From the domestic AI giants to the global top five big model companies, how can cloud take it easy?
In the chaotic era of big models, Cloud has finally taken the first step in its own way.On May 30th, Yun Cong Rong announced the official launch of the public beta for the large model, which was less than two weeks after its debut on May 18th
In the chaotic era of big models, Cloud has finally taken the first step in its own way.
On May 30th, Yun Cong Rong announced the official launch of the public beta for the large model, which was less than two weeks after its debut on May 18th.
In terms of time, it has been over six months since the release of ChatGPT, and it is not too early; But in terms of attention, as the most high-profile AI giant in the capital market in the past few months, Yuncong's every move is enough to be interpreted and analyzed by hundreds of millions of shareholders using a magnifying glass.
So, amidst all the attention, the time from its debut to the official public beta was less than two weeks away,Where exactly does Yun Cong's confidence come from? Compared to ChatGPT, which has attracted hundreds of millions of people's attention, what is the calm performance of Yuncong?
According to the introduction of Yuncong's leisurely big model, it is a Natural language processing assistant developed by CloudWalk Technology to help people complete various tasks more efficiently, such as answering questions, providing suggestions, generating code, etc. Therefore, 'calm' can also be understood as an efficient and convenient way of working.
According to the official website of Congrong, we can find that its functions mainly focus on nine dimensions: knowledge Q&A, creative copywriting, logical reasoning, article writing, common sense Q&A, code writing, language understanding, daily life, and professional knowledge.
So, next, we will conduct a horizontal evaluation and comparison of the six classic problems in the above fields, and take a look. What is the quality of Yuncong's calm big model?Better understandingBetter understandingAnyone who has deeply experienced the cloud from the calm big model can easily come to the conclusion that as a local AI big model trained from a rich Chinese corpus, the calm big model has significant advantages in understanding Chinese semantics and mastering Chinese corpus knowledge.
Firstly, compare the performance of different large models in understanding Chinese semantics and the true deep meanings behind semantics.
Taking the high-frequency response of "you are a good person" in emotional interaction between men and women as an example
ChatGPT, understanding the first and most superficial meaning behind this sentence, you are a decent person. However, there is a certain deviation in the understanding of the second meaning, which does not necessarily indicate that she has a clear response to your feelings or whether she intends to accept your confession.
In fact, in the implicit confession scene in Chinese, sending a good person card is actually a refusal meaning of "using affirmative sentences to express negation". Correspondingly, based on this erroneous judgment, the follow-up questioning style action guidance given is naturally adding mistakes to mistakes.
From the calm model of the cloud, we can see that after understanding the literal meaning of the first layer, even the second layer behind it, "she may only treat you as a friend or an ordinary colleague," has been accurately interpreted;
But what's even more surprising is that after interpreting the second layer of meaning, the third layer of action guidance provided by the calm big model is undoubtedly more appropriate and reasonable - "more action and communication are needed to make her feel your sincerity and sincerity", rather than continuing to ask "what exactly does your response mean.
After accurately understanding the questioner's meaning, the second question we need to consider is,Is it true or false understanding of AI in response to our question.Shortly after the launch of ChatGPT, many users discovered that AI always gave themselves a serious wrong answer with a wrong lip.
Or use a classic Chinese love question to test - 'Why do couples always say when they make wishes,' Be like birds in heaven '?
This is the answer from chatGPT, as you can see, AI gave the wrong steps to solve the problem, but mistook the correct answer. The verses originally belonging to the Song of Everlasting Regret were placed on the Mulan Poems and Nalan Xingde. In addition, a plausible fairy story was also invented, but in the use scene, the answer that symbolizes the expectation of good love was accidentally answered correctly.
Compared to other models, the level of understanding of Chinese in the calm model clearly shows that it is better. Not only did it accurately identify the source of the poem "Song of Eternal Sorrow", but it also provided precise explanations for the contextual background of the poem; At the same time, the description of the usage situation and the true meaning of the poem should be more precise and precise compared to chatGPT.
Of course, Yuncong's calm big model is not omnipotent. It still involves love questions. With a slight change in the way we set the questions, we can see that both foreign chatGPT and Chinese calm big models are immature like elementary school students when faced with the death problem of "your family and I fell into the water at the same time".
More scenarios and more professional answers
Of course, in addition to the most basic semantic understanding and corpus training, what we are more concerned about is actually the transformation and acceleration that cloud computing, as a large model, can bring to our productivity.
Generally speaking, the most commonly used AI text generation scenarios are none other than three categories: 1. Summarizing long article content to save time; 2. Help us write weekly reports, emails, and complete some basic transactional tasks; 3. In some creative copywriting, give us some ideas and inspiration.
Let's first take a look at the summary text. Taking the recent article "Artificial Intelligence, Pragmatic Development is the Right Way" published by the People's Daily as an example, we can see that ChatGPT and Yuncong have their own strengths and weaknesses in their calm performance.
Among them, the summary of chatGPT is more detailed, but relatively lengthy; A calm and large model, with a small overall word count, can also accurately summarize the core ideas and content of the article.
Next, let's take a look at how the two perform in specific scenarios such as writing emails:
It can be seen that regardless of which product, AI knows how to apologize. Compared to other methods, ChatGPT focuses more on root cause analysis and the provision of specific solutions in the future; Yuncong's calm email is more concise and clear, and which one is better depends on our needs.
The last question, inspired by creative copywriting, is to write a new fairy tale about a princess arranging a dragon to steal a marriage.
Among the two responses, the calm model maintained its consistent and concise style. At the end of the story, the setting where the dragon transformed the prince into a frog was eye-catching;
ChatGPT still provides a very rich content filling, and the setting of specific characters' personalities and plot design in the story are more comprehensive.In the above few questions and answers that are not clear about right or wrong, we can see that the cloud style is more concise, while GPT is known for its details. For users with different needs, it may be a good choice to combine the two.From the domestic AI giants to the global top five big model companies, how can cloud take it easy?
To a certain extent, in China.
...
But when it comes to more specific details, we will find that,AI.
.2015z.
AIAI.
2022527AIIPO.
.
2016800.5%50%.
100039.21%.
8.2019CWOS operating system.
2021AI2022The launch of the Rongda model is preciselyCWOS operating system.
.
After understanding the ease of cloud technology,Another question immediately surfaced, why did Yuncong's large model arrive late, but on the ground, it was able to take the lead?
.
.
....
CWOS operating system.
Disclaimer: The content of this article is sourced from the internet. The copyright of the text, images, and other materials belongs to the original author. The platform reprints the materials for the purpose of conveying more information. The content of the article is for reference and learning only, and should not be used for commercial purposes. If it infringes on your legitimate rights and interests, please contact us promptly and we will handle it as soon as possible! We respect copyright and are committed to protecting it. Thank you for sharing.(Email:[email protected])