Yitong AI Studio (HCS version)
AIStudio's one-stop artificial intelligence development and application platform is oriented towards closed-loop data application and has full-process end-to-end AI application development and deployment support from data management and processing, model development and optimization, model deployment and application, system operation and maintenance and management. Capabilities Huawei Cloud Market Yitu Artificial Intelligence Platform provides end-to-end artificial intelligence solutions to users in different industries, allowing users to develop and deploy artificial intelligence at the fastest speed and in the least time.
- The underlying hardware resources are heterogeneous and compatible with servers of different architectures such as underlying X86/ARM, Huawei NPU/NVIDIA GPU, etc.
- Compatible with mainstream deep learning frameworks including Huawei MindSpore, TensorFlow and PyTorch
- Provide end-to-end full-stack AI development, optimization, and inference deployment capabilities such as data management and processing, model development and optimization, model deployment and application, etc.
- Unified access management of all types of data, providing a variety of intelligent data processing tools for ETL, annotation and other intelligent processing of all types of data
- Provide customers with "deep learning + machine learning" full-scenario AI development and application infrastructure services
Yitong AI Studio aims to provide end-to-end artificial intelligence solutions for users in different industries, allowing users to carry out artificial intelligence development and deployment work at the fastest speed and in the least time. Yitong AI Studio customizes a large number of complex artificial intelligence tools, such as multi-machine multi-GPU/NPU training, super-parameter search, and neural network structure search, into the platform to facilitate the use of all cluster users. Yitong AI Studio can support end-to-end AI processing of all scene data (unstructured data, structured data and time series data): data integration, data cleaning, data annotation, model training, model optimization, and model deployment.
The specific functions of each module are as follows:
1. Data management platform
It supports the unified management of unstructured and structured data, provides data channels, data storage, data management, data display and other functions, and can realize the production of data sets to meet model training and evaluation.
(1) Supports processing of data sets such as images, natural language, and audio/voice
(2) Supports multiple data set management, such as COCO, YOLO, MINIST, MINISTinCSV, CVAT, ImageNet, Kitti
(3) Support the import of data sets in any format outside this platform
(4) Supports uploading of very large data sets, data segmentation and cutout, etc.
2. Artificial Intelligence Platform
Provides the development, training, evaluation and release of deep learning models based on unstructured data, provides a professional and flexible development environment, and supports team collaborative development, version management, resource statistics and other functions.
(1) Support multiple computing resources for model development and training
(2) Support elastic scheduling of computing resources
(3) Support multiple deep learning frameworks
(4) Support super parameter tuning
(5) Supports multiple model visualization tools
3. Data annotation platform
An efficient independent data annotation platform that supports multiple data annotation scenarios such as image classification, target detection, image segmentation, text detection, and text recognition. It has powerful automatic annotation, flexible multi-person annotation, and easy-to-use batch annotation functions.
(1) Support multiple types of application scenarios
(2) Support multi-person annotation
(3) Support automatic labeling
(4) Support batch annotation
4. Model factory
Model Factory is a model management center that supports model warehousing, model upload, format conversion, version control, model combination and other management. It also presets a rich and multi-category model to meet different application scenarios.
(1) Model combination of multiple application scenarios to reduce costs and improve efficiency
(2) Hundreds of preset multi-category models, such as CV, NLP, Recommend, MultiModal, etc., can meet the needs of multiple scenarios
(3) Automatic conversion between model formats
5. Reasoning center
Provide inference services adapted to different models, support central reasoning and edge reasoning, provide secure access to massive edge nodes, and help users conduct unified management, monitoring, and operation and maintenance of inference services.
(1) Support heterogeneous hardware access, unified management and operation and maintenance
(2) Flexibly meet users’ reasoning and deployment needs of different models
(3) Support full life cycle management of inference services
(4) Support inference data to be transmitted back to the cloud in real time
(5) Complete closed loop of incremental learning, model update, and push
6. Mirror Center
Supports reasonable and effective image management, such as uploading, downloading, version control, permission control, etc., and provides preset images for use.
(1) Multiple images and super large images can be uploaded to the image center quickly and easily, saving users time
(2) Verify the legitimacy of uploaded images to reduce rework costs
(3) Reasonable version management to facilitate user use
7. Operation and maintenance center
Operation and maintenance center APSC (Apulis Service Center) provides visual resource health monitoring, customized alarm rule combinations, alarm notifications, and supports resource configuration, task management, system settings and other functions.
(1) Comprehensive resource monitoring and visual resource health monitoring
(2) Customized alarm rule combinations and multi-form alarm notifications
(3) Resource management, flexible enterprise-level organization management resources
(4) Flexible and convenient system management, one-click installation and configuration
8. Unified authentication service IAM
The unified identity authentication service IAM (Identity and Access Management) has functions such as single sign-on, authentication management, centralized authorization and auditing based on permission policies.
(1) Organizational management
(2) Role management
(3) User group management
(4) User management
All users
Product pricing
Product price = ordering interval quantity 1 * interval specification unit price 1 + ordering interval quantity 2 * interval specification unit price 2.
For example: the call rate is 0.3 yuan/minute for the part that does not exceed 3 minutes, and 0.2 yuan/minute for the part that exceeds 3 minutes; if the user calls for 8 minutes, the fee charged is 3*0.3+5*0.2=1.9 yuan.
Product price = order quantity * unit price of the range to which the quantity belongs.
For example: the call rate is 0.3 yuan/minute for no more than 3 minutes, and 0.2 yuan/minute if it exceeds 3 minutes; if the user calls for 2 minutes, the fee charged is 2*0.3=0.6 yuan; if the user calls for 8 minutes, the fee charged The cost is 8*0.2=1.6 yuan.
Product price = order quantity * unit price.
The product price is the price displayed on the page.
After-sales service time: 5 * 8 hours After-sales service content: fault recovery, technical support Service hotline: 0755-26559725 Service email: contact@apulis.com
By ordering this product you agree toGoods and Services Agreement》