Babbitt | Metaverse daily must-read: Tencent will announce large models and AI chat robots; Alimama releases two new AI products...

56581bbf1ac2211e12a385e99495ab38.gif

Abstract: Tencent will officially announce its large model and AI chat robot tomorrow. However, the day before the release, WeChat has already revealed the relevant naming and icons. WeChat public accounts and mini programs show that Tencent’s large AI model is named “Tencent Hunyuan Large Model” and the multi-modal dialogue robot is named “Tencent Hunyuan Assistant”.

5b2730d0908f2bdcba5cdd81418bdc11.png

Image source: Generated by Unbounded AI

Hot info

Alimama releases two new AI delivery products, "Wanxiangtai Unbounded Edition" and "Alimama Bailing"

On September 6, Alimama released two new AI advertising products at the 2023m Summit: the one-stop brand management platform "Alimama Bailing" and the one-stop marketing delivery system "Wanxiangtai Unbounded Edition" to create business products for merchants. "Twin engines" of growth that combine efficiency and effectiveness.

Wang Xiaochuan: The total number of downloads in the community has exceeded 5 million since the Baichuan model was open sourced

On September 6, Baichuan Intelligence, a large model company founded by Wang Xiaochuan, held a press conference to introduce the latest progress of Baichuan Large Model in open source. Wang Xiaochuan said that the two large open source models Baichuan-7B and Baichuan-13B have ranked among the best in many authoritative evaluation lists and have been downloaded more than 5 million times. He believes that Baichuan Intelligence, as an entrepreneurial team with experience in search, knows how to select and screen high-quality vertical industry data based on trillions of Internet data, and conduct duplication and anti-spam training. Therefore, Baichuan's ability in training corpus helps After being open sourced, Baichuan-7B's performance on mainstream tasks in Chinese and English surpassed LlaMA2-13B.

Lingdian Youshu released the "Lingdian Model" large model, which has five major functions including smart business analysis

It was reported on September 6 that Lingdian Youshu (301169.SZ) today released the "Lingdian Model" large model, which has five major functions: intelligent consultation and question answering, intelligent summary extraction, intelligent operation assistant, intelligent strategy assistant, and intelligent business analysis. "Zero Point Model" covers Zhizheng Model and Shushang Model. The former mainly serves government customers and the latter mainly serves commercial customers. The two share technical foundations to a large extent, but each has its own characteristics. In addition, "Zero Point Model" is a flexible large model that is scalable, upgradable, and portable. It has four technical features: open source model base transformation, multi-algorithm technology integration, inference speed optimization mechanism, and reduced computing power configuration environment requirements.

Mourui Technology and Muxi jointly released the first AI model "texture super-resolution" technology

It was reported on September 6 that recently, Changsha MoRui Network Technology Co., Ltd. (hereinafter referred to as "MoRui Technology") and Muxi Integrated Circuit (Shanghai) Co., Ltd. (hereinafter referred to as "Muxi") jointly released the first AI model "texture" "Super score" technology.

According to Mu Xi, this technology relies on the computing power of XiSilicon N100, the first product of XiSilicon's N-series AI inference GPU, and combined with the algorithm library resources and AI super-resolution algorithm of MoRui Technology, it has successfully applied AI super-resolution to the field of 3D models for the first time. , with the strength to create domestic software and hardware integrated solutions that are independent, controllable, safe and trustworthy.

WeChat launches the "Tencent Hunyuan Assistant" applet: supports AI Q&A/painting, and the icon resembles the Tai Chi Yin and Yang pattern

It was reported on September 6 that Tencent will officially announce its large model and AI chat robot tomorrow. However, the day before the release, WeChat has already revealed the relevant naming and icons. WeChat public accounts and mini programs show that Tencent’s large AI model is named “Tencent Hunyuan Large Model” and the multi-modal dialogue robot is named “Tencent Hunyuan Assistant”. Among them, the icon of “Tencent Hunyuan Assistant” is quite interesting, and the overall design is similar to the traditional Chinese Tai Chi yin and yang pattern. According to the mini program information, the functions of “Tencent Hunyuan Assistant” include AI Q&A and AI painting. Tencent officials have previously demonstrated the ability to write copywriting. After testing, it was found that the "Tencent Hunyuan Assistant" applet requires invited users to open it, and large-scale testing is not currently available.

Alibaba Cloud launches large video generation model I2VGen-XL

According to news on September 6, Alibaba Cloud recently launched a new large-scale video generation model I2VGen-XL. The person in charge of R&D of the model said that the model can be applied to short video content production, movie production and other scenarios.

