Generating Your First LLM Output

tip

Having problems? Don't worry, reach out on discord and we will help you out.

First, we need to install tokio in our project. Since this is a tutorial we will install the full tokio package crate, in production, of course we should be more selective with what features we install.

cargo add tokio --features full

First, let's start by writing a simple Rust program that generates an LLM output using LLM-Chain and the OpenAI driver:

use llm_chain::{executor, parameters, prompt};

// Declare an async main function
#[tokio::main(flavor = "current_thread")]
async fn main() -> Result<(), Box<dyn std::error::Error>> {
    // Create a new ChatGPT executor
    let exec = executor!()?;
    // Create our prompt...
    let res = prompt!(
        "You are a robot assistant for making personalized greetings",
        "Make a personalized greeting for Joe"
    )
    .run(&parameters!(), &exec) // ...and run it
    .await?;
    println!("{}", res);
    Ok(())
}

Understanding LLM Response

When you run the program, you'll receive an LLM response. The response contains the generated text and other metadata.

Error Handling and Common Issues

One common issue you might encounter is forgetting to set the OpenAI API key. Make sure you have set the API key in your OPENAI_API_KEY environment variable.

export OPENAI_API_KEY="YOUR_OPEN_AI_KEY" # TIP: It stars with sk-

If you don't want to set enviroment variable or want to multiple api-keys. Then you can use a different macro like this.

use llm_chain::{executor, options, parameters, prompt};
use tokio;

// Declare an async main function
#[tokio::main(flavor = "current_thread")]
async fn main() -> Result<(), Box<dyn std::error::Error>> {
    // Create a new ChatGPT executor
    let options = options! {
        ApiKey: "sk-proj-..."
    };

    let exec = executor!(chatgpt, options);
    match exec {
        Ok(exec) => {
            
            let res = prompt!(
                "You are a robot assistant for making personalized greetings",
                "Make a personalized greeting for Joe"
            )
            .run(&parameters!(), &exec) // ...and run it
            .await?;
            println!("{}", res);
        }
        Err(err) => panic!("Unable to create executor: {}", err),
    }
    // Create our step containing our prompt template

    Ok(())
}

In the next tutorial, we'll cover adding parameters to customize the LLM prompt to create more complicated interactions.