Too long to look at the version
Huawei Pangu Grand Model 3.0 was released on July 7
A cross-strait seminar on digital transformation and metaverse development was held in Nanjing
Peking University Law model ChatLaw released
NetEase AI big model accelerates the application and will provide 100,000 people with new AI occupations this year
Meta has launched its benchmark Twitter app Threads, and users are not limited in their reading volume
Israel has initially passed a crypto tax bill that will exempt foreign residents from cryptocurrency capital gains tax
Japan's Ministry of Internal Affairs and Communications is soliciting public opinions on "the use of metaverse technology in the Web3 era"
Gao Wen, academician of the Chinese Academy of Engineering: will build a 200 billion parameter natural language large model base, and the performance will be benchmarked against ChatGPT
Baidu Wang Haifeng: Wenxin 3.5 improves the effect by 50% and the reasoning speed is increased by 30 times
Chairman Grammy: Music that incorporates elements of AI creation is fully eligible to participate and compete for nominations
Zhejiang: Promote the chain, scale and internationalization of the metaverse industry in an all-round way
AI digital identity ecosystem Aspecta completed a new round of strategic financing, with Spartan Group participating
Decentralized data storage startup Hello completed a €170,000 pre-seed round of financing, with participation from Casti Ventures and others
Metaverse headlines
Huawei Pangu Model 3.0 was released on July 7
According to the first financial report, on July 6, at the 2023 World Artificial Intelligence Conference, Ken Hu, Huawei's rotating chairman, delivered a speech revealing that Pangu Model 3.0 will be released soon. Huawei's Pangu CHAT multi-modal 100-billion-level large-model products will be displayed and tested at the HUAWEI CLOUD Developer Conference, mainly for ToB/G government and enterprise customers.
Ken Hu said that the emergence of ChatGPT at the end of last year has pushed artificial intelligence to a new outlet, artificial intelligence will help rewrite everything around them, Huawei's core consideration now is to fully promote artificial intelligence in the next stage.
It is understood that Huawei's Pangu series of basic large models were released in 2021, including NLP, CV, and scientific computing large models, and subsequently released large models for mining, drug molecules, meteorology, and ocean waves.
Among them, the Pangu NLP large model uses the Encoder-Decoder architecture for the first time, which takes into account the understanding and generation capabilities of the NLP large model, and ensures the embedding flexibility of the model in different systems.
In downstream applications, only a small number of samples and learnable parameters are required to complete the rapid fine-tuning and downstream adaptation of hundreds of billions of large-scale models.
The Pangu CV large model is the industry's largest CV large model that realizes on-demand model extraction for the first time, taking into account the discrimination and generation capabilities, and can adaptively extract models of different scales based on the model size and running speed requirements, and AI application development can be quickly implemented.
At the same time, Huawei announced that it will work with 26 enterprises and research institutes to innovate the application of basic and industry large models based on Ascend AI, and jointly release the Ascend AI large model training and promotion integrated solution with partners to accelerate the application of large models in various industries, and 23 Ascend AI partners will launch new AI servers, intelligent edges, and terminals.
Domestic dynamics
A cross-strait seminar on digital transformation and metaverse development was held in Nanjing
According to Xinhuanet, on July 5, a cross-strait seminar on digital transformation and metaverse development was held in Nanjing, Jiangsu Province. With the theme of "Digital Transformation and Metaverse Exploration", this seminar aims to build a cross-strait metaverse industry exchange and cooperation platform, promote technological innovation, application expansion, ecological construction and industrial governance in the metaverse field, and jointly promote the integrated development of the cross-strait metaverse industry.
As an integration of cutting-edge digital technology and an important platform for interaction between the virtual world and real society, the metaverse is becoming a new form and track for the development of the digital economy, showing broad development prospects.
Industries on both sides of the strait have given full play to their respective advantages, strengthened exchanges and cooperation in expanding application scenarios, exploring business models, and promoting talent co-education, and accelerated the innovative application of the metaverse and the construction of a digital ecosystem. Pay attention to the security challenges brought by the metaverse industry and effectively ensure the security of development.
