＂Guidelines for the Construction of Child-friendly Enterprises in Wenzhou (Trial)＂ was released

A few days ago, the reporter learned from the Office of the Leading Group for the Construction of Child-friendly Cities in Wenzhou and the Municipal Bureau of Economic and Information that in order to steadily promote the development of child-friendly industries in Wenzhou and strive to provide safe products and services for children, the city officially issued the Guidelines for the Construction of Child-friendly Enterprises in Wenzhou (for Trial Implementation) (referred to as the Guidelines).

According to the four principles of giving priority to children, guarding growth, ensuring safety, meeting needs, caring for employees, cohesive force, children’s participation, and building and sharing, the Guidelines define the development goals of children and enterprises, and realize the "two-way trip" between enterprises and children through six major construction contents.

Construct a systematic and complete institutional system

The "two-way trip" between enterprises and children should start with "building a systematic and complete system". The newly issued "Guidelines" clearly stated that it is necessary to improve the child-friendly system and work coordination mechanism, and promote the integration of child-friendly concepts into enterprise planning and construction.

Specifically, it is to adhere to the principle of maximizing children’s interests, cut in from the aspects of enterprise management, product and service supply, space creation, development environment, employee care, etc., and promote the deep integration of child-friendly concept and enterprise sustainable development as a whole.

Provide high-quality products and services

The Guidelines emphasize that enterprises should ensure the safety of products and services, and ensure that the products and services they provide are studied and tested according to relevant national and local standards, and will not cause mental, psychological or physical harm to children. In addition, it is necessary to establish a protection mechanism for children’s personal information to avoid information leakage.

At the same time, enterprises should improve the adaptability of products and services, and the products and services provided should be functional, interesting, interactive and iterative. In marketing, we should use marketing methods that respect children’s rights and carry out educational activities for children.

Help the growth of children with special needs

The Guidelines care for the healthy growth of children with special needs. In addition to encouraging enterprises to provide barrier-free products and services, products should have clear barrier-free labels, and support the healthy growth of children with special needs. It is proposed that voluntary service activities should be carried out in a planned, targeted and sustainable way in the life and study of children with special needs.

Especially for children affected by natural disasters, diseases and other emergencies, such as relief materials, medical rehabilitation, academic counseling, psychological counseling, legal services and so on.

Promote the construction of friendly space for enterprises

In terms of strengthening the construction of child-friendly spaces, the Guidelines clarify that the site selection of industrial enterprises should be based on the existing laws, regulations and standards of health, safety in production and environmental protection in the country and the region to avoid adverse effects on children’s living, learning, entertainment and other growth environments. Encourage qualified enterprises to set up children’s activity rooms for employees’ children to read, study and hold activities. Enterprises can set up non-profit nursery places and kindergartens, or entrust a third party to provide universal nursery and preschool education services for their employees.

In addition, promoting the construction of enterprise-friendly space also requires continuous improvement of equipment and facilities, such as encouraging the construction of maternal and child facilities, the third bathroom (family bathroom) and barrier-free facilities.

Deeply cultivate the friendly development of children’s industry

The Guidelines help the construction of child-friendly industries, and propose that enterprises should actively participate in the construction of Wenzhou’s child-friendly industrial chain, combine their own business, exchange needed goods and complement each other’s advantages with other enterprises committed to child-friendly undertakings, and jointly build a child-friendly industrial chain with Wenzhou characteristics.

The Guidelines encourage the government, outstanding child-friendly enterprises and chambers of commerce to take the lead in setting up child-friendly enterprise alliances, co-ordinate high-quality resources, and help build child-friendly industries systematically and professionally.

Create a positive working environment

The Guidelines require enterprises to implement various measures to care for employees, such as unblocking career paths and providing employees with statutory sick pay, overtime pay and social security. Encourage young employees to solve housing problems, set up a love fund for employees with difficulties, and provide decompression facilities and psychological counseling.

For employees with children, enterprises are encouraged to provide them with affordable and high-quality childcare services. Provide grants, entrance awards, children’s business insurance, multi-child family benefits, etc. for employees’ children, and regularly carry out parent-child activities such as summer camp, theme study, parent-child sports meeting, parent-child reading club and family reception days.(Reporter Xia Wei correspondent Li Yangci)

Twenty-five years after its birth, One Piece continues to take risks.

[Global Times special correspondent Luc] Since it was serialized in the Weekly Youth JUMP in 1997, the straw hat Luffy and various adventure stories in the Japanese cartoon One Piece have attracted a large number of readers and audiences all over the world. According to the official Twitter of One Piece on the 16th, the latest theatrical version of this series, The Sea King: The Red-haired Geji, broke the highest box office record of this series in one fell swoop 10 days after it was released in Japan. At present, the domestic box office in Japan has exceeded 7 billion yen (about 350 million yuan), and the number of people watching movies has exceeded 5 million. At the same time, at the ongoing Beijing International Film Festival, this theatrical version, which is almost synchronized with Japan, is also hard to get a ticket, and the ticket price has even been fired to 30,000 yuan.

Stills of Nautical King: The Red-haired Geji

As the 15th theatrical version of the series, Voyager: The Red-haired Geji was officially released in Japan on August 6th, and it made a box office success of 2.25 billion yen in the first weekend, and made persistent efforts in the next weekend, with the box office successfully exceeding 7 billion, which is second only to the 2020 Blade of the Ghost: Infinite Train. "Red-haired Geji" revolves around the new key figure Uta, whose singing from another dimension can make the whole world crazy. As the plot unfolds, Luffy discovers that Wuta was originally the daughter of her benefactor-Janks with red hair, but the relationship between father and daughter is quite subtle … The biggest attraction of the film is that Wuta can control the super power of the "conscious world", and the final duel is also opened in two parallel worlds. Different from the previous theatrical editions, the cartoonist Eiichiro Oda himself is the producer of "The Red-haired Geji" for the first time, which is a reassurance for the "pirates fans" all over the world.

