Argilla

What is Argilla?

Argilla is an open-source collaboration platform for AI engineers and domain experts to build high-quality datasets. It targets language model fine-tuning, RLHF and evaluation, and it has joined Hugging Face.

Key capabilities of Argilla

• Expert annotation: domain experts focus on the data that matters
• Fine-tuning data: datasets for LLM fine-tuning
• RLHF feedback: collect human feedback for alignment
• Evaluation sets: build datasets for model evaluation
• Automation: automates routine tasks from prototype to production
• API and UI: Python API and web interface for NLP datasets

How Argilla works

Engineers create datasets through the API, and domain experts review and label records in the web interface. Automation handles routine steps so curated data moves from prototype to production.

Who uses Argilla?

Research institutions, data science teams and AI engineers curating NLP datasets. A public demo is available and a Discord community supports practitioners.

Argilla pricing

Argilla is open source and the vendor homepage lists no paid plans. It now sits within Hugging Face, so hosted options follow Hugging Face terms.

Argilla alternatives

Segments.ai and CVAT are annotation tools, Toloka and Invisible Technologies provide human data services, and Lightly curates image data.