DexVerse Operation Manual
DexVerse is a cloud platform that combines cloud-native technology with MixedAI capabilities to automatically trigger, manage, schedule, and monitor Deep Learning training tasks. It mainly serves industrial vision model training and embodied intelligence R&D scenarios.
Its core functions and features include:
Task management capability: Supports full lifecycle management of Deep Learning training tasks, including automatic triggering, dynamic scheduling, and real-time monitoring. You can view task status, download data or models, and handle task failure, cancellation, continuation (
resume), and other operations.Technical integration advantage: Based on cloud-native technology, it enables elastic resource scaling and dynamic task scheduling. Combined with MixedAI simulation and machine learning capabilities, it lowers the barrier to using MixedAI, and users do not need to manage MixedAI versions or fine-grained hardware resources by themselves.
Integrated usage scenario: Through the PickWiz "One-Click Connection" function, data generation and model training can be completed in the cloud based on exported training configurations (such as
zipfiles) (supported only in the internal network).Embodied intelligence support: In embodied intelligence R&D, it provides model full lifecycle management (pre-training, fine-tuning, Inference), data engine functions (Real2Sim data projection, generative data augmentation, automated skill augmentation), online adaptive fine-tuning, and high-fidelity simulation validation.
1. Create a Task
（1）Visit the DexVerse training platform address in the production environment, http://192.168.4.135:8081/, open the following login page, and use Feishu to scan the code to log in.
（2）After logging in, enter the "New Task" page in the "Task Management" module, as shown below:
（3）On the page shown below, you can create a new training task. Note: at this point, you must fill in the task name and upload the task data file (the zip file regenerated after modification and recompression in Step2), then click the "Create Task" button to create the new task, as shown below:
（4）After the new task is successfully created, it automatically jumps to the task list page, as shown below:
（5）Click the "Details" button on the far right to open the detailed information page of the corresponding task, as shown below:
（6）Click the circular icon in the upper-right corner to open the Personal Center and Log Out pages. The Personal Center page is shown below:
（7）The Log Out page is shown below:
2. View the Trained Model Type
PickWiz 1.7.1 adds the model type field
3. View Task Details
4. Cancel a Task
Currently, only tasks in the following statuses can be canceled:
Pending Scheduling
Scheduling
Running
5. Continue (resume) a Task
Continuing a task, also called task resume, means that for a task that has already failed, you can restart it again. The task will run again on the server where it was last executed. If the target server has no available resources (GPU) at that time, it will wait in the queue.
6. Download the Model Exported by the Task
7. View Failure Reasons