As early as 2014, "One Piece" was recognized by Guinness World Records as "the comic book with the largest cumulative circulation created by a single author", and now its cumulative circulation has exceeded 500 million copies, and it has sold well in more than 60 countries and regions around the world. This year’s Beijing International Film Festival has prepared the special unit "One Piece: The Sailor King: The First Touch", including three theatrical works: The Battle of Alabastan, The Queen of the Desert and the Pirates, The Mystery of Qiao Ba’s Life: Miraculous Cherry Blossoms in Winter, and The Red-haired Geji, which were "empty" when invoicing last weekend, and even had a platform to fry the package to a sky-high price of 270,000 yuan. Previously, two theatrical versions of Crazy Action (2019) and Huang Jincheng (2016) were released in China, which earned 200 million yuan and 100 million yuan respectively.

In addition to the China market, One Piece is not well-known in Europe and America. The streaming media giant Netflix officially started the production of the live-action version of One Piece at the beginning of this year, and the young actors from different parts of Asia, Africa and Latin America formed the "Straw Hat Pirates". Judging from the stills currently exposed, the "Merry" and the town have a high degree of restoration. However, although Eiichiro Oda personally acted as a live-action producer and participated in casting, the vast number of "pirates fans" are still worried about the level of this drama series-after all, there are more than 1,000 original works, but there are only 10 episodes in the drama series, so whether the people and special effects can meet the imagination in animation will become the key.

Who says elephants can’t dance! Programmed large language model to realize time series prediction of cross-modal interaction ｜ ICLR 2024

The heart of the machine column

Machine zhixin editorial department

Recently, researchers from monash university in Australia, ant group, IBM research institute and other institutions have explored the application of model reprogramming in large language models (LLMs). It also puts forward a brand-new perspective: efficiently reprogramming the large language model for general time series prediction-its proposed Time-LLM framework can realize high-precision time series prediction without modifying the language model, surpassing the traditional time series model in multiple data sets and prediction tasks, and making LLMs perform well in dealing with cross-modal time series data, just like an elephant dancing!

Recently, inspired by the big language model in the field of general intelligence, the new direction of "big model+time series/spatio-temporal data" has made a lot of relevant progress in generate. At present, LLMs has the potential to completely change the way of time series/spatio-temporal data mining, thus promoting the efficient decision-making of typical complex systems such as cities, energy, transportation and remote sensing, and moving towards a more universal intelligent form of time series/spatio-temporal analysis.

Large-scale models, such as language and other related basic models, can be trained and skillfully readjusted to deal with time series and spatio-temporal data in a series of general tasks and applications in special fields. Source: https://arxiv.org/pdf/2310.10196.pdf.

Recent studies have expanded large-scale language models from dealing with natural languages to time series and spatio-temporal tasks. This new research direction, that is, "large model+time series/spatio-temporal data", has given birth to many related developments, such as LLMTime directly using LLMs for zero-sample time series prediction and reasoning. Although LLMs has powerful learning and presentation ability, and can effectively capture complex patterns and long-term dependencies in text sequence data, as a "black box" focusing on natural language, its application in time series and spatio-temporal tasks still faces challenges. Compared with traditional time series models such as TimesNet and TimeMixer, LLMs can be compared with "elephant" for its huge parameters and scale.

Therefore, how to "tame" LLMs, which is trained in the field of natural language, so that it can process numerical sequence data across text patterns and exert its powerful reasoning and prediction ability in time series and spatio-temporal tasks has become the key focus of current research. Therefore, a deeper theoretical analysis is needed to explore the potential pattern similarity between language and time series data, and effectively apply it to specific time series and spatio-temporal tasks.

This paper expounds how to make general time series prediction by reprogramming LLM Reprogramming. It puts forward two key technologies, namely, (1) reprogramming the time sequence input and (2) prefixing the prompt, which transforms the time sequence prediction task into a "language" task that can be effectively solved by LLMs, and successfully activates the ability of large language model to do high-precision time sequence reasoning.

Address: https://openreview.net/pdf?. id=Unb5CVPtae

Paper code: https://github.com/KimMeen/Time-LLM.

1. Background of the problem

Time series data exist widely in reality, among which time series prediction is of great significance in many dynamic systems in the real world and has been widely studied. Different from natural language processing (NLP) and computer vision (CV), a single large-scale model can handle multiple tasks, and the time series prediction model often needs to be specially designed to meet the needs of different tasks and application scenarios. Although the basic model based on pre-training has made great progress in NLP and CV fields, its development in time series field is still limited by data sparsity. Recent research shows that large language models (LLMs) have reliable pattern recognition and reasoning ability when dealing with complex tag sequences. However, how to effectively align the two modes of time series data and natural language, and how to use the reasoning ability of large language model itself to deal with the task of time series analysis is still a challenge.

2. Overview of the paper

In this work, the author puts forward Time-LLM, which is a general framework of LLM Reprogramming. LLM can be easily used for general time series prediction without any training for the large language model itself. Time-LLM first reprogrammed the input time series data with Text Prototypes, and expressed the semantic information of the time series data by using natural language representation, and then aligned two different data modes, so that the large language model could understand the information behind the other data mode without any modification.

In order to further enhance LLM’s understanding of input time-series data and corresponding tasks, the author puts forward a paradigm of Prompt-as-Prefix (PaP), which fully activates LLM’s processing ability in time-series tasks by adding additional context prompts and task instructions before time-series data representation. In this work, the author has carried out sufficient experiments on the mainstream time series benchmark data sets, and the results show that Time-LLM can surpass the traditional time series model in most cases, and has greatly improved the learning tasks of Few-shot and Zero-shot.

The main contributions in this work can be summarized as follows:

1. This work puts forward a new concept of time series analysis by reprogramming large language model, without any modification to the main language model. The author shows that time series prediction can be regarded as another "language" task that can be effectively solved by ready-made LLM.

2. This work puts forward a general language model reprogramming framework, namely Time-LLM, which includes reprogramming the input time series data into a more natural text prototype representation, and enhancing the input context through declarative prompts (such as domain expert knowledge and task description) to guide LLM to conduct effective cross-domain reasoning. This technology provides a solid foundation for the development of multi-modal time series basic model.

3. The performance of Time-LLM in mainstream forecasting tasks always exceeds the performance of the best existing models, especially in scenes with few samples and zero samples. In addition, Time-LLM can achieve higher performance while maintaining excellent model reprogramming efficiency. Greatly release LLM’s untapped potential in time series and other sequential data.

