Auto-Labeler

An AI-powered data labeling library for Python.

Key Features

• 🔍 Discovery: Automatically suggest labels/taxonomy using Iterative or Embedding-based discovery.
• 🏷️ Assignment: Labels your dataset using LLMs (Gemini, OpenAI, Anthropic, etc.).
• ⚡ Batching & Async: High-throughput processing for large datasets.
• 💾 Disk Caching: Save costs and time with local persistence.
• 💰 cost Tracking: Real-time USD cost estimation for every run.
• 🛡️ Validation: Pydantic-powered fail-fast checks for your data.

Installation

pip install auto-labeler-ai

Quick Start

from auto_labeler import AutoLabeler
import pandas as pd

labeler = AutoLabeler(model_name="gemini/gemini-2.5-flash")
results = labeler.label_dataset(
    pd.read_csv("data.csv"),
    labels=["Urgent", "Billing", "General"],
    context="Customer support tickets"
)
print(results[["text", "label"]])  # output column is always 'label'
print(labeler.get_usage())

Documentation

For full guides on Advanced discovery, Caching, and API reference, visit our: 👉 Documentation Site

Testing

pytest

License

MIT