At Baidu World 2023 on October 17, version 4.0 of Wenxin Large Model was officially released! Baidu directly said: Wenxin Big Model 4.0 is currently the most powerful Wenxin Big Model. At the meeting, Baidu Chairman Robin Li showed us the power of Wenxin Big Model 4.0 in search, maps, business intelligence, smart meetings, smart videos, and multi-round conversations. So is Wenxin Big Model 4.0 really that good? We obtained the test qualification as soon as possible and conducted actual testing.
Compared with Wenxin Big Model 3.5 when it first came out, Wenxin Big Model 4.0 has now evolved to have more functions. There are 8 plug-ins alone, including Yijing Liuying (text to video), illustration and painting (see Pictures speak), E-Words (visualized data analysis), etc.
These plug-ins can also be freely combined to complete more complex tasks.
Today we will start from a more basic point of view, including Chinese understanding ability, video generation ability, and Chinese dialect ability, to introduce to you the video capabilities of Xixin Big Model 4.0 and the plug-in Yijingliuying.
Comprehension ability, especially Chinese understanding ability
Let’s first take a look at the understanding ability of Wenxin Large Model 4.0. The main test here is its ability to deal with "language traps" and its "recognition ability" of technical terms. The answer of Wenxin Large Model 4.0 is very concise and gives the answer directly.
The answer is Docker is an open source application container engine in the computer and its role.
Generating capabilities, especially the ability to generate videos
So in the next wave, we will test the ability of Wenxinyiyan 4.0 to generate videos to see if it has this magical ability. At this time we have to take a look at Wen Xin Yi Yan’s plug-in, Yijing Liuying (text to video), which is the ability of Wen Xin Yi Yan 4.0 to generate videos.
"Yijing Liuying" is a plug-in in Wenxin Yiyan 4.0. It uses a technical means to automatically convert text content into video form, so that the text can be presented in a more vivid and vivid way. This technology can be applied to various text contents, such as news reports, novels, poems, etc., to transform them into dynamic images with visual effects, bringing a richer and more intuitive viewing experience to the audience. One-mirror streaming technology can greatly improve the expressiveness and attractiveness of content, helping to better convey the information and emotions contained in the text.
Let’s try next and use the Yijingliuying (text to video) function to generate a video. Using Wenxinyiyan 4.0's Yijingliuying, copywriting: What is docker, generate a 30-second video with detailed explanation.
Let's take a look at the video generated by the Yijing Liuying plug-in. It not only generates a clip of the docker container, but also unexpectedly has the copywriting and subtitles and voices, which is very complete. Seeing this, I have to lament the advancement of technology. Video editing is no longer just for professional operators. We use the Wenxinyiyan 4.0 plug-in One Shot Streaming to quickly produce and output a high-quality promotional video.
Mandarin version of docker explanation
dialect ability
Then, there is the dialect proficiency test. Wenxin Big Model 4.0 is said to have upgraded its mathematical computing capabilities, so we directly tested it on the dialects to see if it can support various dialects:
The following is a video script introducing Docker broadcast in Sichuan dialect: [Background music starts, relaxed and lively] Narrator (Sichuan dialect): What is Docker? Docker is an open source application container engine that allows us to package applications and dependency packages into a portable image and publish it to any popular Linux or Windows operating system. [Camera switches to show the Docker icon and official website] Narrator (Sichuan dialect): Docker can help us deliver applications faster, isolate the application and infrastructure layers, and manage infrastructure like a hypervisor. [Camera switches to show an example of Docker container running] Narrator (Sichuan dialect): Using Docker, we can package, test and deploy applications faster, shortening the cycle from writing to deploying and running code. [Camera switches to show the application of Docker in the development process] Narrator (Sichuan dialect): In short, Docker is an artifact that allows us to easily deploy and manage applications!
Sichuan dialect version of docker explanation
Cantonese version of docker explanation
Summarize
At this point in the test, I might as well make a small summary: overall, the Chinese understanding ability, video generation ability, and Chinese dialect ability are very strong. First of all, it can generate a video script based on our conversation; secondly, it can convert the video script into a video through a plug-in; finally, it can also change the language at will, and can support Chinese dialects.
From the release of Wenxin Yiyan in March, to the 3.5 version update of the Wenxin model in the middle of the year, and now the stunning debut of version 4.0, the iteration speed of Baidu Wenxin model has been rapid. In this review, the amazing performance in video generation is indeed amazing, subverting traditional video editing. Once again, it reflects Baidu’s profound technical accumulation in the field of large models.
Looking forward to the new era of AI native, Wen Xinyiyan's rapid progress in the basic capabilities of large models, as well as the subversive operation and performance of AI native application development, all make people sigh. The future is an era of technology and intelligence.