3. Model framework

As shown in ① and ② in the model frame diagram above, the input time series data is normalized by RevIN, and then cut into different patch and mapped to the hidden space.

There are significant differences between time series data and text data in expression, and they belong to different modes. Time series can neither be directly edited nor described in natural language, which brings great challenges to prompting)LLM to understand time series. Therefore, we need to align the time series input features to the natural language text domain.

A common method to align different modes is cross-attention, as shown in ③ in the model frame diagram, which only needs to make a cross-attention for the embedding and time sequence input features of all words (in which the time sequence input feature is Query, and the embedding of all words is Key and Value). However, LLM’s inherent vocabulary is very large, so it can’t directly align the time series features to all words effectively, and not all words have aligned semantic relations with time series. In order to solve this problem, this work combines vocabularies linearly to obtain text prototypes, in which the number of text prototypes is much smaller than the original vocabulary, and the combination can be used to represent the changing characteristics of time series data, such as "short rise or slow decline", as shown in the above figure.

In order to fully activate LLM’s ability to specify time-series tasks, this work puts forward the paradigm of prompt prefix, which is a simple and effective method, as shown in the model frame diagram. Recent progress shows that other data patterns, such as images, can be seamlessly integrated into the prefix of prompts, thus making effective reasoning based on these inputs. Inspired by these findings, in order to make their methods directly applicable to real-world time series, the authors put forward an alternative question: Can hints be used as prefix information to enrich the input context and guide the transformation of reprogramming time series patches? This concept is called Prompt-as-Prefix (PaP). In addition, the author also observed that it significantly improved LLM’s adaptability to downstream tasks and supplemented the reprogramming of patches. Generally speaking, it means that some prior information of time series data set is used as prefix prompt in the way of natural language, and it is spliced with aligned time series features to feed to LLM. Can it improve the prediction effect?

The figure above shows two prompting methods. In Patch-as-Prefix, the language model is prompted to predict the subsequent values in the time series, which are expressed in natural language. This method encounters some constraints: (1) Language models usually show low sensitivity when dealing with high-precision numbers without the assistance of external tools, which brings great challenges to the accurate processing of long-term forecasting tasks; (2) For different language models, complex customized post-processing is needed, because they are pre-trained on different corpora and may use different word segmentation types when generating high-precision numbers. This leads to the prediction expressed in different natural language formats, such as [‘0′,’.’,’ 6′,’ 1′] and [‘0′,’.’,’ 61′], representing 0.61.

In practice, the author identified three key components to construct effective hints: (1) the context of data set; (2) Task instruction, which makes LLM adapt to different downstream tasks; (3) Statistical descriptions, such as trends and time delays, enable LLM to better understand the characteristics of time series data. The following figure shows an example of a prompt.

4. Experimental effect

We have conducted a comprehensive test on eight classic public data sets in long-range forecasting. As shown in the following table, the benchmark comparison of Time-LLM is significantly superior to the previous best results in the field. In addition, compared with GPT4TS which directly uses GPT-2, Time-LLM which adopts the idea of reprogramming and prompts as-prefix is also significantly improved, which shows the effectiveness of this method.

In addition, we evaluated the zero-shot learning ability of reprogrammed LLM within the framework of cross-domain adaptation. Thanks to the reprogramming ability, we fully activated LLM’s prediction ability in cross-domain scenarios. As shown in the following table, Time-LLM also showed extraordinary prediction effect in zero-shot scenarios.

Step 5 summarize

The rapid development of large-scale language models (LLMs) has greatly promoted the progress of artificial intelligence in multimodal scenes, and promoted their wide application in many fields. However, LLMs’s huge parameter scale and its design mainly aimed at natural language processing (NLP) scenarios have brought many challenges to its cross-modal and cross-domain applications. In view of this, we put forward a new idea of reprogramming large model, aiming at realizing the cross-modal interaction between text and sequence data, and this method is widely used to deal with large-scale time series and spatio-temporal data. In this way, we expect LLMs to be like a flexible dancing elephant, which can show its powerful ability in a wider application scenario.

Interested friends are welcome to read the paper (https://arxiv.org/abs/2310.01728) or visit the project page (https://github.com/KimMeen/Time-LLM) for more information.

This project has been fully supported by NextEvo, the AI innovation R&D department under the Intelligent Engine Division of Ant Group, especially thanks to the close cooperation between the language and machine intelligence team and the optimization intelligence team. Under the leadership and guidance of Zhou Jun, Vice President of Intelligent Engine Division, and Lu Xingyu, Head of Intelligent Optimization Team, we have successfully completed this important achievement hand in hand.

Baidu won again! The domestic AI model has made a new breakthrough.

Science and technology are the primary productive forces.

People who master advanced scientific and technological tools are not only "promoted and raised", but also "one step ahead of others".

The AI big model is the most advanced technology tool that is popular all over the world.

"The emergence of intelligence brought by the big model is the basis for us to develop AI native applications." On October 17, Li Yanhong said at Baidu World 2023. On the same day, Li Yanhong delivered a speech on the theme of "Teach You to Do AI Native Applications by Hands", released version 4.0 of Wenxin Big Model, and brought more than ten AI native applications such as new search and new maps.

First, the AI ​ ​ big model, Wenxin 4.0

Li Yanhong is full of energy on the podium.

At the "2023 Baidu Yunzhi Conference" just past September 5, Li Yanhong, founder, chairman and CEO of Baidu, pointed out in the award speech of Baidu "Wenxin Cup" Entrepreneurship Competition that "AI native application is the meaning of the existence of big models".

Li Yanhong said that the model itself does not directly generate value, and the application developed based on the basic big model is the meaning of the model. For entrepreneurs, "the big model is meaningless, and the application opportunities are greater". He analyzed that there are only Android and iOS operating systems in the mobile Internet era, but there are many particularly successful applications. Only by generating enough AI native applications on the basis of large models is a healthy ecological environment.

At today’s "Baidu World Congress" on October 17th, Li Yanhong officially released Baidu’s Wenxin Big Model 4.0 with the theme of "Teaching you how to make AI native applications by hands".

