A quick look at news points in one minute!
1. Tang Daosheng from Tencent: Large-scale industry model is a better path for enterprises to embrace AI
2. WPSAI is officially named, and the official website is launched simultaneously
3. Hu Guoping of HKUST Xunfei: On October 24, the general large-scale model was released, fully benchmarked against ChatGPT, and fully surpassed in Chinese
4. China Mobile will release the "Nine Heavens" 1+N large model, led by Feng Junlan, Chief Scientist of China Mobile Research Institute
5. GPT-4API is fully open for use!
6. Microsoft denies that 30 million customer account credentials were stolen
7. Lithography machine giant ASML: No special edition lithography machine has been launched to the Chinese market
8. Yahoo, the "living fossil" of the Internet, is about to go public again
9. Microsoft could make at least $90 billion from ChatGPT
10. ChatGPT traffic declines for the first time, analysts: the novelty is gradually disappearing
11. Bunv0.6.13 released
12. Shanghai AI Lab released the open source model InternLM-7B
domestic news
1. Tang Daosheng from Tencent: Large-scale industry model is a better path for enterprises to embrace AI
At the 2023 World Artificial Intelligence Conference Industry Forum, Tang Daosheng, Senior Executive Vice President of Tencent and CEO of Tencent Cloud and Smart Industry Business Group, said that the general large model has strong capabilities, but it cannot solve the specific problems of many enterprises. Enterprise large-scale model applications need to comprehensively consider factors such as industry expertise, data security, continuous iteration, and comprehensive costs. Based on the large industry model, building your own exclusive model may be a better option for enterprises.
2. WPSAI is officially named, and the official website is launched simultaneously
On July 6, Kingsoft Office brought its large language model-based intelligent office assistant WPSAI to the 2023 World Artificial Intelligence Conference. WPSAI's official website was launched simultaneously, and a channel for recruiting intelligent office experience officers was opened. WPSOffice individual users and WPS365 enterprise users can apply . Zhang Qingyuan, CEO of Kingsoft Office, said that Kingsoft Office will position WPSAI as the application side of the large language model, and will anchor AIGC (content creation), Copilot (smart assistant), and Insight (knowledge insight) in the future. (36 krypton)
3. Hu Guoping of HKUST Xunfei: On October 24, the general large-scale model was released, fully benchmarked against ChatGPT, and fully surpassed in Chinese
On July 6, at the 2023 World Artificial Intelligence Conference, Hu Guoping, Senior Vice President of iFLYTEK and director of the National Key Laboratory of Cognitive Intelligence, revealed at the Ascend Artificial Intelligence Industry Summit Forum that on October 24, iFLYTEK will release the GM AI. Model, fully benchmarked against ChatGPT, and fully surpassed in Chinese. (The Paper)
China Mobile will release the "Nine Heavens" 1+N large model, led by Feng Junlan, Chief Scientist of China Mobile Research Institute
According to media reports, China Mobile will release the "Nine Days" 1+N large-scale model in the near future, which is aimed at two major application scenarios of government affairs and customer service. The person in charge of China Mobile's large-scale model technology and product research and development is Feng Junlan, chief scientist of China Mobile Research Institute and general manager of Artificial Intelligence and Smart Operation Center. (Pin Wan)
international news
1. GPT-4API is fully open for use!
On July 7, OpenAI announced on its official website that the GPT-4 API is fully open for use. Now all paid API users can directly access GPT-4 in 8K context without any waiting. It is expected that by the end of July, OpenAI will open the GPT-4 API to new developers. (API detailed instruction address: https://platform.openai.com/docs/guides/gpt) In addition, OpenAI also launched GPT-3.5Turbo, DALL·E and WhisperAPI.
