Skip to main content

LLM Call Tool

The LLM Call Tool allows users to interact with a language model (LLM) by providing a prompt and selecting a model ID.

Steps to Use the LLM Call Tool​

Step 1: Access the Tool​

  • Click on Add User Step.
  • From the available options, select the LLM Call Tool.

Step 2: Input the Prompt​

You need to provide a prompt for the LLM. You have two options for entering the prompt:

  • Manual Entry: Type the entire prompt directly into the input field.
  • User Input Variable: Use a user input variable to dynamically provide the prompt. Use the syntax {{variable_name}} to access the value of an input variable. For example, if there is an input parameter named “promptText”, you can access its value using {{promptText}}.

Step 3: Select the Model​

Choose the model from the available list of LLM models. Ensure you select a model that best fits your use case.

Step 4: Configure Guardrails Validators (Optional)​

You can set up guardrails validators to ensure the model's output meets certain criteria. Set up validators based on your specific needs. This may include rules for output quality, relevance, or other criteria.

Step 5: Execute the LLM Call​

Once you have entered the prompt and selected the model ID, initiate the LLM call:

  1. Run the tool: Click the Run button to start the LLM call process.
  2. Results: The tool will process your request and interact with the model based on the provided inputs.

Step 6: View the Output​

The output from the LLM will be displayed in the results section. The results will be presented in a structured format, often in plain text or JSON, depending on the nature of the output.

Step 7: Utilize the Results​

You can now use the LLM's output for further processing, analysis, or integration into your application.

Tips for Effective Use​

  • Clear Prompts: Provide clear and specific prompts to get the most relevant output from the model.
  • Appropriate Model Selection: Choose the model ID that best fits your use case for accurate results.
  • Effective Guardrails: Apply guardrails validators to ensure the quality and relevance of the output.

If you have any questions or need further assistance, please contact the helpdesk([email protected]).