Summary: It has been half a year since the release of Midjourney v5. When will the next major version v6 be released? The answer is: later this year. David Holz, CEO of Midjourney, recently stated at an event that the leap from v5 to v6 will be greater than the leap from v4 to v5. However, Holz did not provide an exact release date, but indicated that it will be released soon, definitely within this year.
Image Source: Generated by Wujie AI
Hot News:
Two Highs and One Department: Using "Deep Synthesis" and other generative AI technologies to commit online violent crimes, will be punished more severely according to law
Bitpush News, September 25th - The Supreme People's Court, the Supreme People's Procuratorate, and the Ministry of Public Security jointly issued the "Guiding Opinions on Punishing Online Violent Crimes in Accordance with the Law." The "Opinions" require that those who commit online violent crimes with one of the following five situations shall be punished more severely according to law: those targeting minors and disabled persons; those organizing "water armies," "hired thugs," or other personnel; those fabricating "sexual" topics to infringe upon the dignity of others; those using "deep synthesis" and other generative artificial intelligence technologies to disseminate illegal information; and those initiated and organized by network service providers.
Microsoft collaborates with Johns Hopkins University to launch ALMA, a large model focused on translation work
According to PingWest on September 25th, the Arxiv page shows that researchers from Johns Hopkins University recently collaborated with Microsoft to launch a new large language model ALMA, which is dedicated to translation work. ALMA's full name is Advanced Language Model-based trAnslator, developed based on LLaMA-2. The model introduces a new fine-tuning method, eliminating the need for a large amount of parallel data traditionally relied upon by translation models by initially fine-tuning on monolingual data and then subsequent fine-tuning on a small amount of high-quality parallel data. Experimental results show that ALMA outperforms traditional models in translation work.
Baichuan Intelligence releases Baichuan2-53B large model, commercializing the B-end market
According to "Science and Technology Innovation Board Daily" on September 25th, Baichuan Intelligence released the closed-source large model Baichuan2-53B, which enhances mathematical and logical reasoning capabilities and reduces model hallucinations through a high-quality data system and enhanced search. Baichuan Intelligence also opened API interfaces, officially entering the ToB field and commencing the commercialization process. It is understood that Baichuan2-53B is the 6th large model released by Baichuan Intelligence. Previously, Baichuan Intelligence released the Baichuan 2 large model.
Du Xiaoman opensources "Xuanyuan 70B" financial large model
According to the webmaster's home on September 25th, Du Xiaoman launched the new financial industry large model "Xuanyuan 70B." "Xuanyuan 70B" is an important financial dialogue large model project aimed at providing excellent natural language processing capabilities. The project is built on the basis of the Llama2-70B model, and after extensive pre-training in Chinese and a large amount of general and financial domain pre-training, a series of powerful financial large models have been formed.
"Xuanyuan 70B" is an open-source large model with 70B parameter level and a context length of over 8k. It incorporates a large amount of professional financial corpus in the pre-training phase, aiming to significantly enhance financial professionalism while improving general capabilities. This model has been fully open in the open-source community, including model weights and training corpus, providing the financial industry with an efficient Chinese base model. Users can also customize fine-tuning as needed.
Alibaba Cloud's Zhou Jingren: The new open-source large model Qwen-14B provides stronger cognitive capabilities
According to 36Kr on September 25th, Zhou Jingren, Chief Technology Officer of Alibaba Cloud Intelligence, announced at the Tongyi Qianwen open-source release conference that the new open-source large model Qwen-14B, based on 140 billion parameters and 3 trillion token training, provides stronger inference, cognitive, planning, and memory capabilities. In addition, compared to the previous open-source version, the new QWEN-7B (v1.1) model has generally improved capabilities, with the core indicators increasing by up to 22.5%.
AI startup company Writer, which raised $100 million, open sources its large language model Palmyra
Bitpush News, according to the public account "AIGC Open Community," the artificial intelligence startup company Writer has open sourced its large language model Palmyra on huggingface, with 8 models including small, base, 20b-chat, Instruct-20b, med-20b, etc., which are commercially available and support data fine-tuning.
It is reported that Palmyra's technical highlights include powerful features with only a small amount of parameters, which is very helpful for small and medium-sized enterprises and individual developers without computing resources; it has been trained on commercial writing and marketing data, mainly targeting enterprise users; enterprise-level data security, with multiple security barriers built in. In addition to generating text, it can also extract content summaries from videos, PDFs, and audio; support data fine-tuning, allowing enterprises to create their own "ChatGPT" assistant, etc. Earlier reports stated that the AI startup company Writer completed a $100 million financing, with a valuation exceeding $5 billion.
Investment and Financing:
Amazon to invest $4 billion in AI startup Anthropic
According to Devdiscourse, Amazon announced on Monday that it will invest $4 billion in cash in the artificial intelligence startup Anthropic to compete with its growing AI cloud competitors. As part of the deal, Amazon's employees and cloud customers will have early access to Anthropic's technology and will integrate it into their own businesses.
Anthropic also pledged to rely heavily on Amazon's cloud services, including using a large number of proprietary chips purchased from Amazon to train its future AI models. The heads of Amazon's cloud division and Anthropic stated in a joint interview that the immediate investment will be $1.25 billion, and both parties have the right to trigger an additional $2.75 billion in financing from Amazon. In addition, they refused to disclose how much stake Amazon will now hold in Anthropic, and were unwilling to disclose the latest valuation of the startup.
It is reported that Anthropic was founded by former OpenAI researchers in 2021 and is headquartered in San Francisco, USA. The company has developed a chatbot named Claude and has received investments from tech giants such as Google and Salesforce. It completed a $450 million Series C financing in May. Earlier reports stated that the US AI company Anthropic, supported by Google, received a $100 million investment from SK Telecom.
AI-native international human resources service provider TTC completes tens of millions of yuan in Pre-A round financing
According to "Science and Technology Innovation Board Daily" on September 25th, recently, AI-native international human resources service provider TTC completed tens of millions of yuan in Pre-A round financing, with the leading investor being Innovation Works.
Metaverse Recommended Reading:
"With only 1% of the parameters, the effect surpasses ControlNet, a new AI painting control master is coming"
"New AI Painting Detail Control Master" ControlNet-XS is here! The key point is that it only requires 1% of the parameters of the original ControlNet. ControlNet-XS was developed by the Computer Vision Laboratory at Heidelberg University. Currently, related papers and pre-trained models have not been released. However, researchers have stated that the ControlNet-XS FID score is significantly better than ControlNet. The code for controlling Stable Diffusion-XL and Stable Diffusion 2.1 will be open sourced soon.
https://www.8btc.com/article/6833834
"Encirclement" of NVIDIA by China and the United States, the trillion-dollar AI chip market is difficult to dominate alone
As NVIDIA takes the lead in the AI large model computing chip market and surpasses a trillion-dollar market value, Intel, AMD, and Chinese GPU chip companies are quietly competing, hoping to get a share of the AI computing chip market.
https://www.8btc.com/article/6833824
"In response to OpenAI's challenge, Midjourney will release the V6 version upgrade this year"
It has been half a year since the release of Midjourney v5. When will the next major version v6 be released? The answer is: later this year. David Holz, CEO of Midjourney, recently stated at an event that the leap from v5 to v6 will be greater than the leap from v4 to v5. However, Holz did not provide an exact release date, but indicated that it will be released soon, definitely within this year. Regarding what new things Midjourney v6 will bring, Holz promised to bring better text understanding, meaning that image generation will be closer to the described prompts and will better capture the details in the prompt wording. It may also bring text rendering functionality.
https://www.8btc.com/article/6833878
免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。