Li Yanhong explained the four core abilities of Wenxin Big Model: understanding, generation, logic and memory, and once again stressed:

"Without rich AI native applications built on the basic model, the basic model has no value."

At the Baidu World Congress alone, more than a dozen important AI upgrades of Baidu products were released in one breath, and nearly 50 immersive AI technology exhibitions were held at the venue.

Let’s walk into the venue of China AI Big Model 4.0 era.

Second, Wen Xin big model, application trust-mart.

Wenxin model is an industrial-level knowledge enhancement model independently developed by Baidu, which has formed a three-layer model technology system of foundation-task-industry, including basic models such as natural language processing, vision, cross-modal, task models such as dialogue, cross-language, search, information extraction, bio-computing field model, industry model and tool platform to support the application of large models.

The overall model effect, training efficiency, training performance and reasoning performance of Wenxin Big Model 4.0 have been greatly improved.

It is worth mentioning that Baidu released a number of products based on Wenxin model at this conference, such as: Baidu New Search.

"Search" is the first product reconstructed by Baidu through a large model.

Li Yanhong said that generative AI and search are a perfect match. Search, although still starting from that simple box, has surpassed and reconstructed the original search mode in the next step, which not only improves the search quality, but also breaks through the application scope of search and expands the product boundary. Such AI interactive search will lead the intergenerational transformation of the search industry.

There is also Baidu GBI (Generative Business Intelligence). Baidu GBI is the first generative business intelligence product in China.

Baidu GBI (Generative Business Intelligence) organically combines Wenxin Big Model with BI-like scenes, supports interactive dialogue through natural language, performs data query and analysis tasks, realizes "any table, any question", and establishes a new data analysis paradigm of "dialogue is insight" for corporate customers.

Traditional BI tools are difficult to find data, use tools and have limited index coverage. Both Excel and data analysis platform are designed for high-frequency (preset) scenes, so they cannot flexibly deal with various problems at any time. The analysis and insight of a complex problem often takes hours or days to complete. At the same time, these analyses are also a part of business decision-making. In the fierce and rapidly changing market, such a long decision-making cycle is usually unacceptable. The main contradiction is the contradiction between the high threshold of traditional BI and data-driven business requirements.

The emergence of AI large model technology has brought a brand-new "problem-solving idea" to these problems.

Baidu GBI allows users to "ask any one or more data tables with any questions and get a conclusion quickly".

At present, Baidu GBI has prefabricated three flexible access schemes for customers with different preparation levels.

? Quick experience, verification effect: For customers who expect quick experience effect, you can upload, query and analyze data directly on Baidu GBI page.

? Complete access, independent display: customers have their own front-end interface of the system, and the underlying capabilities related to the number of applications are all provided by Baidu GBI, thus achieving end-to-end integration and unified result presentation.

? Tool collaboration and ability enhancement: If the customer already has a perfect database and BI tool construction, he only needs to enhance the ability of "table inquiry", and can only access Baidu GBI’s intelligent number inquiry module to improve the effect and user experience, and the functions of data source access and presentation are still realized in the customer’s original system.

Among them, Baidu’s new library is indispensable. The new Baidu Library is a product that Baidu has reconstructed relatively thoroughly at present.

Baidu Library has 1 billion high-quality documents. Based on the understanding of these contents, it can write truly practical and clear manuscripts, which will neither be pure fiction nor general empty talk. Li Yanhong pointed out that, based on the four capabilities of Wenxin Big Model, it is no exaggeration to say that Baidu Library fundamentally changed its life and completed the evolution from "content tool" to "productivity tool".

For example, the function of "document generating PPT" is more complicated in understanding and deconstructing the key points, writing logic and content style of the article, and the difficulty is far beyond "generating PPT in one sentence"; "Wensheng Chart" charts data-based content and quickly forms analytical insight; Professional charts can be generated intelligently in PPT, which supports online editing and flexible modification; Long articles with ten thousand words can be summarized with one click, and personalized questions and answers can be conducted based on the article, so as to fully understand and learn the content of the article, and the functions of intelligent abbreviation, continuation and expansion can break through technical difficulties, allowing users to realize "intelligent creation" at will.

In just over a month, the new function of Baidu Library AI has been widely welcomed and praised by users, with the cumulative number of users exceeding 10 million and the cumulative number of uses being nearly 100 million. Accumulated PPT is over 2 million, and generated content is over 20 million.

Also, Baidu released the spiritual realm plug-in platform.

? Spiritual Matrix is a plug-in development platform for all kinds of developers, and it is the first plug-in development platform in China to realize the whole process automation.

The spirit realm plug-in platform faces three kinds of objects:

? Developers with development ability: provide a complete set of technical solutions.

? Developers with rich data: provide code-free solutions for visual layout, suitable for developers with limited development ability but data.

? Developers with ideas: provide plug-in creation methods for prompt debugging, and it only takes 5-10 minutes to complete development.

Li Yanhong said that the plug-in is a special AI native application, and it is also the lowest threshold and the easiest to use.

At present, 27,000 developers have applied to settle in the Lingjing plug-in platform for one month, covering 20+ fields, including legal consultation, resume generation, brain mapping, oral practice, contract review and so on. Individual developers account for over 30%.

As a support, Baidu released the "Star Start Plan" to provide developers with a huge flow pool, tens of millions of computing power support, billions of funds and complete ecological resources.

At the same time, Baidu AI Cloud Qian Fan Big Model Platform is the world’s first one-stop enterprise-level big model production platform launched by Baidu AI Cloud. The successful practical experience of building AI native applications is summarized into tools and methodologies, which are deposited on the world’s first one-stop big model service platform "Qian Fan" to help entrepreneurs and enterprises build their own AI native applications with low cost and high efficiency. Qian Fan not only provides big model services based on ERNIE Bot or third-party open source big models, but also provides a full set of tool chains and development environment to help enterprises develop their own exclusive big models.

There are 41 data sets and 226 sets of high-quality Prompt templates preset on Qian Fan platform, and 42 domestic and foreign mainstream large models are managed, providing Chinese enhancement, performance enhancement and context enhancement. For example, foreign big models such as Llama2, which used to have a good dialogue effect in English, are just as good in Chinese now.

