Navigating the Workspace
The EdgeBench Workspace serves as a central hub and personal storage area where users can manage their projects and models. It includes two main tabs:
- Projects – dedicated to viewing and managing project builds for inference tasks.
- My Models – provides access to user-uploaded models via the BYOM feature.
Users can also explore pre-trained models via the Model Zoo tab.
Common user controls
These controls are consistently available across both Projects and Models sections in the Workspace, as well as in the model zoo. They help users manage lists and individual items efficiently.
Filter panel
Located on the left side of the interface, the Filter Panel allows users to narrow down lists using the following criteria:
Modality
- Audio
- Vision
- Time-series
Framework
- Filter by AI training framework (e.g., TensorFlow, PyTorch, ONNX)
Size
- Use the horizontal slider to filter models by file size, ranging from >1 KB to >50 MB.
Clear All
- Click this button to remove all selected filters
Sort options
Accessible via dropdown menus, sorting helps organize entries based on:
Date Last Modified
- Latest First (Descending)
- Oldest First (Ascending)
File Size
- Largest to Smallest
- Smallest to Largest
File Name
- A to Z (Ascending)
- Z to A (Descending)
Upload Date (for models only)
- Newest First
- Oldest First
Search bar
A keyword-based search tool that helps users locate specific projects or models by name or function. Available in:
- Workspace → Projects
- Workspace → My Models
- Model Zoo
Card actions
Each card (project or model) includes an ellipsis menu offering quick actions:
- Rename
- Edit the name of the item.
- Delete
- Permanently remove the item from the list.
Note
For BYOM models, deletion is only allowed if the model is not linked to any projects.
Top bar overview
| No. | Control | Function |
|---|---|---|
| 1 | Workspace | Workspace provides each user with 10 GB of personal cloud storage to store AI projects and models, including BYOM uploads. It allows users to view and manage projects, builds, and related resources in a centralized location. |
| 2 | Model Zoo | Contains ADI-verified and pre-trained models available for deployment and evaluation. |
| 3 | Theme Toggle | Switch between light and dark mode using the sun and moon icons. |
| 4 | Notification Bell | Available to keep users informed about ongoing and completed activities. Whenever a compilation or inference process is initiated, a corresponding notification is generated and displayed. |
| 5 | Profile Avatar | The Profile menu includes the following options: My Profile – Allows users to view and edit their personal profile details. Online Help – Directs users to support resources and documentation for guidance on using the application. Send Feedback – Enables users to share feedback, report issues, or suggest improvements. EngineerZone – Provides access to a community and knowledge-sharing platform for technical discussions and support. Log Out – Allows users to securely sign out of the application. |
Projects tab
The Projects Tab is the central dashboard for managing AI projects within the Workspace. It presents all user-created projects in a visually organized card layout, making it easy to browse, access, and manage work at a glance.
Project list management
Refer to Common User Controls for:
- Filter Panel
- Sort Options
- Search Bar
- Rename
- Delete
Project card info
The Projects Tab displays all user-created projects as cards. Each card displays the following information:
| Displayed Info | Description |
|---|---|
| Project Title | Name of the project |
| Task Type | Model task type (e.g., classification, detection) |
| Framework | AI training framework used (e.g., TensorFlow, PyTorch) |
| Model Name | Associated model name |
| Project Size | Total storage used by the project |
| Last Update | Timestamp of the most recent modification |
Project details page
When a user clicks on a project card in the Workspace → Projects tab, they are taken to the Project Detail Page, which is divided into the following key sections:
This section provides high-level information about the selected project:
| Field | Description |
|---|---|
| Project Name | Editable field showing the name of the project. |
| Task Type | Indicates the model's task (e.g., classification, detection). |
| Framework | Shows the training framework used (e.g., TensorFlow, PyTorch). |
| Model Name | Displays the model associated with the project. |
| Project Size | Total storage used by the project files. |
Clicking on a Project card opens the Build Version listing.
Build version listing
This dashboard provides a comprehensive log of all model evaluation runs performed on each ADI hardware under the selected project. This view enables users to trace, audit, and compare inference performance across different build versions and configurations.
