Ant Group
2月29
-
Ant Group Develops Multimodal Remote Sensing Base Model SkySense
09:45Ant Group announced the launch of SkySense, a multimodal remote sensing base model with 2 billion parameters. the model Ant Bailing Big Model is the latest research result in the field of multimodal, and its related paper has been received by the world's top computer vision conference CVPR 2024. According to the data, SkySense's performance in 17 test scenes exceeds that of similar international products, and can be applied in the fields of geomorphology, crop observation and interpretation, effectively assisting agricultural production and operation.
1月02
-
Ant Group: open source China's first medical specialty inference dataset
09:36Ant Group has officially open-sourced the first medical specialty reasoning dataset in China. In order to solve the problems of lack of repeated exploration of symptoms, phantom problems and insufficient reasoning ability of general-purpose language models in medical questioning, Ant Group and a team of urology experts from Shanghai Renji Hospital jointly developed this dataset, and based on the clinical experience of a team of doctors, the Chinese medical specialty Q&A reasoning dataset RJUA-QA was launched by constructing simulated case data, which is the industry's first clinical specialty dataset in the industry. The dataset consists of three parts: training, validation, and testing, and contains a total of 2,132 Q&A pairs, each of which is written by a physician...
-
Ali launches AnyText, an AI drawing model that can write!
09:35Alibaba has launched an AI drawing tool called AnyText that can accurately add text to images and supports four languages, Chinese, English, Japanese and Korean. The tool's features include arbitrary specification of text position, support for multiple text styles and material imitations, and the ability to modify text already in the image. It is based on the diffusion model development, using a modular approach to generate text and embedded in the image to improve the writing accuracy and consistency of text and background. Meanwhile, AnyText's deployment method and comparison test results demonstrate its advantages in text accuracy and image generation.
12月14
-
Ant Group's 20 papers selected for AI Top Session NeurlPS
09:17Ant Group excelled at NeurlPS, the top conference in AI and machine learning recently, with 20 papers featured, covering a wide range of AI and machine learning fields such as computer vision, natural language processing, graph neural networks, and image processing. Among them, more than 70% of the papers focused on solving the challenges and dilemmas encountered by generative AI in the process of rapid development. Ant Group officially open-sourced OpenASCE (Open All-Scale Causal Engine), the industry's first distributed all-link causal learning system, at a workshop on the first day of the conference. It is reported that Ant Group ...
12月09
-
Ali Quark Big Model Passes Filing, Helping Quark App Become AI Assistant for Young People
10:13Alibaba Intelligent Information Business Group's self-developed Quark Big Model has passed the filing, and will gradually upgrade the content products and intelligent tools in the fields of general knowledge, health, and creation. The Quark Big Model has a wide range of application prospects for search products and intelligent tools in general knowledge, health, creation and other areas. As a popular information service product among young people, Quark App will accelerate its development into an AI assistant for young people with the help of its self-developed big model. Quark Big Model has made new breakthroughs in dialog and problem solving capabilities in education, healthcare and other vertical fields, and has become one of the outstanding representatives of home-grown self-research Big Model.
12月04
-
AliCloud's First Tongyi Thousand Questions AI Challenge
09:21At the Tongyi Qianqian conference, Aliyun announced the first "Tongyi Qianqian AI Challenge", which is divided into two tracks: algorithmic and agent. Participants have free access to the Tongyi open source model family, including the newly released 72 billion parameter model Qwen-72B. The challenge is divided into two tracks: algorithmic and agent. The Algorithm track aims to explore the upper limit of the code capability of the open source models by fine-tuning the training of the Qwen models, while the Agent track requires participants to develop next-generation AI applications based on the Qwen models and the Agent-Builder framework of the Magic Hitchhiker community, in order to promote the practical application of the models in a wide range of industries. ...
10月31
-
AliCloud Tongyi Thousand Questions App goes live on Android App Market
09:29October 30th, it was reported that recently, Ali Cloud's big model Tongyi Qianqian App shelved on the major Android application market, version number 1.0.2.The application introduction shows that Tongyi Qianqian is a super-large scale pre-training model, which is able to provide all-around assistance to the user in many aspects, such as creative copywriting, office assistants, learning assistants, and interesting life. Currently, Android users can download and install the app and register for the experience via their cell phone number.
10月18
-
Ant Group Takes Stake in Cognitive Big Model R&D Firm Smart Spectrum AI
19:35Beijing Zhipu Huazhang Technology Co., Ltd. has undergone an industrial and commercial change, and the new shareholder is Ant Group's subsidiary Shanghai Yunfu Enterprise Management Consulting Co. The company's official website shows that Zhi Spectrum AI is committed to building a new generation of cognitive intelligence general model, cooperation in research and development of bilingual hundreds of billions of mega-scale pre-training model GLM-130B, and based on this to build ChatGLM, but also launched the cognitive big model platform Bigmodel.ai.