Argilla

Developer Documentation

Argilla provides a full-featured Python SDK and REST API for integrating dataset curation into your MLOps pipeline.

Installation

pip install argilla

Quickstart Example

import argilla as rg

# Connect to your Argilla server
rg.init(api_url="http://localhost:6900", api_key="YOUR_API_KEY")

# Create and log a dataset
dataset = rg.DatasetForTextClassification.from_pandas(df)
rg.log(dataset, name="my-classification-dataset")

Key Capabilities for Developers

• Distilabel integration — Use the companion distilabel library for scalable AI-assisted data generation and synthetic labeling pipelines.
• Hugging Face Hub — Push and pull datasets directly to/from the Hugging Face Hub using native integrations.
• Active learning support — Plug in your models to prioritize uncertain samples for annotation, reducing labeling costs.
• REST API — Full REST API available for custom integrations and programmatic dataset management.
• Self-hosting — Deploy via Docker with full control over your data and infrastructure.

Refer to the Argilla Docs and Distilabel Docs for full API references and advanced guides.