Detect

A method for real-time evaluations or continuous monitoring to detect quality issues in LLM-generated text.

Parameters:

• generatedText (string): The text generated by the LLM that needs to be analyzed.
• context (array of strings): An array providing relevant context to analyze the generatedText against.
• userQuery (string, optional): The user query or prompt that resulted in the generatedText.
• config (object, optional): A dictionary of configuration options for the detector. Default value is { hallucination: { detector_name: 'default' } }.
• instructions (string, optional): Additional instructions provided to the LLM.
• asyncMode (boolean, optional): If true, the detect() function will return immediately with a DetectResult object. Default is false.
• publish (boolean, optional): If true, the payload will be published to AIMon and can be viewed on the AIMon UI. Default is false.
• applicationName (string, optional): Required if publish is true. Specifies the name of the application in AIMon.
• modelName (string, optional): Required if publish is true. Specifies the name of the model in AIMon.

Usage Example

import Client from "aimon";

// Create the AIMon client using an API Key (retrievable from the UI in your user profile).
const aimon = new Client({ authHeader: "Bearer API_KEY" });

const runDetect = async () => {
  const generatedText = "your_generated_text";
  const context = ["your_context"];
  const userQuery = "your_user_query";
  const config = { hallucination: { detector_name: "default" } };

  // Analyze the quality of the generated output using AIMon
  const response = await aimon.detect(
    generatedText,
    context,
    userQuery,
    config,
    undefined, // Optional instructions
    false, // asyncMode
    true, // publish
    "YourApplicationName", // Required if publish is true
    "YourModelName" // Required if publish is true
  );

  console.log("Response from detect:", response);
};

runDetect();