Adala: Autonomous Data Labeling Agent Framework

Introduction

Adala is a cutting-edge Autonomous Data (Labeling) Agent framework designed to revolutionize the way AI professionals approach data processing and labeling tasks. This robust framework enables the implementation of specialized agents that can autonomously acquire and refine skills through iterative learning, making it an invaluable tool for AI engineers, machine learning researchers, data scientists, and educators alike.

Key Features

🌟 Reliable Agents

Adala's agents are built upon a solid foundation of ground truth data, ensuring consistent and trustworthy results for all your data processing needs.

🎮 Controllable Output

For each skill, users can configure desired outputs and set specific constraints with varying degrees of flexibility. This allows for tailored results that meet exact project requirements.

🎯 Specialized in Data Processing

While excelling in diverse data labeling tasks, Adala agents can be customized for a wide range of data processing needs, making them versatile tools in any AI professional's toolkit.

🧠 Autonomous Learning

Adala agents are not just automated; they're intelligent. They develop skills iteratively and independently based on their environment, observations, and reflections.

✅ Flexible and Extensible Runtime

The runtime environment in Adala is highly adaptable. A single skill can be deployed across multiple runtimes, facilitating dynamic scenarios like student/teacher architectures. The framework's openness invites community extensions, ensuring continuous evolution.

🚀 Easily Customizable

Quickly customize and develop agents to address specific challenges without facing a steep learning curve, making Adala accessible to both beginners and experts.

Who Can Benefit from Adala?

• AI Engineers: Architect and design AI agent systems with modular, interconnected skills.
• Machine Learning Researchers: Experiment with complex problem decomposition and causal reasoning.
• Data Scientists: Apply agents to preprocess and postprocess data, with native Python notebook integration.
• Educators and Students: Use Adala as a teaching tool or as a base for advanced projects and research.

Getting Started

Installation

pip install adala

For the latest updates, install directly from GitHub:

pip install git+https://github.com/HumanSignal/Adala.git

Prerequisites

Set your OpenAI API key:

export OPENAI_API_KEY='your-openai-api-key'

Quickstart Example

import pandas as pd
from adala.agents import Agent
from adala.environments import StaticEnvironment
from adala.skills import ClassificationSkill
from adala.runtimes import OpenAIChatRuntime

# Set up your agent
agent = Agent(
    environment=StaticEnvironment(df=train_df),
    skills=ClassificationSkill(
        name='sentiment',
        instructions="Label text as positive, negative or neutral.",
        labels={'sentiment': ["Positive", "Negative", "Neutral"]},
        input_template="Text: {text}",
        output_template="Sentiment: {sentiment}"
    ),
    runtimes = {
        'openai': OpenAIChatRuntime(model='gpt-3.5-turbo'),
    },
    default_runtime='openai'
)

# Train and run the agent
agent.learn(learning_iterations=3, accuracy_threshold=0.95)
predictions = agent.run(test_df)

Available Skills

Adala offers a variety of pre-built skills, including:

• ClassificationSkill
• SummarizationSkill
• QuestionAnsweringSkill
• TranslationSkill
• TextGenerationSkill
• OntologyCreator
• Math Reasoning

Each skill comes with example notebooks and Colab integration for easy experimentation and learning.

Conclusion

Adala represents the future of autonomous data processing and labeling. By providing a flexible, powerful framework for creating intelligent agents, it empowers AI professionals to tackle complex data challenges with ease and efficiency. Whether you're conducting cutting-edge research, developing production-level AI systems, or educating the next generation of data scientists, Adala offers the tools and capabilities to drive your projects forward.