Each row represents a build created for a specific device and includes the following information:
| Field | Description |
|---|---|
| Status | Displays the current build state such as Compilation Completed, In Progress, or Failed. |
| Last Updated Timestamp | Shows when the build was last modified. |
| File Size | Size of the generated build artifact. |
| Build Actions | Options to Export Build, Open Build, or Delete Build. |
Clicking on a build entry opens the Build Details Panel on the right.
Status tags overview
Status tags provide real‑time visibility into the progress and outcome of both the build and inference workflows. Each step in the process generates a corresponding status indicator, allowing users to quickly understand what the system is doing, whether an action is ongoing, completed, or requires attention.
Note that these tags are also visible during the build process, shown at the upper right portion of the window.
| Status Tag | Description |
|---|---|
| Created | Indicates the beginning of the build process. The project already exists, and the user can proceed with device selection. |
| Ready | States that device, model, and configuration settings have been fully set. |
| Compilation Queued | Compilation request has been initiated; waiting to begin processing |
| Compilation Failed | Errors were encountered during compilation. User needs to check the logs for more details and address the errors for the compilation to continue. |
| Compilation Running | Compilation is currently in progress. |
| Compilation Successful | The compilation completed with no errors. |
| Compilation Aborted | Triggered when the user aborts the “Start Compilation”, this action stops the compilation process. |
| Compilation Completed | All configurations for the selected device and model have been successfully compiled and applied. This unlocks the next step in the build process. This status is reflected through Steps 3 (Uploading Input Data), 4 (Pre-processing Script), and 5 (Post-processing Script). |
| Inference Queued | This status shows after clicking the “Start Inference” button; the request for deployment to the selected device has begun. |
| Inference Failed | Indicates an unsuccessful inference attempt. Logs should be checked for details. |
| Inference Aborted | Inference started but was stopped (user-initiated stop). |
| Inference Running | The inference operation is currently being executed on the selected device. |
| Inference Completed | Inference ran successfully on the selected device. User can now view the results. |
Build details side panel
This panel provides detailed information about the selected build version:
| Section | Description |
|---|---|
| Build Details | Displays the current status of the build. This section also provides two key actions: • Export Build – Allows the user to export the compiled build artifact to their local machine; this enables to download the build for deployment or manually share with another user. • Open Build – Directs the user to the active build step for further configuration or review. |
| Compilation Parameters | Displays flags and settings used during model compilation. |
| Open Step (Compile) | Allows users to go back to Step 2 - Compilation and modify compilation settings. |
| Inference Details | Shows deployment status, file count, and inference result status. |
| Open Step (Inference) | Enables users to open Step 6 - Deploy and Inference and modify inference settings. |
Note: The Open Step buttons are disabled for builds with inactive status.
Available controls
| Control | Function |
|---|---|
| Rename Project | Click the pencil icon next to the project name to edit. |
| Delete Build | Click the trash icon next to a build entry to remove it. |
| Delete Project | Use the top-right button to delete the entire project. |
Models tab
The Models Tab in Workspace displays all user-uploaded models. Each model is presented as a card with key information for quick reference and management.
Model list management
Refer to Common User Controls for:
- Filter Panel
- Sort Options
- Search Bar
- Rename
- Delete
Note
BYOM models can only be deleted if not linked to any projects.
Model card info
Each model card includes the following fields:
| Field | Description |
|---|---|
| Model Name | Name of the uploaded model |
| Task Type | Model task type (e.g., classification, detection) |
| Framework | Model training framework used (e.g., TensorFlow, PyTorch) |
| Model Size | File size of the model |
| Upload Date | Timestamp of when the model was uploaded |
| Associated Projects | Number of projects linked to this model |
Model detail page
Clicking a model card opens the Model Detail Page, which is divided into the following components:
Model overview
This section provides high-level information about the selected model:
| Element | Description |
|---|---|
| Model Name | Displayed prominently; editable if needed |
| Task Type | Indicates the AI task type |
| Framework | Shows the AI training framework used |
| Model Size | File size of the model |
Associated projects listing
The Associated Projects section provides a quick overview of all projects linked to a specific model.
Model properties side panel
This panel displays basic metadata about the model:
| Property | Description |
|---|---|
| Task | General task category (e.g., vision, audio) |
| Task Type | Specific model task (e.g., object detection) |
| Framework | Model training framework used |
| Owner Name | User who uploaded the model |
| Upload Date | When the model was added to the Workspace |
| Last Updated | Most recent modification timestamp |