Northeastern University research team released TechGPT, a large language model
According to CaiLian News Agency on July 6, recently, the scientific research team led by Associate Professor Ren Feiliang of the School of Computer Science and Engineering of Northeastern University released TechGPT, a large-language model that was fully fine-tuned and open sourced.
Compared with other types of large language models, TechGPT mainly strengthens the analysis and processing capabilities of natural language texts in more than ten vertical professional fields such as computers, materials, machinery, and metallurgy.
At present, TechGPT supports single-round dialogue and multi-round dialogue, covering logical reasoning Q&A, suggestion consultation Q&A, text comprehension Q&A, Chinese and English version translation, information extraction and other natural language understanding and text generation capabilities.
Peking University Law model ChatLaw released
According to the Science and Technology Innovation Daily, ChatLaw was developed by the Shenzhen Graduate School of Peking University and the Rabbit Exhibition Intelligent AIGC Joint Laboratory. Its data is mainly composed of forums, news, legal articles, judicial interpretations, legal consultations, legal examination questions, judgment documents, and through cleaning, data enhancement, etc. to construct dialogue data.
At present, Chatlaw has open sourced 3 models: ChatLaw-13B, ChatLaw-33B, and ChatLaw-TextVec.
NetEase AI big model accelerated the application and will provide 100,000 people with new AI occupations this year
According to China Securities Network, at the 2023 World Artificial Intelligence Conference, NetEase's many artificial intelligence large model landing products made their debut, focusing on the leading innovation achievements in the artificial intelligence infrastructure layer, engine layer, platform layer, model layer and application layer.
It is reported that since the beginning of this year, NetEase has adhered to pragmatism and rapidly promoted the landing of AI large models in key industrial scenarios such as games, education, industry, and music. It is understood that the NetEase AI large model is accelerating to cover more than 100 industrial application scenarios.
While accelerating innovative applications, NetEase Fuxi Youling crowdsourcing platform will also provide 100,000 people with new AI occupations this year, including excavator remote drivers, AI painters, AI expression binding artists and other human-machine collaboration jobs.
Overseas news
Meta has launched its benchmark Twitter app Threads, which users can read without restrictions
According to BlockBeats, on July 6, Facebook's parent company Meta has officially launched its new application Threads, which uses Mastodon's decentralized social media protocol ActivityPub integration, users can log in through Instagram accounts, focusing on Musk's Twitter.
In order to limit Twitter crawling, Musk imposed a read limit on Twitter users last week: verified accounts can read up to 10,000 tweets per day; 1000 unverified accounts; 500 new unverified accounts, which caused user dissatisfaction. At the end of June, Musk had an online agreement with Meta founder Zuckerberg to engage in offline MMA fighting, which has not yet been carried out.
Israel has initially passed a crypto tax bill that will exempt foreign residents from cryptocurrency capital gains tax
According to CoinDesk, the Knesset on Wednesday preliminarily passed a bill that would exempt foreign residents from capital gains tax on the sale of digital currencies and reduce the tax on crypto options (similar to stock options) for employees from 50 percent to about 25 percent. The bill is supported by a coalition government led by Israeli Prime Minister Benjamin Netanyahu.
Last November, Israel's Ministry of Finance issued a series of recommendations for the regulation of digital assets, calling for the creation of a new regulatory infrastructure to regulate the issuance of collateralized digital assets, including stablecoins, and the financial services provided through them, and to pass legislation to give the Bank of Israel the authority to regulate digital assets "with significant stability or monetary impact."
At the same time, the guide also proposes to tax overseas cryptocurrencies through the Bank of Israel.
Japan's Ministry of Internal Affairs and Communications is soliciting public opinions on "the use of metaverse technology in the Web3 era"
According to CoinPost, Japan's Ministry of Internal Affairs and Communications is currently soliciting public comments on the draft report summarized by the "Research Council on the Use of Metaverse and Other Technologies for the Web3 Era". The purpose of the seminar is to sort out the issues related to information and communication management in the application of the metaverse in the Web3 era. The public consultation period runs from 17 June to 9 July and the final report will be prepared based on the results.
