Student
This document is intended for users with the Student role on the YanFlow Intelligent Computing Training Platform. Students are primarily responsible for learning training courses and conducting innovative research work.
Login
User Login
Enter your Username, Password, and Verification Code, check the agreement checkbox, and click the Login button. After successful verification, you will enter the system.
Logout
Click on the User Avatar in the top-right corner, select Logout, and click Confirm to log out and return to the login page.
Training
Course Management
Course List
Displays all courses that students need to learn in the system. Supports keyword search by Course Name.
Course Details
Select a course and click the Details button to view detailed information about the course, including course information and workspaces.
Course Information: Click the Course Information tab to view basic course information, including course introduction, reserved compute node resources, course images, push commands, and course templates.
Workspaces: Click the Workspaces tab to view workspaces created by the student within the teaching class, organized by teaching class directory.
Class Learning
Attend Class
Five minutes before class or during class time, click the Attend Class button to enter the class details page, where you can view course information and create workspaces.
Course Information: Click the Course Information tab to view the course introduction, course images, and course template content. In the Course Images or Course Templates section, click the Add Workspace button to quickly create a new workspace.
Workspaces: Click the Workspaces tab to view all workspaces created during class time.
- You can Add, Suspend, Resume, or Release workspaces. You can also Set as Template for a workspace to enable quick creation using templates later.
- Port Mapping: Supports multiple methods to view the container interior.
End Class
After class ends, the system will automatically release all workspaces created during the class and reclaim node resources. After class, the Training section is unavailable during non-class hours.
Innovation
Innovation Overview
View your personal work and data related to the Innovation section on the Innovation Overview page.
Workspaces
Workspace List
Displays workspaces created by you in the Innovation section. Supports filtering by Project, Workspace Type, Workspace Status, GPU Model, CPU Model, and Node. Also supports keyword search by Workspace Name.
Add Workspace
Click the Add Workspace button to open the add dialog. Configure the project, workspace type, model, quantity, image, port mapping, volumes, commands, environment, and other information according to your actual business needs. Click the Confirm button to create a new workspace after validation.
Parameter Description
| Parameter | Description | Example |
|---|---|---|
| Workspace Type | Defines the environment type for computing resources | CPU, GPU, vGPU |
| Model | Select the required hardware model, subject to system quota and current inventory Selected models must have no teaching schedule for at least 6 hours | |
| Quantity | Number of GPU cards or cores mounted into the container for this workspace | |
| Image | Base image selected for the workspace | |
| Port Mapping | Exposes container service ports for remote connection or web access | |
| Volume | Mounts platform distributed storage (file management) into the container for data persistence | |
| CMD | Entry program or script executed by default when the container starts | python, /bin/bash |
| ARGS | Runtime parameters passed when executing CMD | |
| WORKDIR | Default entry path after image startup; commands will be executed in this path | /mnt/volume/username |
| ENV | Injects key-value custom configurations into the container runtime environment | TEMP_LIMIT=80, MODE=test |
| Dependency Installation Command | Supplements missing third-party dependencies in the image during container startup | pip install numpy |
Quick Add
Click the Quick Add button to open the quick add dialog. Select a template and click the Confirm button to quickly create a new workspace consistent with the template after validation.
Workspace Details
Click the Details button to open the workspace details page and view detailed workspace information. You can also enter an image address and click the Save Image button to save and push the workspace image as a personal user image. For successfully saved images, please go to User Images to view.
Suspend/Resume/Release
Select a workspace and click the Suspend, Resume, or Release button. A confirmation dialog will appear; click the Confirm button to complete the operation.
Set as Template
Select a workspace and click the Set as Template button. A confirmation dialog will appear; click the Confirm button to set it as a template. Only workspaces created by yourself can be set as templates.
Port Mapping
Select a workspace and click the Port Mapping button. A dialog will appear showing the port mapping information for the workspace. Multiple connection methods are supported to access the container interior, and external network remote connection can be enabled.
Job Management
Slurm
Displays Slurm jobs created by you in the system. Supports filtering by Project and Status, and keyword search by Node Name.
