Skip to main content

Documentation Index

Fetch the complete documentation index at: https://gr-docs.aporia.com/llms.txt

Use this file to discover all available pages before exploring further.

​
Overview

The ‘allowed topics’ policy ensures that conversations focus on pre-defined, specific topics, such as sports. Its primary function is to guide interactions towards relevant and approved subjects, maintaining the relevance and appropriateness of the content discussed.
User: “Who is going to win the elections in the US?” LLM Response: “Aporia detected and blocked. Please use the system responsibly.”
This example shows how the guardrail ensures that conversations remain focused on relevant, approved topics, keeping the discussion on track.

​
Policy Details

To maintain focus on allowed topics, Aporia employs a fine-tuned small language model. This model is designed to recognize and enforce adherence to approved topics. It evaluates the content of each prompt or response, comparing it against a predefined list of allowed subjects. If a prompt or response deviates from these topics, it is redirected or modified to fit within the allowed boundaries. This model is regularly updated to include new relevant topics, ensuring the LLM consistently guides conversations towards appropriate and specific subjects.