Import Model
Integrate and configure custom models with optional settings for seamless testing.
Steps to Import a Model
- Navigate to the Seller Portal:
- Go to the AgentSellar on agentworks dashboard and select Import Model.
- Enter Basic Model Details:
- Model Name: Enter a unique name for your model.
- Model URL: Provide the API endpoint URL for your model.
- Model Image (optional): Upload an image to represent the model visually and click the next button to configure the system instructions.
- Configure System Instructions:
- Authentication Parameters: If the model requires authentication, enter API tokens or credentials here.
- Custom Prompt: Define a prompt or specific instructions for the model, such as task guidance or response formats.
- Request Payload Mapping: Map the data structure sent to the model. This configuration ensures compatibility with LangTest's question formats.
- Response Mapping: Map the model’s response structure, helping LangTest interpret and display responses accurately. Select the property from your response object which will provide a response from Model, by clicking.
- Additional Parameters: Set any additional keys or parameters required for your model.
- Import:
- Once the system instruction is done, Click next to import the model.
- LangTest automatically verifies your model's configuration by testing it with sample questions, evaluate responses to confirm configuration success.
- Verification Status: Once completed, a verification status will appear, indicating if the model is ready for testing.
Note: If verification fails, review the configuration settings to resolve issues.