Managing Fine-Tuned Models¶
The Fine-Tuned Models tab in the Models section allows you to create, customize, and manage models fine-tuned to your specific use cases.
- Create a Model: Fine-tune a Kore-hosted model or import one from Hugging Face. Configure training parameters, datasets, and hardware resources as needed.
- Deploy a Model: Once fine-tuning is complete, deploy the model within the Agent Platform or externally using the generated API endpoint. You can also fine-tune further on top of an already fine-tuned model.
Once deployed, the model is listed on the Fine-tuned models page, showing its deployment count and other details.
Viewing the Model List¶
The Fine-Tuned Models listing page displays all available models along with the following details:
Field | Description |
---|---|
Model Name | Name of the fine-tuned model. |
Base model | The model used as the base for fine-tuning. |
No. of deployments | Number of times the model has been deployed. |
Tags | Labels associated with the deployment. |
Added by | User who created or imported the model. |
Updated on | Date of the deployment. |
Select a model in the list to open the model’s detail view. By default, this opens the Overview page. The left navigation pane includes the following sections:
- Overview– View general information, training progress, and test information.
- Deployments – Manage all deployments of the selected model.
- Configurations– Edit model name and description, add tags, adjust the model endpoint timeout duration, or delete the model.
Managing Model Deployments¶
Each model can have multiple deployments, which are tracked independently.
Click the Deployments tab to see all deployments for the selected model, along with the following details:
Field | Description |
---|---|
Deployment Name | Name given by the user during deployment. |
Deployment ID | System-generated ID (not editable). |
Status | The status of deployment: Deploying, Optimizing, Failed, Ready to Deploy, or Deployed. | Deployed By | User who performed the deployment. |
Added On | Date of deployment. |
Actions |
Copy cURL – Copies the cURL command for invoking this deployment. Manage API Keys – Opens the API key management tab. Re-trigger – Restarts the deployment (Option available only if it has failed or stopped). |
Managing Deployment Details¶
Selecting a specific deployment on the Deployment page opens its detail view, where you can view and manage the configuration, endpoint, and API keys for that deployment.
- Model Endpoint – View or manage the live endpoint; re-deploy the model if needed (specific to this deployment).
- API Keys – Generate and manage keys scoped to this deployment. API keys are isolated per deployment for secure access control.
- Configurations – Edit the description and tags, or undeploy/delete the model.