home page
About this site
recommend
To be kicked
Search
one
Global avatar Gravatar application course
42941 Reading
two
How many kilometers does Chaoshan dialect pave the way?
29806 Reading
three
What is Chaoshan Bayin
20858 Reading
four
Common DNS resolution servers in Guangzhou
18850 Reading
five
How to view the saved FTP site password of FileZilla
15385 Reading
Online knowledge
front-end technology
Electronic Commerce
operating system
Application software
Microknowledge
Household Encyclopedia
Sign in
Search
Tag Search
windows
WeChat
mysql
twelve thousand three hundred and six
Acetaminophen
ibuprofen
win10
browser
data base
excel
dns
febrifuge
two thousand and twenty-four
Rural medical insurance
ncms
License plate information
vpn
l2tp
qq
office
Laolang Knowledge Base
Accumulated writing
one hundred and forty-five
Articles
Accumulated receipt
twenty-six
Comments
home page
column
Online knowledge
front-end technology
Electronic Commerce
operating system
Application software
Microknowledge
Household Encyclopedia
page
About this site
recommend
To be kicked
Search to
two
Article and
Results for
2024-06-26
OpenAI Announces Termination of Service to China
On the morning of June 25, many netizens shared the latest announcement of OpenAI in their circle of friends, which said: "Our data shows that your organization has API traffic from regions that OpenAl does not currently support." It is understood that on June 25, 2024,
OpenAI sent an email to Chinese API users, saying that it would take measures to block API traffic from regions on the list of countries and regions that are no longer supported.
From July 9, OpenAI will block unsupported national APIs.
If users use their APIs in countries where OpenAI does not allow them, they will be blocked.
This means that OpenAI unilaterally announced for the first time that it would stop providing API services to China.
The direct impact of this decision is that companies that had hoped to start their businesses with the help of OpenAI's big model will face a devastating blow.
From another perspective, OpenAI will force all Chinese enterprises to use domestic models.
In fact, in many scenarios, many domestic open source models are no worse than closed source models, which are fully enough to support the development of Chinese AI applications.
For example, in terms of Chinese knowledge, Ali has better understanding than Llama 3.
At present, many start-ups use open source models with 10 million parameters to train their vertical models.
Therefore, China's domestic models may usher in a new wave of growth!
Related Background On June 21, the U.S. government just released the so-called targeted draft rules that prohibit the U.S. from making certain investments in China's artificial intelligence (AI) and other technology fields that may threaten the national security of the United States, or require notification of these investments.
1. Americans are prohibited from participating in transactions involving certain technologies and products that pose a particularly serious threat to the national security of the United States;
2. Americans are required to notify the Treasury of other transactions involving certain technologies and products that may pose a threat to the national security of the United States.
In response, the Ministry of Commerce said on June 24 that China had noticed the release of relevant documents by the United States.
The Ministry of Commerce said that the United States should respect the laws of market economy and the principle of fair competition, stop politicizing and weaponizing economic and trade issues, and cancel restrictions on investment in China, so as to create a good environment for Sino US economic and trade cooperation.
It is understood that the United States will complete the public opinion collection on this ban before August 4, which means that the United States is expected to implement this new regulation before the end of the year as expected.
The draft regulates certain US investments in semiconductor, microelectronics, quantum computing and AI, with the aim of preventing US proprietary technology from helping China develop cutting-edge technology or dominate the global market.
June 26, 2024
26 Reading
0 Comments
1 Like
2024-05-16
Mysterious code time. sleep (6), appearing in Huawei's big model release
Today, Huawei demonstrated the Wensheng diagram capability of the big model at the press conference. After the first run of the demonstration, pressing Ctrl-C interrupted, and a line of code was displayed on the screen: time. sleep (6).
This has aroused many netizens' curiosity about what happened when they discussed time. sleep (6).
1. The difference between the demonstration and the actual product operation At the product launch, the manufacturer usually carefully prepares the demonstration link to show the best state and function of the product.
However, this demonstration does not necessarily fully reflect the operation of the actual product.
The manufacturer may take some measures, such as pre recording video, using analog data, etc., to create a smoother and better presentation effect.
For example, at Baidu Wenxin Yiyan's press conference, Robin Li said that in order to "save time", the display was recorded in advance.
Similarly, when Google released the Gemini big language model, it also revealed that the demonstration video was not recorded in real time, but "pieced together" through still image frames and text prompts.
This practice may give the audience the illusion that the product performance is better.
2. The user is concerned about the actual performance of the product. Although there may be some differences in the demonstration, the user is most concerned about the actual performance after the product is released.
Even if the demonstration scheme is pre recorded, it will not affect the public's curiosity and willingness to use the product.
As shown by ChatGPT of OpenAI and Wenxin Yiyan of Baidu, even if there are some "cheating" practices in the demonstration, users will still pay high attention and praise as long as the product can finally meet user needs and solve practical problems.
3. Users' feedback after the product is released. Users are the real "judges" after the product is released.
Whether it is a large model or other software products, users will judge the quality of the products through actual use.
If the product is easy to use and can solve users' problems, users will naturally give positive feedback and high praise.
On the contrary, if the product cannot meet the user's needs, the user will criticize and complain without mercy.
Take Google Gemini as an example. Although there is some suspicion of "cheating" in the demonstration process, Nat Friedman said that Gemini's traffic has reached 1/4 of ChatGPT, which shows that users still have high expectations and interest in the product.
4. The manufacturer's marketing strategy product release is an important channel for the manufacturer to display and promote products.
In this process, the manufacturer will try to show the best state of the product to attract the attention and interest of users.
This does not mean that there is a problem with the product, but a common marketing strategy of the manufacturer.
The key is that even if there are some "cheating" practices in the demonstration, as long as the product can finally meet user needs and solve practical problems, the user is the real judge.
User feedback and evaluation is the key factor to determine the success or failure of the product.
May 16, 2024
499 Reading
0 Comments
0 likes