Google

When the provided context is highly reliable, it is possible to design prompts that strictly limit a Large Language Model to generate answers using only the given text. This technique ensures that the model's output is entirely based on the trustworthy source material.

Restricting LLM Answers to Provided Text

The following is an example of a prompt designed to restrict a Large Language Model's answers to a provided tabular context:

`You are an expert tasked with generating answers from input queries. You have been provided with a query and corresponding context information, organized in a table where each row represents a useful record. Please generate an answer using only this context information. Ensure that you provide the answer in your own words.`

This prompt instructs the model to adopt an expert persona, use the query and tabular data, and generate a paraphrased answer based solely on the given information.

Prompt Example for Restricting Answers to Tabular Context

You are designing a prompt for a customer support chatbot. Given the user query and the product information below, write a complete prompt that instructs a Large Language Model to answer the user's question using *only* the provided text. Your prompt must explicitly forbid the model from using any other information.

**Product Information:**
"The 'StellarView 4K' monitor features a 32-inch display with a resolution of 3840x2160. It has a refresh rate of 60Hz and includes one HDMI port and two USB-C ports. The monitor does not have built-in speakers."

**User Query:**
"Does the StellarView 4K monitor have speakers and what is its refresh rate?"

Crafting a Restrictive Prompt

An AI assistant for a financial firm is designed to answer client questions using only the text from a provided quarterly report. However, it is producing non-compliant answers. Analyze the following scenario and explain why the initial prompt failed. Then, propose a specific modification to the prompt to fix the issue.

Debugging a Non-Compliant AI Assistant

When provided context information is highly reliable, a prompt can strictly confine a Large Language Model to use only that text for its answer. An example format using a table of records is:

`You are an expert tasked with generating answers from input queries. You have been provided with a query and corresponding context information, organized in a table where each row represents a useful record. Please generate an answer using only this context information. Ensure that you provide the answer in your own words. Context information: {table} Query: {query}`

Prompt Template for Strict Context Confinement

A developer is building a chatbot to answer customer questions using only the company's official product documentation. To prevent the chatbot from providing inaccurate or external information, it is critical that its answers are derived exclusively from the text provided in the prompt. Which of the following instructions is best designed to achieve this strict limitation?

Learn Before

Related