decision.rs

use llm_client::prelude::*;

/// Multiple reason requests across a temp gradient with the winner being returned. Currently, decision is bound to the one_round reasoning workflow. Workflows relying on grammars are only supported by local LLMs.
#[tokio::main(flavor = "current_thread")]
pub async fn main() {
    // Using a preset model from Hugging Face
    let llm_client = LlmClient::llama_cpp()
        .mistral7b_instruct_v0_3()
        .init()
        .await
        .unwrap();

    // A boolean decision request
    let response = llm_client
        .reason()
        .boolean()
        .decision()
        .set_instructions("Is the sky blue?")
        .return_primitive()
        .await
        .unwrap();
    assert!(response);

    // An integer decision request
    let mut reason_request = llm_client.reason().integer();
    // Settings specific to the primitive can be accessed through the primitive field
    reason_request.primitive.lower_bound(0).upper_bound(10);
    let mut decision_request = reason_request.decision();
    let response = decision_request
        .set_instructions("How many fingers do you have?")
        .return_primitive()
        .await
        .unwrap();
    assert_eq!(response, 5);

    // Options
    let mut decision_request = llm_client.reason().integer().decision();
    // Set the number of 'votes', or rounds of reasoning, to be conducted
    decision_request.best_of_n_votes(5);
    // Uses a temperature gradient for each round of reasoning
    decision_request.dynamic_temperature(true);

    // An integer request, but with an optional response
    let response = decision_request
        .set_instructions("How many coins are in my pocket?")
        .return_optional_primitive()
        .await
        .unwrap();
    assert_eq!(response, None);
}