As a result, Qian Fan platform has greatly lowered the threshold for enterprises to use, train and reason the big model. Enterprises can quickly optimize the model effect for their own business scenarios. In the past, it took customers at least a week to verify the effect of a large model. Now, based on a one-stop tool chain, it takes only one day to run through a model and see the effect.

By the beginning of September, there were nearly tens of thousands of monthly enterprises on the platform of Baidu AI Cloud Qian Fan, covering more than 400 scenes in manufacturing, energy, transportation and other industries.

In addition, nearly 50 immersive AI technology exhibitions were arranged at the venue of the Baidu World Congress in Shougang Park.

What makes me feel interesting is naturally the intelligent hardware that can interact on the spot. Here are a few examples:

[Add a girlfriend machine]

Drama-chasing artifact: a 27-inch super-large flat plate that can stand, sit and watch, and push and run at any time.

Fitness artifact: equipped with a small degree of AI fitness function, it has a wealth of somatosensory games and fitness courses, and can also score AI recognition actions.

[Qinghe learning mobile phone]

It integrates multiple functions, such as mobile phone, dictionary pen, telephone watch, translator, listening treasure, word card, etc., deeply integrates ERNIE Bot model, and provides full-link counseling such as interactive AI topic presentation, multi-disciplinary AI homework assistant AI oral practice, and AI precision.

I guess the school won’t let students take it even if they are killed, hahaha ~

[Small Smart Screen X9]

A new generation of small smart screen, equipped with a horizontal rotating pan/tilt, with a panoramic horizontal field of view of 360 and a super wide-angle field of view of 120; Support infrared night vision function, can automatically turn on/off infrared fill light according to the intensity of ambient light, and can realize 24-hour full-time and all-round home care.

Third, Baidu AI, pride of chinese

Artificial intelligence is an important driving force to lead the new round of global scientific and technological revolution and industrial transformation, and it is of great significance to promote social development.

Artificial intelligence can greatly improve production efficiency, help improve the quality and efficiency of social services and improve people’s lives.

In manufacturing, agriculture, service and other industries, we will create more value for the society by optimizing processes, accurate forecasting and automatic decision-making.

In addition, the development of artificial intelligence will also help promote technological innovation and industrial upgrading.

The application of artificial intelligence technology needs corresponding technical support, such as big data, cloud computing, Internet of Things, etc. The development of these technologies can in turn promote the progress of artificial intelligence technology, thus forming a virtuous circle and promoting technological innovation and industrial upgrading.

Finally, artificial intelligence can promote sustainable development.

Artificial intelligence can help human beings manage and utilize resources better, conduct production and life in an environmentally friendly way and achieve sustainable development.

At present, various countries have introduced relevant policies and measures to accelerate the development of artificial intelligence industry, trying to gain the upper hand in the global scientific and technological competition, and the world’s top enterprises are also racing against time.

The US government has put forward the National Artificial Intelligence Initiative, which plans to invest billions of dollars in artificial intelligence research and development and encourage private enterprises to participate. In addition, the United States has established a number of artificial intelligence research centers and alliances to strengthen the development of this field.

The British government has announced that it will invest hundreds of millions of pounds in the field of artificial intelligence and establish a number of artificial intelligence research institutes and centers to strengthen the development of this field.

The German government put forward the "Artificial Intelligence Innovation Plan", which aims to strengthen research and development and innovation in the field of artificial intelligence. The German government also plans to invest billions of euros to support innovative projects and start-ups in the field of artificial intelligence.

China government has listed artificial intelligence as a national strategy. China is also actively promoting international cooperation in the field of artificial intelligence, aiming to play a leading role in the global field of artificial intelligence.

And how will China’s artificial intelligence route develop?

Li Yanhong said, "The big model technology continues to evolve, and a large number of AI native applications will continue to emerge, eventually reshaping the industrial form and competitive landscape."

The big model and the AI native application, just like the operating system and App in the mobile Internet era, complement each other.

Baidu has made breakthroughs in these two aspects, taking the lead in making achievements, which not only opens up unlimited growth space for Baidu in the AI era, but also leads and enlightens China AI. At the moment when the global AI model is evolving iteratively against time, it has increased the core competitiveness for the AI industry in China.

As the leader of AI industry in China, the release of Wenxin Big Model 4.0 not only represents the stage peak of AI technology in China, but also reflects the pride of independent AI technology capability in China, and helps the new development of generative AI industry in China.

Author: Zhang Dongwei (senior Internet person, marketing expert, college student employment and entrepreneurship tutor)

This article first appeared on WeChat WeChat official account: Zhang Dongwei. The content of the article belongs to the author’s personal opinion and does not represent Hexun.com’s position. Investors should operate accordingly, at their own risk.

Reporting/feedback

The National Day box office broke through 2.3 billion, and ＂As Hard as a Rock＂ temporarily topped the list.


1905 movie network news According to the data of Lighthouse Professional Edition, as of 17: 22 on October 4, the box office list of new National Day movies in 2023 (including point screening and pre-sale) has reached 2.3 billion yuan. The competition is fierce, and it is temporarily in the top three in the rankings.


The director’s "Rock Solid" continued the box office growth momentum of the previous days, temporarily ranking first with 755 million yuan. The addition of many powerful actors such as,,,,,, adds a lot of highlights to this film. The infighting between Zhang Guoli and Yu Hewei in the play outlines various crises hidden behind the appearance of a prosperous city.


With the theme of the emotional state of contemporary urbanites, "Ex-4: Early Marriage" drives the box office with its high reputation. Meng Yun, who plays, tries a blind date, while Yu Fei, who plays, and his girlfriend create their own "marriage cooling-off period" to preview their marriage life in advance. The audience gained laughter and moved, but also had a deeper understanding of feelings. As of press time, "Ex-4: Early Marriage" has reached 585 million at the box office, temporarily ranking second.


The war giant "Volunteer Army: Attack by Heroes" has always attracted people’s attention with the highest score in the box office list of the new National Day movies in 2023. As of press time, the box office of the film has reached 423 million, temporarily ranking third. As the first part of the trilogy "Volunteers", the film tells the story from the decision to send troops to Chinese people’s Volunteer Army’s entry into the DPRK and the first and second battles in a more grand narrative way. It fills the blank of similar works in conception and perspective.


