Crate prediction_guard 

Prediction Guard api client. Used to communicate with Prediction Guard API.

You must have an API key to use the client. Once you have your API key you create an instance of client::Client. This will allow access to all the endpoints.

Example

use prediction_guard::{chat, client};

#[tokio::main]
async fn main() {
    let clt = client::Client::new().expect("client value");

    let req = chat::Request::<chat::Message>::new("Neural-Chat-7B".to_string())
                .add_message(
                    chat::Roles::User,
                    "How do you feel about the world in general?".to_string(),
                )
                .max_tokens(1000)
                .temperature(0.85);

    let result = clt.generate_chat_completion(&req)
                    .await
                    .expect("error from generate chat completion");

    println!("\nchat completion response:\n\n {:?}", result);
}

See the /examples directory for more examples.

Modules

chat

Data types that are used for the chat endpoints, including chat completions, chat vision and chat events.

client

Used to connect to the Prediction Guard API.

completion

Data types that are used for the completion endpoints.

embedding

Data types to be used in the embedding endpoint.

factuality

Data types that are used for the factuality endpoints.

image

Utility module used to download and base64 encode an image.

injection

Data types used for the injection endpoint.

models

Data types that are used for the factuality endpoints.

pii

Data types for the PII detection endpoint.

rerank

Data types that are used for the rerank endpoint.

tokenize

Data types that are used for the tokenize endpoint.

toxicity

Data types for the toxicity endpoint.

translate

Data types used for the translate endpoint.

Type Aliases

Result