In addition, the Ministry of Internal Affairs and Communications and other government agencies will implement necessary measures to improve the convenience of Japanese metaverse users and expand use cases for virtual space in response to the policy issues outlined in this report. It is reported that a number of companies in Japan have announced their entry into the "meta-universe".
Metaverse perspective
Gao Wen, academician of the Chinese Academy of Engineering: will build a 200 billion parameter natural language large model base, and the performance will be benchmarked against ChatGPT
According to IT House reported on July 6, Gao Wen, academician of the Chinese Academy of Engineering and director of Pengcheng Lab, said at the Ascend Artificial Intelligence Industry Summit Forum at WAIC2023 that Pengcheng Lab has launched the Mental Big Model Plan, aiming to build the first completely independent innovation, open source and open natural language pre-training large model base in China, with parameter levels reaching 200 billion and performance benchmarking ChatGPT.
Baidu Wang Haifeng: Wenxin 3.5 improves the effect by 50% and the reasoning speed is increased by 30 times
According to the Securities Times, on July 6, at the 2023 World Artificial Intelligence Conference Industry Forum, Baidu Chief Technology Officer Wang Haifeng said that the effect, function and performance of Wenxin Model 3.5 have been comprehensively improved, and the basic model upgrade, fine-tuning technological innovation, knowledge point enhancement, logical reasoning enhancement, etc. have been improved, and the model effect has been increased by 50%, the training speed has been increased by 2 times, and the reasoning speed has been increased by 30 times.
Chairman Grammy: Music that incorporates elements of AI creation is fully eligible to participate and compete for nominations
According to the Webmaster House, recently, Harvey Mason Jr. He said that as the music industry continues to adapt to the new technology of artificial intelligence, the Grammys will adjust accordingly. In short, AI or music that incorporates elements of AI creation is fully eligible to participate and compete for a Grammy nomination.
"As long as human beings have made more than a minimum contribution, which is meaningful to us, they can always be considered for nomination or award," he said. "Of course, they don't want to see technology replace human creativity, and they want to make sure that technology enhances, glorifies or complements human creativity." He also said that this year's results can verify whether such works will be nominated.
Institutional perspective
Zhejiang: Promote the chain, scale and internationalization of the metaverse industry in an all-round way
According to Zero One Finance reports, recently, Zhejiang held a provincial conference on the high-quality development of the platform economy and issued the "Implementation Opinions on Promoting the High-quality Development of the Platform Economy".
The opinions mentioned that innovation in future scenarios of the platform economy should be encouraged. Platform enterprises are encouraged to use innovative technologies such as blockchain, digital twin, and extended reality to create future-oriented diversified application scenarios. Accelerate the construction of new advantages in the future industry of the metaverse, support diversified entities to build a comprehensive test platform for the metaverse, strengthen the application of the metaverse in multiple scenarios, and comprehensively promote the chain, scale, and internationalization of the metaverse industry.
Financing news
AI digital identity ecosystem Aspecta completed a new round of strategic financing, with Spartan Group participating
According to BlockBeats news, on July 6, Aspecta, an AI-driven digital identity ecosystem focused on developers, announced the completion of a new round of strategic financing, the specific amount of which has not been disclosed, and this round of financing is invested by Spartan Group.
The new funding will focus on improving Aspecta ID, Aspecta's core application, and expanding the developer ecosystem infrastructure globally. The team's goal is to establish a new paradigm of data ownership, interoperability, and user experience using artificial intelligence and blockchain technology.
Decentralized data storage startup Hello completed a €170,000 pre-seed round of financing, with participation from Casti Ventures and others
According to BlockBeats, on July 6, according to EU-Startups, decentralized data storage startup Hello announced the completion of a 170,000 euro pre-seed round of financing, with the participation of Esade BAN, Casti Ventures and Bcombinator, as well as several European angel investors, and the new financing will be used for strategy execution such as product development.
Hello is an early decentralized Web3 ecosystem that includes applications, networks, and infrastructure based on blockchain technology, decentralized storage, and distributed databases. Hello aims to provide a sustainable, scalable, efficient, and user-controlled approach to data storage that addresses the limitations of traditional systems.