The box office adapted from real events continued to grow steadily, reaching 311 million, temporarily ranking fourth. The extreme confrontation between harmony and others has become a topic of great interest to the audience. Sports comedies temporarily ranked fifth, with a box office of 91.512 million as of press time.


This vaccine will be replenished in the near future! Once caught, you may be paralyzed for life →

The baby’s attention is at home! !

Polio vaccine! Answer after answer.

As of June 20th.

Inner Mongolia Autonomous Region to carry out relevant population

Replanting Activities of Inactivated Polio Vaccine (IPV)

First, IPV vaccine replanting began

In order to maintain the polio-free status in our region and effectively protect people’s health, according to the requirements of the Notice on Replanting the Second Dose of Inactivated Polio Vaccine for the Population jointly issued by the National Bureau for Disease Control and Prevention and the Implementation Plan for Replanting Inactivated Polio Vaccine for the Population in Inner Mongolia Autonomous Region jointly issued by the five departments of the Regional Bureau for Disease Control and Prevention, the inactivated polio vaccine (hereinafter referred to as inactivated polio vaccine) was comprehensively launched in Inner Mongolia from March 20 to June 30, 2024.

Second, the school-age population replanted with IPV vaccine

Children born between March 1, 2016 and September 30, 2019 who have only received one dose of inactivated polio vaccine are replanted with the second dose of inactivated polio vaccine; Children born in this age group who have not been vaccinated with inactivated polio vaccine have been supplemented with 2 doses of inactivated polio vaccine.

Third, why should we replant IPV vaccine?

Polio is an acute infectious disease caused by poliovirus infection, which mainly affects children under 5 years old. There are three serotypes of poliovirus, namely type I, type II and type III. Polio virus is mainly transmitted through oropharyngeal secretions. After infection, it can cause fever, limb pain, and even limb paralysis, respiratory muscle paralysis and other serious consequences.

Vaccination is an effective means to prevent polio. Before May 1st, 2016, the immunization program of polio vaccine used in our district was to vaccinate school-age children with trivalent attenuated polio vaccine (tOPV) for four times, which can prevent three types of polio virus infection at the same time. After May 1, 2016, according to the requirements, the immunization program of polio vaccine in Inner Mongolia Autonomous Region was adjusted to 1 dose of IPV+3 doses of bOPV, and after December 31, 2019, it was adjusted to 2 doses of IPV+2 doses of bOPV as required, which is still in use today. IPV vaccine is the abbreviation of inactivated polio vaccine, which can prevent three types of polio virus infection at the same time; BOPV vaccine is the abbreviation of bivalent live attenuated polio vaccine, which can prevent both type I and type III poliovirus infection, but has no cross-protection effect on type II poliovirus infection.

In recent years, polio outbreaks have occurred in many countries around the world, especially those caused by poliovirus derived from type II vaccine. Although China has been polio-free for many years, it still faces the risk of polio virus import and transmission. IPV replanting work is to improve the immune level of poliovirus type ⅱ of the population concerned, especially among children who have only been vaccinated with one dose of IPV vaccine or have not been vaccinated with IPV vaccine because the country has adjusted the immunization program of polio vaccine twice. Replanting can improve children’s immunity to poliovirus and reduce the risk of infection.

Iv. time and place of IPV replanting

IPV replanting time is determined by local conditions. Parents can pay attention to the notice of the local health department or the Center for Disease Control and Prevention to know the specific time and place of replanting. Generally speaking, IPV replanting work will be carried out in designated community health service centers, township hospitals and other inoculation units.

V. Precautions for replanting IPV

1. Parents should ensure that their children are in good health before taking their children for vaccination. If they have fever, diarrhea and other discomfort, they should suspend vaccination.

2. You should bring your child’s vaccination certificate when you are vaccinated, so that the staff can verify the vaccination information.

3. Observe for 30 minutes after inoculation to see if the child has allergic reaction or other discomfort. If there is any abnormality, you should inform the doctor in time.

4. Pay attention to keep the inoculation site clean and dry after inoculation to avoid infection.

At the beginning of this year, six departments, including the National Bureau of Disease Control and Prevention, jointly issued the Notice on Replanting the Second Dose of Inactivated Polio Vaccine for the Population. Children born between March 1, 2016 and September 30, 2019, who have only received one dose of inactivated polio vaccine, need a second dose of inactivated polio vaccine to strengthen the immune protection effect against poliovirus type II.

Source: Pentium Rong Media WeChat WeChat official account, Inner Mongolia CDC

Editor: Zhao Qing

Disclaimer: Except for the original content, the text, pictures, audio and video of the manuscript are all from the Internet and major media. The copyright belongs to the original author. If you think the content is infringing, please leave a message at the bottom of the article to contact us to delete or pay the manuscript fee.

Original title: "This vaccine has been quickly replenished in the near future! Once you are recruited, you may be paralyzed for life → "

Read the original text

Wangsu Technology： Investors are looking forward to boosting the stock price, and the company says it is concerned about the AI ​ ​ development adaptation plan.

Source: Ask Secretary Dong.

Investors ask questions:

Your company underperformed the Growth Enterprise Market index by 40 points a year, and you haven’t come up with measures to boost the stock price. Is it against the national policy of boosting the capital market? I hope the company can give a clear and reasonable explanation to the above problems, re-examine the current capital operation strategy and take measures that are more conducive to maintaining the confidence of investors in the secondary market. Today, with the vigorous development of AI infrastructure industry, NetScience should have seized the opportunity to maximize shareholder value through more scientific capital operation, instead of dampening investor confidence through a series of puzzling operations.

Secretary-General’s Answer (NetScience SZ300017):

Hello, thank you for your advice. The company will pay close attention to the changes brought by the development and application of new technologies to the industry and actively embrace the AI era. The company continues to invest in the construction of edge computing platform, and gradually upgrades CDN nodes to edge computing nodes with storage, calculation, transmission and security functions, and actively pays attention to the development of large models, agents and AI applications in the market, new industries and application scenarios, and pays attention to customers’ capital expenditure in this regard. According to the customer’s demand, adapt and polish the company’s technical solutions synchronously, and keep up with the pace of industry development. thank you

See more questions and answers from the Secretary-General > >

Disclaimer: This information is extracted from public information by Sina Finance and does not constitute any investment advice; Sina Finance does not guarantee the accuracy of the data, and the content is for reference only.