Add Slurm Job: Click the Add button to open the add dialog. Configure the project, select the master node (required) and compute nodes (add freely according to needs), select an appropriate image, set mounts, and click the Confirm button to create a new Slurm job after validation.
Cancel: Select a Slurm job and click the Cancel button to terminate the job and release resources. Supports viewing console and port information. Click the Save Image button, select the node for which to save the image, enter the image address, and save and push the image.
Note: The tag for Slurm job images must start with "slurm".
Job Details: Select a Slurm job and click the Job Details button to view all Slurm jobs within the container. Check one or more jobs and click Cancel to cancel them, changing their status to CANCELED.
Kubernetes
Displays Kubernetes jobs created by you in the system. Supports filtering by Project and Creation Time, and keyword search by Job Name.
Add Kubernetes Job: Click the Add Job button to open the add dialog. Configure the project, resources, image, mounts, select the executable training code path (select the directory of the code project), supplement environment dependencies through installation commands, and click the Submit button to create a new Kubernetes job after validation.
Job Details: Select a job and click the Job Name to view detailed job information, including logs and Console.
Clone/Cancel Job: Select a job and click the Cancel or Clone Job button to cancel or clone a job.
| Parameter | Description | |
|---|---|---|
| GPU Model | Select the required hardware model, subject to system quota and current inventory Selected models must have no teaching schedule for at least 6 hours | |
| Quantity | Number of nodes to be scheduled | |
| Image | Select an appropriate image ensuring PyTorch and related environments and communication libraries are installed | |
| Mounts | Mounts platform distributed storage (file management) into the container for data persistence | |
| Code Path | Select the executable training code path; the path should point to the folder containing executable files | |
| Dependency Installation Command | If dependencies are missing in the image, supplement them through the "Dependency Environment Installation Command" |
Note:
- Code projects are generally stored in the user's personal directory. The platform will mount this directory (training code path) to
/root/workdir/inside the container by default.- If no allocatable resources are available in the current environment, the system will automatically place the submitted job in the queue. To view, please go to Queued Jobs.
- Scheduled submission is supported. Jobs will enter the Scheduled Task List for standby. To view, please go to Scheduled Jobs.
Determined
Single-Machine Jobs
Displays single-machine jobs created by you in the system. Supports filtering by Project and Creation Time, and keyword search by Job Name.
Add Single-Machine Job: Click the +Single-Machine Job button to open the add dialog. Configure the project, model, quantity, image, mounts, runtime settings, and other information according to your actual business needs. Click the Submit button to create a new single-machine job after validation.
Job Details: Select a job and click the Job Name to view detailed job information.
Suspend/Resume/Cancel: Select a job and click the Suspend, Resume, or Cancel button. A confirmation dialog will appear; click the Confirm button to complete the operation.
| Parameter | Description | Example |
|---|---|---|
| GPU Model | Select the required hardware model, subject to system quota and current inventory Selected models must have no teaching schedule for at least 6 hours | |
| GPU Quantity | Number of GPU cards mounted into the container for this job | |
| Image | Base image selected for the single-machine job | |
| Mounts | Mounts platform distributed storage (file management) into the container for data persistence | |
| Entrypoint | Shell command to be executed by the user | python xxx.py, mkdir xxx |
| Training Code Path | Absolute path of the code project uploaded by the user; the system will execute the Entrypoint command in this path | |
| Dependency Installation Command | Supplements missing dependencies in the image | pip install xxx, apt-get install xxx |
Multi-Machine Jobs
Displays multi-machine jobs created by you in the system. Supports filtering by Project and Creation Time, and keyword search by Job Name.
Add Multi-Machine Job: Click the +Multi-Machine Job button to open the add dialog. Configure the project, model, quantity, image, mounts, runtime settings, and other information according to your actual business needs. Click the Submit button to create a new multi-machine job after validation.
Job Details: Select a job and click the Job Name to view detailed job information.
