Public Resource Management
Public resources refer to resources that are visible to all users and bound to the running container. HyperAI Through a special account HyperAI To publish public resources.at present HyperAI There are three types of public resources supported:
- Public Tutorial
- Public dataset
- Publicly pre trained models
Public Tutorial
For safety reasons, The tutorial that is currently running will not be visible to other users, Only the closed container with the highest serial number will be displayed for execution or snapshot. How to create a snapshot? See [file](/docs/gear/workspace/#Create a snapshot for the workspace).
Creation of Public Tutorial
Public tutorials usually refer to executable code. Jupyter Collection of workspaces, about HyperAI Regarding the account. One "Public Tutorial" Actually, it's just a closed one "Execution of containers" .Follow these steps to create a public tutorial.
1. Switch to HyperAI account number. Create a new container
As mentioned above, **One "Public Tutorial" Actually, it's just a closed one "Execution of containers" **
2. Store the resources required for the tutorial in a container
There are multiple methods to upload the corresponding resources to the container:
- adopt GitHub Download to container
- adopt Jupyter The upload function of the editor uploads resources from the local to the container
All resources should be stored in containers /hyperai/home Under the directory, Files saved in other directories cannot be saved after the container is closed.
3. provide README.md file (Optional)
Default in container /hyperai/home A path can be provided README.md of Markdown The formatted file serves as an instructional document for the tutorial. This feature is related to GitHub The warehouse's README.md The function is consistent.
After completing the editing and organization of the files inside the container, you can close the container.
4. Set the container as a public type
As shown in the above figure. Set this tutorial as a public resource. It can be done here "Public Tutorial" I have seen the corresponding resources on the page.
There is also a limit to the number of public resources created by users. Administrators can configure the following settings "Public container" Modify the number of items:
Update of Public Tutorial
When there are multiple executions in the container. The content displayed in the public tutorial is always the latest execution or snapshot content. If you need to update the content in the public tutorial, simply create a new version in the corresponding public container.
For safety reasons, The tutorial that is currently running will not be visible to other users, Only the closed container with the highest serial number will be displayed for execution or snapshot. How to create a snapshot? See [file](/docs/gear/workspace/#Create a snapshot for the workspace).
The removal of public tutorials
Reset the visible range of the public container to "private" You can remove the public tutorial from the platform.
Public dataset & Publicly pre trained models
The content displayed in the public dataset and public pre trained model is HyperAI Publicly available under the account data set Types of data and Pre trained model Type of data. The type of dataset can be determined by its "set up" Page modification.
The public dataset has an additional option as shown in the following figure:
If checked "Allow batch download of files" Other users can download the entire dataset in compressed form. Compare and select as shown below "Allow batch download of files" Effect of Whether or Not:
Set up a public dataset & Publicly pre trained models
and "Public Tutorial" The logic is similar, take HyperAI After setting the dataset under the account to public type, it will be "public resource " Display in the corresponding type.
After setting the dataset to public. All versions of the dataset under this dataset will be publicly available. But unlike "Public Tutorial" Just showing the latest execution.
There is also a limit to the number of public resources created by users. Administrators can configure the following settings "Public datasets" Modify the number of items:
Provide README.md file (Optional)
and "Public Tutorial" similar. One can be stored in the root directory of each version of the dataset README.md file. Used to describe the version of the dataset.
Public dataset & The removal of publicly available pre trained models
Reset the public dataset & The visible range of publicly available pre trained models is "private" The public resource can be taken down.