Reporting/feedback

The popularity of the unpopular plate reunited! More than 30 stocks hit a new high in the year, and the market prefers the defensive direction. These stocks have been funded.

Cailian News on May 13th (Editor Zilong),Today (May 13th), the public utilities sector strengthened again, and individual stocks were also active collectively. At the close, Gansu Energy, Xichang Power, Star Power, Hong Tong Gas and Leshan Power had daily limit, while Shouhua Gas rose by nearly 12%. Among them, Xichang Power won 4 boards in 6 days, and Star Power and Leshan Power both advanced to 2 boards.

The public utilities sector has warmed up, and many shares have reached a new high in the year.

For a long time, due to the slow growth of the industry and the lack of imaginable space, the public utility sector is relatively unpopular compared with various growth tracks and themes, and its related concept stocks are also unattractive to the market. However, with its relatively anti-pressure performance at the beginning of the year and its obvious performance in the near future, the popularity of public utility stocks has reunited again and has become the focus of investors’ attention in recent days. As of today’s close (May 13th), the utility sector has increased by nearly 11.3% during the year, ranking in the top 20% of the sector rankings.

In terms of specific stocks, many targets also hit new highs in the year today. As of today’s close, a total of 37 utility stocks have set an intraday high this year, among which Guodian Power, Datang Power Generation, Guiguan Power, Shenneng, hubei energy and other heavyweights are listed. In terms of performance during the year, Ganneng shares increased by nearly 64.6% this year, and eight shares of Xichang Power, hubei energy, Shenneng, Lantian Gas, Baoxin Energy, Star Power, Datang Power Generation and Huayin Power also increased by more than 30% during the year.

Among the new high stocks in these years, Xichang Electric Power was the most active, with four daily limit boards in the past six trading days, with a cumulative increase of nearly 56%. According to the data of Xichang Electric Power’s Dragon and Tiger List (3rd) today, its recent surge was attended by multi-channel hot money. Among the top five seats in the trading list, CITIC Jiantou Securities Beijing Guangqumennei Street, Guotai Junan Securities Shanghai Songjiang Zhongshan East Road and Yongxing Securities Cixi Xincheng Avenue North Road are listed. In addition, the headquarters of Guotai Junan Securities, a common seat for overseas institutions, is also located in the buying list.

The defensive sector continued to dominate, and these targets were funded to increase their positions.

Since the beginning of this year, the defensive sector has continued to dominate. Among the top industries in the year, household appliances, banking, coal, nonferrous metals, petrochemicals and public utilities all have defensive attributes, while the previously active track direction collectively showed a cooling trend. Boosted by this, defensive ETFs also rose sharply during the year. Among them, Bosera Fund’s leading home appliance ETF, Guangfa Fund’s construction machinery ETF and Guangfa Fund’s home appliance ETF ranked in the top three. In addition, resources, banks, central enterprises’ energy and energy ETFs also ranked in the forefront during the year.

From the perspective of funds, Public Offering of Fund, one of the main players in the market, also strengthened its defense direction in the first quarter. According to the statistics of the recent research report of Guojin Securities, regardless of the A-share market or the Hong Kong stock market, active partial-share funds have obviously reduced their positions (such as technology and medicine), especially small and medium-sized stocks, and increased their positions in the upstream and cycle (non-ferrous, public utilities, energy and raw materials), especially large-cap stocks, which is reflected in the period of increasing market volatility, institutions have obviously reduced their risk appetite and favored the defensive allocation of "large-cap value".

At present, the defense direction has become the focus of the market, and many stocks have also been actively funded recently. As of May 10th, the number of shares held by Colorful Chemical, Oriental Materials, Baodi Mining, Midland New Materials, SINOMACH Seiko, Huicheng Environmental Protection, Suneng, Cinda Securities, SINOMACH, and Ningbo Ocean Shipping in the northbound direction increased in the first place this month. In terms of financing funds, Yili, Changjiang Power, Jiangsu Bank, Chuanning Bio, Founder Securities, COSCO Haikong, China Ship, Lu Kang Medicine, China Petroleum, China Rare Earth, etc. will be paid close attention to during the month or by financiers.

(Cailian Zilong)
Reporting/feedback

Two cases of abnormal detection of COVID-19 nucleic acid were found in Zhongshan, Guangdong.

　　CCTV News:According to "Healthy Zhongshan" WeChat WeChat official account news, on October 16th, two cases of COVID-19 nucleic acid detection were found abnormal in Cuiheng New District (Nanlang Street), Zhongshan City, Guangdong Province. According to the needs of epidemic prevention and control, the relevant measures are hereby notified as follows:

　　1. Temporary traffic control will be implemented on all roads entering and leaving Ma ‘an Island (tentatively for three days), and bus service within Ma ‘an Island and across regions will be suspended. During the period of control, personnel can only enter and leave in principle. If it is really necessary to enter and leave because of emergency medical treatment for major diseases, specific official duties, production guarantee of key industrial chains, and supply guarantee of basic living materials, they must hold a negative nucleic acid test certificate within 24 hours, and enter and leave with a pass after the consent of the COVID-19 Epidemic Prevention and Control Headquarters of Cuiheng New District (Nanlang Street).

　　Second, delimit the control area

　　High-risk area: The construction sites (including living quarters) of Ma ‘an Island Dongwuwei Joint Inspection Building, Wanbin Shuangyue Building, Wanbin Wanchao Plaza, Wanbin Qianwan Garden, Convention and Exhibition Center and Wanbin Sales Department are classified as high-risk areas, and the management measures of "staying at home and serving at home" are strictly implemented.

　　Medium-risk area: other areas outside the high-risk area in Dongwuwei and Dongliuwei of Ma ‘an Island are classified as medium-risk areas, and the management measures of "people don’t leave the area and take things at wrong peaks" are strictly implemented.