2. Microsoft denies that 30 million customer account credentials were stolen
Hacker group AnonymousSudan claims to have breached Microsoft servers and stolen a database containing credentials for more than 30 million customer accounts. Microsoft has denied this claim. AnonymousSudan is known for launching DDoS attacks, and it is believed to be associated with organizations such as the pro-Russian Killnet. Microsoft acknowledged last month that its services such as Azure, Outlook and OneDrive were temporarily down due to a DDoS attack by AnonymousSudan. AnonymousSudan is selling a stolen Microsoft customer database for $50,000, but a Microsoft spokesman says there is no evidence its customer data was compromised.
3. Lithography machine giant ASML: No special edition lithography machine has been launched to the Chinese market
In response to the rumor that "a special version of the DUV lithography machine will be launched in China", the Dutch lithography machine giant ASML responded, "ASML has always complied with the applicable laws and regulations, and we have not launched a special version of the lithography machine for the Chinese market. machine". A week ago, the Dutch government officially announced new regulations restricting the export of certain advanced semiconductor equipment, which will take effect on September 1. (Observer Network)
4. Yahoo, the "living fossil" of the Internet, is about to go public again
Yahoo CEO Jim Lanzon said recently that Yahoo plans to go public again and is already financially prepared with a good balance sheet and profitability. Yahoo currently has more than 30 business units, including finance, sports, news, mail and TechCrunch and other sites. Jim Lanzon also said that Yahoo will actively consider seeking merger and acquisition opportunities in related industries that have not yet established a business, and artificial intelligence will also provide a new opportunity for all businesses. (later Latepost)
Microsoft could make at least $90 billion from ChatGPT
On July 6, Morgan Stanley released a research report, raising Microsoft's target price from $335 per share to $415 per share. Several preliminary trend lines point to Microsoft's strongest position in enterprise software that could benefit from generative artificial intelligence, analysts said. They also expect (artificial intelligence) to represent an incremental opportunity for the company of up to $90 billion by fiscal year 2025 at minimum penetration. As such, Morgan Stanley believes that given Microsoft's strong position in generative AI, a business model that can quickly respond to changing demand (especially Azure's consumption model), and a still reasonable GAAP P/E ratio, it can further expand profits rate potential, Microsoft is the stock of choice among large software companies. (Futu Niu Niu)
5. ChatGPT traffic declines for the first time, analysts: the novelty is gradually disappearing
According to data from the analysis company Similarweb, ChatGPT, an artificial intelligence chat robot that became popular around the world after its release in November last year, saw its first monthly decline in website traffic and unique user visits in June this year. David Carr, Senior Insights Manager at Similarweb, said the traffic changes indicated that the novelty of the chatbot was fading. Rishi Jaluria, an analyst at RBC Capital Markets, explained that these data point to an increased demand for the real-time information generation capabilities of generative AI. (Sina Technology)
Programmer Zone
Bunv0.6.13 released
Bun, a modern JavaScript runtime similar to Node and Deno, has released the latest Bunv0.6.13 version, which has made several changes to Bun, such as the grand launch of the JavaScript bundler bunbuild, performance improvements, and 20% faster JSON .parse, Proxy with up to 2 times faster speed, and arguments, etc. For details, please refer to the official release notes: https://bun.sh/blog/bun-v0.6.13
Shanghai AI Lab released the open source model InternLM-7B
Shanghai AI Lab released the open source model InternLM-7B, the Chinese name Shusheng·Puyu Large Model, which contains 7 billion parameters. According to the developer, “InternLM is a multilingual 100 billion parameter pedestal model trained on over trillions of token data. Through multi-stage progressive training, the InternLM pedestal model has a relatively high level of knowledge, and is capable of reading comprehension in Chinese and English, Reasoning tasks and other scenes that require strong thinking ability have excellent performance, and have outstanding performance in a variety of comprehensive tests designed for humans. On this basis, through high-quality human-labeled dialogue data combined with RLHF and other technologies, InternLM can be used in Respond to complex instructions when talking to humans, and show responses that conform to human ethics and values." The source code is open source under the Apache-2.0 agreement, and the weight is completely open to academic research. Commercial use is also allowed after obtaining official written permission. (Solidot)
Original link: https://blog.csdn.net/csdngeeknews/article/details/131589952