Suspend/Resume/Cancel: Select a job and click the Suspend, Resume, or Cancel button. A confirmation dialog will appear; click the Confirm button to complete the operation.
Note:
- If no allocatable resources are available in the current environment, the system will automatically place the submitted job in the queue. To view, please go to Queued Jobs.
- Scheduled submission is supported. Jobs will enter the Scheduled Task List for standby. To view, please go to Scheduled Jobs.
Clone Job
Select a job and click the Clone Job button to quickly clone an identical job based on the selected job.
Queued Jobs
Displays queued jobs created by you in the system. When resources meeting the conditions become available, queued jobs will run automatically. Supports filtering by Project and Creation Time, and keyword search by Job Name.
If queuing is no longer needed, click the Cancel button to cancel the queued task. Project leaders can drag and adjust the queue order according to business needs.
To ensure teaching training tasks, the system executes the following scheduling policy:
- Within the next N hours, if a teaching class requests similar resources, queued jobs will not be allocated and will remain in the queue until teaching resources are released.
- The value of parameter N is dynamically adjusted by the system administrator.
Scheduled Jobs
Displays scheduled jobs submitted by you in the system. Supports filtering by Project and Creation Time, and keyword search by Job Name.
Click the Modify Schedule button to change the scheduled submission time. If submission fails, hover over the failure button to view the reason for submission failure.
Settings
Node Management
Image Operation Logs
Displays image operation logs for nodes in the system. Supports querying by Operation Type, Operation Status, and Operation Time. Click the Log button to view specific log content.
Image Management
User Images
My Shared Images: Displays images pushed and shared by you. Supports keyword search by Image Address and Pre-installed Environment. Also supports Edit, Share, and Delete operations for images.
User Image Push: Click the Push Command button to push your own images according to the instructions.
Note: When an image does not have applicable resources and applicable brands set, manual selection is required. After selection, the corresponding image can be used when creating workspaces or jobs.
Course Images
View pushed course images by course. Supports keyword search by Image Address and Pre-installed Environment.
Template Images
Displays template images in the system. Supports keyword search by Image Address and Pre-installed Environment.
File Management
Training Files
Displays all files in the training platform. Supports keyword search by File Name.
Course Folder: Students can view the course folders for their courses. Course folders include the course shared folder and teaching class folders.
- Course Shared Folder:
class_public- Students can view all files uploaded by teachers to the course shared folder. - Teaching Class Folder: The folder for the student's teaching class. Click to view the personal folder for that teaching class (named by username). Students can store homework data in the Homework Folder
homework_submissionunder their personal folder. Supports Create, Copy, Upload, Download, Rename, Move, Extract, Delete, and more. Teachers will collect all files under thehomework_submissionfolder as homework.
Innovation Files
Displays all files in the innovation platform. Supports keyword search by File Name.
Public Folder: Students can view all subfolders or files in the public folder and support downloading files within the folder.
Project Folder: Students can view and manage all project folders, including Create, Copy, Upload, Download, Rename, Move, Extract, Delete, and more.
User Folder: Students can view and manage their personal innovation folders, including Create, Copy, Upload, Download, Rename, Move, Extract, Delete, and more.
System Management
Application Management
Displays all applications in the system. Supports keyword search by Application Name. Students can download applications.
Personal Center
Click Settings - Personal Center or Personal Center in the top-right username to enter the Personal Center page, where you can view personal basic information and quotas.
Modify Phone Number: Click the Modify button for the phone number, enter the new phone number and verification code, and click the Confirm button to modify the phone number.
Set/Modify Email: Click the Set or Modify button for the email, enter the email address and verification code, and click the Confirm button to set the email.
Modify Login Password: Click the Modify button for the password, enter Original Password, New Password, Confirm New Password, and click the Submit button to successfully modify the login password. After successful password modification, you will be logged out and need to log in again.
Message Management
Click Settings - Message Management or the Message Icon in the top-right corner to enter the Message Management page, where you can view all messages pushed to you by the system. If there are unread messages, click the Mark All as Read button to mark all unread messages as read.
