Unified Management of Development Elements
The platform provides four elements of AI development: unified management of model code, data set, model and execution environment.
* Only show the dynamics of open source projects
Recommended Organizations
These excellent organizations are using the OpenI AI Collaboration Platform for collaborative development of projects. To show your organization here, Click here to submit.
More OrganizationsCommunity Activities
The community has prepared a wealth of activities, waiting for you to participate!
Recommended Projects
Excellent AI projects recommendation. To show your project here, Click here to submit.Click here to explore more projects.
智算网络
人工智能算力网络推进联盟已接入11家智算中心，算力总规模1924P
Collaborative Development Environment
Provide a collaborative development environment for AI development, which is the biggest highlight that distinguishes the OpenI AI Collaboration Platform from other traditional Git platforms.
Data Collaboration and Sharing
By uploading data sets in the project, many project members cooperate to complete data preprocessing. You can also establish a better model with community developers by setting the data as a public dataset.
Model Management and Sharing
Associate the model with the code version, you can adjust the model in different ways based on the historical version of the code and save the results. The trained model can be open and shared, so that more people can use the model to test and give feedback.
Once Configuration, Multiple Reuse
Provide execution environment sharing, Once Configuration, Multiple Reuse. Lower the threshold of model development, and avoid spending repetitive time configuring complex environments.
OpenI AI Collaboration Platform
OpenI AI collaboration platform has cooperated with Pengcheng cloud brain and China computing power network (c ² Net) can be used to complete AI development tasks by using the rich computing resources of Pengcheng cloud brain and China computing network.
Pengcheng CloudBrain's existing AI computing power is 100p FLOPS@FP16 (billions of half precision floating-point calculations per second), the main hardware infrastructure consists of GPU servers equipped with NVIDIA Tesla V100 and A100, and Atlas 900 AI clusters equipped with Kunpeng and shengteng processors.
China computing power network (c ² Net) phase I can realize high-speed network interconnection between different artificial intelligence computing centers, and realize reasonable scheduling of computing power and flexible allocation of resources. At present, 11 intelligent computing centers have been connected, and the total scale of computing power is 1924p OPS@FP16 。 Qizhi AI collaboration platform has been connected to Pengcheng Cloud Computing Institute, Chengdu Intelligent Computing Center, Zhongyuan Intelligent Computing Center, Hefei brain and other nodes.
Developers can freely select the corresponding computing resources according to the use needs, and can test the adaptability, performance, stability, etc. of the model in different hardware environments.
If your model requires more computing resources, you can also apply for it separately.