Babbitt | Metaverse Daily Must-Read: Nvidia Wins Ma, Talal Future Releases Hundreds of Billion-Level Mathematical Models

1c1218a624b62d4fefee35eb8ea485d4.gif

Abstract: No one doubts that Nvidia is the biggest winner of this wave of AIGC boom, not one of them. However, beyond people's expectations, Nvidia can make such money. The net profit rose 8 times, and announced the repurchase of 25 billion US dollars of stock at the high level of the stock price. What is the pattern, this is the pattern.

052ff9a490abd7a0a134f5af790f1e35.png

Image source: Generated by Unbounded AI

Hot info

NAVER, South Korea's largest web portal, releases a large language model for artificial intelligence

According to Jiemian News citing Yonhap News Agency, South Korea’s largest portal website NAVER released the Korean artificial intelligence large language model “HyperCLOVA X” on August 24. HyperCLOVA X is an upgraded version of HyperCLOVA that will be launched in 2021. It is a large language model tailored for Korean. Choi Soo-yeon, CEO of NAVER, said that various artificial intelligence-based technologies will be incorporated into services such as search, shopping, appointment, user creation, maps, and videos.

TAL's self-developed 100-billion-level mathematical model MathGPT is online, open for public testing from now on

According to a DoNews report on August 24, during the 20th anniversary live broadcast event, TAL (Xueersi) CTO Tian Mi announced that MathGPT, a 100-billion-level large-scale model in the mathematics field developed by TAL, was officially launched and opened for public testing. From now on, users can apply for a free trial experience through the official website to register an account.

CNKI releases AI intelligent writing platform and other "big model + AIGC" products

According to news from the Webmaster’s Home on August 24, CNKI recently officially released a series of big data knowledge management products based on "big model + AIGC", including a big data governance platform, digital standard intelligent application platform, and HowNet intelligent writing platform , China's economic and social big data research platform. Among them, HowNet AI Intelligent Writing is an intelligent document production system developed based on large models and knowledge bases using AIGC technology. AI can generate a report outline based on the document name and subject, and generate all chapters of the document with one click according to the outline AI, visually displaying the entire process of document generation.

Du Xiaoman Xuanyuan Large Model Version 2.0 will be released in October

According to the "Kechuangban Daily" report on August 24, version 2.0 of Du Xiaoman's Xuanyuan model will be released in October. It is understood that the 2.0 version has improved capabilities in scenarios such as dialogue and financial understanding. In the incremental pre-training and instruction fine-tuning stages, this version has added massive financial data; in terms of dialogue capabilities, the context length of "Xuanyuan 2.0" has been expanded to 8k, which can handle longer financial reports, research and analysis. In May of this year, Du Xiaoman open-sourced the first 100-billion-level Chinese financial model "Xuanyuan". Since then, hundreds of financial institutions have applied for trial use.

The total size of Nvidia's procurement contracts reached US$11.15 billion, an increase of 53% from the previous quarter

According to Taiwan's Economic Daily reported by "Kechuangban Daily", as the sales of Nvidia's AI chips surged, the market was originally worried that the production of the supply chain could not keep up. In this regard, Colette Kress, Nvidia's chief financial officer, pointed out in the latest earnings conference that supply is increasing quarter by quarter and will continue all the way to next year. Nvidia's purchase commitments and other contracts totaled $11.15 billion, a 53% jump from the previous quarter's $7.3 billion. She said the increase showed companies spending more on long-term supplies. The supply is expected to increase every season until next year.

Samsung may disclose details of its own ChatGPT-like generative AI in September: code optimization, email writing

According to IT House citing Korean media Korea Daily, Samsung plans to hold the Real Summit 2023 event on September 12, and will launch its own version of generative AI. During the event, Samsung will disclose information about generative AI to the public, the media, and Samsung customers, with many skills similar to ChatGPT.

DIGITIMES Research: Shipments of high-end artificial intelligence servers are expected to reach 337,000 units in 2024

According to the "Kechuang Board Daily" report on August 24, according to DIGITIMES Research, shipments of high-end artificial intelligence servers are expected to reach 337,000 units in 2024, and the generative artificial intelligence market will reach 109 billion by 2030 USD, with a compound annual growth rate (CAGR) of 35%. According to Eric Huang, vice president of DIGITIMES, the large amount of data generated by the cloud and hardware has greatly increased the amount of data transmitted, which opens up opportunities not only for artificial intelligence, but also for the entire ICT industry. The next step is to enter the era of edge AI.

