Module specialized 

Specialized AI APIs for ranking, moderation, and classification.

This module provides unified interfaces for specialized AI tasks that go beyond text generation, including document ranking/reranking, content moderation, and text classification.

Ranking Example

use llmkit::{RankingProvider, RankingRequest};

let provider = CohereProvider::from_env()?;

let request = RankingRequest::new(
    "rerank-english-v3.0",
    "What is the capital of France?",
    vec![
        "Paris is the capital of France.",
        "Berlin is the capital of Germany.",
        "London is the capital of England.",
    ],
);

let response = provider.rank(request).await?;
for result in &response.results {
    println!("Score: {:.4} - {}", result.score, result.document.as_ref().unwrap());
}

Moderation Example

use llmkit::{ModerationProvider, ModerationRequest};

let provider = OpenAIModerationProvider::from_env()?;

let request = ModerationRequest::new("omni-moderation-latest", "Some user content to check");

let response = provider.moderate(request).await?;
if response.flagged {
    println!("Content was flagged!");
    for category in response.flagged_categories() {
        println!("  - {}", category);
    }
}

Classification Example

use llmkit::{ClassificationProvider, ClassificationRequest};

let provider = CohereClassifyProvider::from_env()?;

let request = ClassificationRequest::new(
    "embed-english-v3.0",
    "I love this product!",
    vec!["positive", "negative", "neutral"],
);

let response = provider.classify(request).await?;
let top = &response.predictions[0];
println!("Predicted: {} (score: {:.4})", top.label, top.score);

Structs

ClassificationExample

Example for few-shot classification.

ClassificationPrediction

A classification prediction with confidence score.

ClassificationRequest

Request for text classification.

ClassificationResponse

Response from a classification request.

ModerationCategories

Moderation category flags.

ModerationModelInfo

Information about a moderation model.

ModerationRequest

Request for content moderation.

ModerationResponse

Response from a moderation request.

ModerationScores

Moderation category confidence scores (0.0 to 1.0).

RankedDocument

A ranked document with its relevance score.

RankingMeta

Metadata from ranking API response.

RankingModelInfo

Information about a ranking model.

RankingRequest

Request for ranking/reranking documents.

RankingResponse

Response from a ranking request.

Enums

ModerationInput

Input types for multi-modal moderation.

Statics

MODERATION_MODELS

Registry of known moderation models.

RANKING_MODELS

Registry of known ranking models.

Traits

ClassificationProvider

Trait for providers that support text classification.

ModerationProvider

Trait for providers that support content moderation.

RankingProvider

Trait for providers that support document ranking/reranking.

Functions

get_moderation_model_info

Get moderation model info by ID.

get_ranking_model_info

Get ranking model info by ID.