Shanghai Jiao Tong University releases CodeApex, a large-model bilingual programming assessment benchmark

The APEX Laboratory of Shanghai Jiao Tong University launched CodeApex, a bilingual benchmark data set focused on assessing the programming understanding and code generation abilities of LLMs.

To assess the programming understanding of large language models, CodeApex has designed three types of multiple-choice questions: conceptual understanding, common sense reasoning, and multi-hop reasoning. In addition, CodeApex also uses algorithmic problems and corresponding test cases to evaluate the code generation capabilities of LLMs. CodeApex evaluated a total of 14 large language models on coding tasks. Among them, GPT3.5-turbo shows the best programming ability, achieving approximately 50% and 56% accuracy on these two tasks respectively. It can be seen that large language models still have a lot of room for improvement in programming tasks.

Zoom launches generative AI assistant AI Companion

On September 5, local time, the video conferencing platform Zoom announced the official release of the generative AI assistant-AI Companion, which is currently available for free to paid Zoom users.

It is reported that Zoom’s joint artificial intelligence approach dynamically integrates its own large-scale language model and third-party models such as Meta Llama 2, OpenAI and Anthropic to provide powerful real-time digital assistant functions to help users improve work efficiency and collaborate more effectively. Work. Since Zoom launched generative AI in early June, thousands of companies have tried Team Chat compose and Meeting summary for free. Starting this fall, Zoom will launch AI Companion on its platform, significantly expanding its artificial intelligence generation products.

Investment and Financing

Gleen AI receives $4.9 million in funding to use enterprise data to solve AI hallucination problem

According to reports on September 6, the startup Gleen AI recently announced that it has received US$4.9 million in financing. Investors include Slow Ventures, 6th Man Ventures, South Park Commons, Spartan Group and other venture capital companies and angel investors, including former Facebook /Meta Platform Vice President of Product Management Sam Lessin continues to develop anti-hallucination data layer software for enterprises by filtering the enterprise's own internal data, converting it into a vector database, and using this data to improve the quality of AI model answers.

AI productivity platform Kindo raises US$7 million in financing

According to reports on September 6, artificial intelligence startup Kindo recently announced that it has received US$7 million in seed round financing to launch a secure AI productivity platform for enterprises.

The goal of the Kindo platform is to empower enterprises by harnessing the power of any AI model, whether commercial, open source, or proprietary. By integrating AI into workflows, companies can significantly increase productivity while ensuring security, compliance and centralized management. The seed funding will help Kindo expand its business and enable more enterprises to adopt AI.

It is reported that Kindo was founded in September 2022 by Williams and technology industry veterans, bringing together the expertise of former chief security officers and IT executives from unicorn companies such as Bird Rides, Clover Health, and Riot Games, as well as from Google, The talent at OpenAI and Square.

Recommended Reading of the Metaverse

"Most of the core original creators of LLaMA are leaving, and the inside story of Meta AI is exposed!" The competition for computing power was fierce, and the members of the large model team changed for three consecutive rounds》

In the past six months, the popularity of ChatGPT has completely made Meta unable to sit still.

From the release of the open source large model series LLaMA in February, to the upgrade of Llama 2, and then to the coding model Code Llama, Meta is betting everything on ALL IN AI. In the open source community, the free research and commercial use of Llama series models directly ignited the fission of alternative models.

However, while Meta is in glory, the AI ​​team is facing a wave of resignations. Due to internal disputes over computing power, most of the team members of the LLaMA and OPT projects have resigned.

https://www.8btc.com/article/6832012‌

"Scammers all over the United States are robbing Nvidia"

Many start-up AI companies are outright “charlatans,” with no scalable revenue model and burning staggering amounts of capital expenditure on AI with no results. "Liar" companies across the United States are pursuing the H100 chip, thereby creating a phenomenon to the outside world that they will have an advantage in the AI ​​field. With all these overvalued, overhyped stocks, it's hard to say when the bubble will burst, but it will.

https://www.8btc.com/article/6831962‌



2424bdb719afdb7d9d78031e09c84cd5.png

Babbitt Park is open for cooperation!

89519aa36edd08d859ead38cb5fe758e.png

fe403e1e4af4f0989f8aeb8265d9fe1f.jpeg

812c156886f42633a2ca1cf96cce478c.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

7f8a55d24a0a7479f8e03f1d1eab7bbe.jpeg

Guess you like

Origin blog.csdn.net/weixin_44383880/article/details/132726257