Investment and Financing

Sophon Engine, a multi-modal large-scale model company, completed an angel+ round of financing worth tens of millions of yuan

According to Jiemian News on August 24, recently, the multi-modal large-scale model company "Sophon Engine" completed an angel + round of financing. This round was jointly led by a fund under CICC Capital and a leading investment institution. With the investment, Jinghe Capital acted as the exclusive financial advisor.

"Harmony Juli" has completed the C1 round of financing of tens of millions of yuan, and will focus on the application of generative AI technology

According to 36 Krypton’s report on August 24, “Heqi Juli” has completed the C1 round of financing of tens of millions of yuan, invested by Dening Capital and Kaola Fund. This round of financing is mainly used for product development and market development. The company will invest heavily in the low-code aPaaS platform and the application of generative AI technology, and continue to expand its business and deploy in more provinces. According to reports, Heqijuli was established in 2013, focusing on digital technology, and providing basic education with software applications and comprehensive solutions covering the three major scenarios of educational affairs, teaching, and governance.

Metaverse recommended reading

"Artificial intelligence has become a powerful "tourist partner"? May still need to improve in many aspects"

August is the peak season for tourism. Whether AIGC can make tourism planning and services more convenient and intelligent deserves our attention. The People’s Data Research Institute focuses on the four general AIGC platforms of Wenxin Yiyan, Xunfei Xinghuo, ChatGPT and 360 Zhinao[1], and evaluates the answer quality of large language models in the field of tourism from three dimensions: target selection, itinerary planning, and explanation services , to discover the ability and possibility of improvement for tourists to use the AIGC platform to solve travel problems and obtain services conveniently.

https://www.8btc.com/article/6830670

"AI Large Model Value Alignment: What, Why, and How?" "

As large models (also known as base models) begin to perform a wide range of language understanding and content generation tasks like humans, people need to face a most fundamental and scientifically challenging problem: how to make the capabilities and behavior of large models comparable to those of humans. Values, true intentions, and ethical principles are consistent to ensure safety and trust in the collaboration between humans and artificial intelligence. This problem is known as "value alignment".

https://www.8btc.com/article/6830668‌

"Profits have soared 8 times, and "AI FOMO" has made Nvidia profitable"

Everyone is predicting that Nvidia will make a lot of money because of this wave of AI, but no one can imagine that this AI infrastructure company will make such an exaggerated profit. Nvidia's processors A100 and H100 dedicated to AI computing power have sold like crazy, so that revenue has soared by 88% compared to the same period last year when AI had not yet "resurrected". Nvidia's net profit in the second quarter was directly bursting. The company's net profit in the same period last year was only 656 million US dollars, but in the second quarter of this year, this figure has directly increased to 6.19 billion US dollars, an increase of more than 8 times.

https://www.8btc.com/article/6830684‌

"Hundreds of billions, dedicated to mathematics, the MathGPT large model has begun public beta"

MathGPT brings together TAL's years of education, teaching and research data accumulation, focusing on the field of mathematics. Among the test results of six public mathematics evaluation collections including CEval-Math, AGIEval-Math, APE5K, CMMLU-Math, Gaokao Mathematics and Math401, MathGPT achieved the highest scores in multiple tests. At the same time, MathGPT also performed well on the general test set of C-Eval's middle and high schools.

https://www.8btc.com/article/6830696‌

Babbitt Park is open for cooperation!

ac43087042309f26a9efafcffd550951.png

67bd69d98cd2146d1603308b17e4e5ea.jpeg

e495cd87c6b7c6d07acf8319206b55e1.gif

Chinese Twitter: https://twitter.com/8BTC_OFFICIAL

English Twitter: https://twitter.com/btcinchina

Discord community: https://discord.gg/defidao

Telegram channel: https://t.me/Mute_8btc

Telegram community: https://t.me/news_8btc

e48ff4d3a9c37c634f65e46313a42758.jpeg

Guess you like

Origin blog.csdn.net/weixin_44383880/article/details/132485249