　　Three, Ma ‘an Island to suspend meetings, forums, training, performances, sales promotion and other gathering activities. Strengthen the prevention and control of key places, chess and card rooms, mahjong halls, Internet cafes, foot baths, massage spas, beauty salons, gyms and other closed business places, stadiums, playgrounds, open-air swimming venues, tourist attractions and other venues are closed, schools, nursery institutions, offline training institutions and other offline services are suspended, and dining places are cancelled.

　　4. Ma ‘an Island will carry out regional nucleic acid detection from 18:00 to 23:00 on October 16; In Cuiheng New District (Nanlang Street), regional nucleic acid detection was carried out in other areas except Ma ‘an Island from 19:00 to 24:00 on October 16th, and there were 35 nucleic acid sampling points in the whole district (see attached table for details). Please actively cooperate with the masses to do a good job in nucleic acid testing, observe the on-site order of sampling points, keep a distance of one meter, do a good job in personal protection, and carry out nucleic acid testing at designated places in an orderly manner. If COVID-19 vaccine is inoculated before nucleic acid testing, it must be tested 48 hours later.

　　Fifth, gather less, don’t get together, and adhere to good living habits such as wearing masks, washing hands frequently, cleaning frequently, and using public chopsticks. If you have symptoms such as fever, dry cough, fatigue, sore throat, and decreased sense of smell, go to the fever clinic of Nanlang Hospital immediately, and wear a mask all the way to avoid taking public transportation.

　　The above prevention and control measures will be implemented from 18: 00 on October 16, 2022, and will be dynamically adjusted according to the needs of epidemic prevention and control.

The ＂two-in-one＂ operation process of the latest regulation of health code and travel code for Beijing epidemic entering and leaving Beijing is coming.

Recently, many people who left Beijing for business trips, tourism and other reasons failed to buy tickets when they returned to Beijing or found that the pop-up window of Health Bao showed that they could not confirm their health status for the time being. So what about the policy of returning to Beijing now? What should I do if I need to return to Beijing urgently? After leaving the high-risk area in Beijing, is it considered as a high-risk area to return to Beijing? How to operate the "two codes in one" of health code and travel code?

The "two-in-one" operation process of the latest regulation of health code and travel code for Beijing epidemic entering and leaving Beijing is coming.

1. Can you leave Beijing in the near future?

Recently, there have been many outbreaks in COVID-19 in China, involving 16 provinces, autonomous regions and municipalities directly under the Central Government, reminding the general public and friends not to leave the country or Beijing. Do not travel to high-risk areas and areas where confirmed cases are reported. High-risk groups, such as people in the closed control area and the control area, should be isolated at home and stay indoors, and carry out nucleic acid testing according to the requirements of relevant departments.

2. Can those who have already left Beijing return to Beijing?

Those who have not returned to Beijing on business or tourism should do a good job in personal protection such as traveling and dining, and pay close attention to the epidemic information in various regions of China. If there is time and space overlap with the published itinerary of reported cases in China, please report the itinerary to the community (village), unit and hotel immediately, isolate yourself locally and cooperate with various control measures. If you don’t return to Beijing, please stop returning to Beijing and cooperate with local prevention and control measures. On the evening of October 31, the Beijing Municipal Commission of Health and Wellness also issued an emergency reminder on epidemic prevention and control: on October 30, Shangrao City, Jiangxi Province, Heihe City, Heilongjiang Province and Hangzhou City, Zhejiang Province also newly reported local cases of COVID-19. Those who come to Beijing via Shangrao City, Jiangxi Province on October 11, Heihe City, Heilongjiang Province on October 12, Jiujiang City, Jiangxi Province on October 29, and Hangzhou City, Zhejiang Province since October 30 should immediately register with their units and communities of residence.

3. Will the health code change color when you come back from other places? What if the green code has a pop-up window?

If you visit other counties in the prefecture-level city where there is more than one local county (city, district or banner) infected with COVID-19 within 14 days, Beijing Health Bao will automatically compare your nucleic acid test records according to the relevant requirements of negative proof of nucleic acid test within 48 hours. If there is a negative record that meets the requirements of nucleic acid detection, the trip verification will get a green code.

The "two-in-one" operation process of the latest regulation of health code and travel code for Beijing epidemic entering and leaving Beijing is coming.

However, if the county (city, district, flag) has more than one case of local COVID-19 infected person’s living history within 14 days, Beijing Health Bao will give a pop-up window prompt, and the green code cannot be obtained. Please stop coming to Beijing according to epidemic prevention requirements.

4. What should I do if I can’t buy a ticket for the health treasure green code?

China Railway 12306 responded that if the health code green code still can’t buy tickets, it may be a big data screening error. Passengers can register their ID numbers, so that the background can confirm and verify them again.

Under the influence of the epidemic, we often use health codes and communication travel codes to enter and leave some places. But when both yards need to be shown, it is necessary to switch pages back and forth, which makes many people worry. In order to make it more convenient for everyone to travel, the national government service platform has realized the integration of health code and travel code, and achieved a nationwide one-page access. Now, as long as you open Alipay and set it up easily, you can greatly improve your travel efficiency!

Method one

Step 1: Open Alipay, search the national government service platform on the homepage, and click Enter.

Step 2: Click the epidemic prevention health information code service after entering the page.

Step 3: Enter this page. If you have not received it, click Received; If you have received it, click to view it now.

Step 4: After entering, it is a health code, and the travel code needs to be verified by clicking, and the mobile phone number and verification code are entered.

Step 5: Generate a green code, with the upper QR code as the health code and the lower one as the travel card, so as to realize the integration of the two codes.

The "two-in-one" operation process of the latest regulation of health code and travel code for Beijing epidemic entering and leaving Beijing is coming.

Method 2

Step 1: Open Alipay and search the health code on the home page.

Step 2: After clicking Enter, click the National Platform Health Code on the right.

Step 3: If it is not received, click Received; If you have received it, click to view it now.

Step 4: After entering, it is a health code, and the travel code needs to be verified by clicking, and the mobile phone number and verification code are entered.

Step 5: Generate a green code, with the upper QR code as the health code and the lower one as the travel card, so as to realize the integration of the two codes.