two months."

Li Yanhong also said, "according to the team's current analysis, our level is about the level of ChatGPT in January this year. But people have long forgotten what it looks like in January. Today, we are used to GPT-4,GPT-4 technology that is only one day behind us, and it is difficult for other big companies to come up with a comparison."

Robin Li said that although there are still many shortcomings in Wen Xin's words, he is proud that the Baidu team was able to take the lead in launching such a product among the big companies around the world, because the market demand is so strong that countless customers want to try and cooperate as soon as possible.

Li Yanhong revealed that Baidu has been doing a big language model since 2019, and has gone through the iterations of Wenxin 1.0, 2.0 and 3.0. After OpenAI released ChatGPT on November 30 last year, the Baidu team felt great pressure and a sense of crisis, and felt that the gap with the international level had been further widened. As a result, they worked day and night to speed up the research and development of Wen Xin, and invited internal testing on March 16 this year.

Robin Li admitted that he made a decision two months before the release that it must be issued in March, and publicly announced this time node. He did this on purpose to put some pressure on the team so that they could move faster and improve faster. "at that time, I took a shot, saying that it must be issued in March," he said. "the team was not so confident and worried that it would not be done in March. I said that even if it could not be done at that time, let's go and explain to the public that we delay a little bit, and I think we are also OK. But we have to talk about it now, and we will send it in March."

Robin Li also admitted that he had encountered some criticism after the release of Wen Xin's words, but this was what he expected. He said at least three times at the press conference that "a word of literary heart is not perfect." He chose this time to release, the main reason is the strong market demand. "countless people-people who didn't have much contact or the industry is far away from me in the past are now asking how we can work with Baidu and how to try it out as soon as possible," he said. after the customer puts forward the demand to us, we carry on the targeted optimization and iteration, which can soon become very useful. "

Li Yanhong also analyzed why other big companies, including Google, Facebook and Amazon in the United States, did not launch products similar to Wen Xin's words. He thinks there are two reasons: one is that they didn't pay so much attention to it before, and the other is that there is a lot of pressure if it is done as the second of a big company and is not as good as a start-up.

CTOnews.com noted that at the press conference of Wen Xin, Li Yanhong not only emphasized the huge demand of Baidu's customers and Baidu itself for Wen Xin, but also kept lowering his figure, admitting that Baidu did not have a "complete Ready" and that he wanted to mark ChatGPT or even GPT-4. The threshold is very high.

Tags: Wen Xin Li Yan month team big demand big factory two gap level pressure customer exuberant continuous product reason press conference that is difference market Apple Docker Huawei Linux macOS MariaDB Microsoft MySQL NVidia OPPO Reno Microsoft macOS Shulou Information Docker vpn