AI Singularity Network comprehensive report on July 8丨AI Singularity Network丨The whole network account has the same name丨
Alibaba Cloud released an AI drawing authoring model
On July 7, at the 2023 World Artificial Intelligence Conference, Alibaba Cloud announced that the AI painting creation model can assist humans in image creation, and can be applied to application scenarios such as art design, e-commerce, games, and cultural creativity in the future.
Among them, in the Wensheng graph capability, Tongyi Wanxiang improves the training stability through prior data to ensure the rationality of the results, and uses multilingual predictive enhancement training to improve the semantic comprehension ability, optimizes the high diffusion process based on the signal-to-noise ratio index, and improves the richness of the details of the generated images.
The application scenario of the style transfer capability is that when users are satisfied with the semantics of the picture and have special requirements for the style, the universal meaning can automatically realize a series of migration of the image style.
Lookalike image generation is for situations where users want to get multiple similar image results. Universal meaning can use robust visual representation, fix the semantics and style of the image, and then generate based on diffusion to achieve the diversity of results.
SenseTime's new big model upgrade: surpassing ChatGPT
On July 7, SenseTime released the latest upgraded SenseGPT model, which surpassed ChatGPT in three global authoritative evaluation benchmarks, based on SenseChat.
In addition, based on the other major AIGC platforms of Ridaynew, this time there is also a big upgrade in one breath:
- SenseMirage 3.0: The number of parameters has been increased to 7 billion to achieve professional photography-level image detail portrayal.
- SenseAvatar 2.0: Improves voice and lip fluency by more than 30% to achieve 4K HD video effects.
- Qiongyu 2.0 (SenseSpace): 20% improvement in space reconstruction efficiency and 50% improvement in rendering performance.
- SenseThings 2.0: The rendering accuracy has been greatly improved, and the restoration of item textures and materials has reached millimeter-level fineness.
SenseTime hopes to continue to promote the leap and improvement of AI infrastructure capabilities through "big model + big device", not only to create a basic model with more powerful general capabilities, but also to further efficiently integrate the expertise of different vertical fields to build a professional large model with better understanding of the industry and more expertise, fundamentally reduce the downstream application cost and threshold of the large model, and let the industrial value of the large model bloom in thousands of industries.
HUAWEI CLOUD released Pangu Model 3.0
Recently, HUAWEI CLOUD released Pangu Model 3.0. Zhang Pingan, executive director of Huawei and CEO of HUAWEI CLOUD, said that Pangu Model 3.0 is an industry-oriented large model series, including a "5+N+X" three-layer architecture.
- The L0 layer includes five basic large models of natural language, vision, multimodality, prediction, and scientific computing, providing a variety of skills to meet the needs of industry scenarios.
- HUAWEI CLOUD can provide industry-specific large models trained using industry public data, or train your own proprietary large models based on industry customers' own data.
- The L2 layer provides customers with more detailed scene models and provides customers with "out-of-the-box" model services.
The official website of WPS AI was launched, and the recruitment of intelligent office experience officers was opened
Kingsoft Office brought its intelligent office assistant WPS AI based on large language model to yesterday's 2023 WAIC, WPS AI official website (ai.wps.cn) was launched simultaneously, and opened the channel to recruit intelligent office experience officers, WPS Office individual users and WPS 365 enterprise users can apply.
As the first ChatGPT-like application in the domestic collaborative office track, WPS AI is connected to multiple office product components of Kingsoft Office, and the original light documents, light forms, and forms are upgraded after being connected to WPS AI, and are renamed WPS Intelligent Documents, WPS Smart Forms, and WPS Smart Forms.
During the World Artificial Intelligence Conference, Kingsoft Office, MiniMax, Baidu Wenxin, and Zhipu AI three large language model partners appeared on the same stage to conduct practical demonstrations and explanations of WPS AI in office scenarios.
China Telecom released TeleChat large model
According to the Securities Times e company, at today's 2023 WAIC China Telecom "Computing-Network Integration and Sunac Future" sub-forum, China Telecom Digital Intelligence Technology Branch officially released China Telecom's large-language model TeleChat, and demonstrated products in three directions: data middle office, intelligent customer service and smart government affairs empowered by the large model.
In addition, TeleChat-E, an educational version of the large model based on TeleChat, ranks seventh on the global large-model comprehensive examination evaluation list C-Eval, and the top few include GP4, ChatGPT and other well-known large models.
[This article is from 丨AI Singularity Network丨The whole network account with the same name丨Welcome to search and follow】