Skip to main content

Model Details Page

The Model Details Page in the Gesund.ai platform presents all relevant information and configuration options for a deployed or draft model. It consolidates technical metadata, live system usage, user access settings, and linked resources, offering a centralized hub to manage and assess your models.

1. Model Overview​

This section summarizes key attributes of the model:

  • Model Name & Status: View model name and current deployment state (Deployed, Archived, or Stopped).
  • Framework & Modality: Shows the technical environment (e.g., PyTorch, TensorFlow) and data type (CT, MRI, etc.).
  • Inference Speed: Indicates how quickly the model performs predictions.
  • Class Mapping & Tags: Lists mapped labels and any associated keywords for indexing or filtering.

2. Resource Monitoring​

Track system utilization in real time:

  • CPU and Memory Gauges: Visual indicators show how much of the allocated system resources are being consumed.
  • Useful for evaluating system load and scaling needs for batch predictions.

3. Container Logs​

Access detailed system logs generated by the model’s container:

  • Includes logs for:
    • Batch prediction events
    • Scaling actions
    • Warnings or errors
  • Each log entry is timestamped and filterable for easier debugging.

4. Source Files​

Explore the files that constitute the uploaded model package:

  • View a tree structure of all source code and metadata files (e.g., .py, .txt, requirements.txt).
  • Individual files can be opened and reviewed inline.
  • Enables transparency into the model’s internal logic and dependencies.

5. Linked Validations​

Check which validation runs are associated with this model:

  • Each row displays:
    • Validation type and progress
    • Linked dataset
    • Modality and run status
  • If no validations are attached, the system will show a placeholder message.

6. Access Management​

Manage access control for your model:

  • Add or remove user permissions by email address.
  • Choose visibility level:
    • Public: All users can access.
    • Restricted: Access limited to specific users with assigned rights (e.g., read-only, edit).

7. Output Mapping​

Configure how model outputs are interpreted and formatted:

  • Map outputs like:
    • Segmentation masks
    • Prediction class
    • Confidence score
    • Status flags
  • Customize output structure for better downstream compatibility.

8. Model Source Info​

Display metadata about where the model came from:

  • Information includes:
    • Origin name
    • Contact email or website
    • Reference institution or dataset

9. Inference Explorer​

Track individual predictions made by the model:

  • Browse inference history with:
    • Batch ID
    • User
    • Dataset used
    • Time taken per image
  • Filter by validation run or time range.
  • Download results for external analysis.

10. Model Actions & Utilities​

Manage the model via the contextual action menu:

  • Available Actions:

    • Watch / Unwatch
    • Archive
    • Mark as Preemptive
    • Edit Metadata
    • Delete Permanently

  • Status Badge: Displayed prominently, indicating the model’s operational state.

This page acts as a control center for any model deployed to the platform, providing both technical depth and administrative control for efficient model lifecycle management.