elluminate 1.1.6

The Official Elluminate SDK

elluminate SDK

elluminate SDK is a Software Development Kit that provides a convenient way to interact with the elluminate platform programmatically. It enables developers to evaluate and optimize prompts, manage experiments, and integrate elluminate's powerful evaluation capabilities directly into their applications.

Installation

Install the elluminate SDK using pip:

pip install elluminate

📚 Full Documentation

The full documentation of elluminate including the SDK can be found at: https://docs.elluminate.de/

Quick Start

Prerequisites

Before you begin, you'll need to set up your API key:

1. Visit your project's "Keys" dashboard to create a new API key
2. Export your API key and service address as environment variables:

export ELLUMINATE_API_KEY=<your_api_key>
export ELLUMINATE_BASE_URL=<your_elluminate_service_address>

Never commit your API key to version control. For detailed information about API key management and security best practices, see our API Key Management Guide.

Basic Usage

Here's a simple example to evaluate your first prompt:

from elluminate import Client

# Initialize the client
client = Client()

# Create a prompt template
template, _ = client.get_or_create_prompt_template(
    name="Concept Explanation",
    messages=[{"role": "user", "content": "Explain the concept of {{concept}} in simple terms."}],
)

# Generate evaluation criteria for the template
template.get_or_generate_criteria()

# Create a collection with test cases
collection, _ = client.get_or_create_collection(
    name="Concept Variables",
    defaults={
        "description": "Template variables for concept explanations",
        "variables": [{"concept": "recursion"}],
    },
)

# Run a complete experiment (generates responses + rates them)
experiment = client.run_experiment(
    name="Concept Evaluation Test",
    prompt_template=template,
    collection=collection,
    description="Evaluating concept explanation responses",
)

# Print results
for response in experiment.responses():
    print(f"Response: {response.response_str}")
    for rating in response.ratings:
        print(f"  Criterion: {rating.criterion.criterion_str}")
        print(f"  Rating: {rating.rating}")

Alternative Client Initialization

You can also initialize the client by directly passing the API key and/or base url:

client = Client(api_key="your-api-key", base_url="your-base-url")

Advanced Features

Batch Evaluation with Experiments

For evaluating prompts across multiple test cases:

from elluminate import Client
from elluminate.schemas import RatingMode

client = Client()

# Create a prompt template
template, _ = client.get_or_create_prompt_template(
    name="Math Teaching Prompt",
    messages=[{"role": "user", "content": "Explain {{math_concept}} to a {{grade_level}} student using simple examples."}],
)

# Generate evaluation criteria
template.get_or_generate_criteria()

# Create a collection with multiple test cases
collection, _ = client.get_or_create_collection(
    name="Math Teaching Test Cases",
    defaults={"description": "Various math concepts and grade levels"},
)

# Add test cases in batch
collection.add_many(
    variables=[
        {"math_concept": "fractions", "grade_level": "5th grade"},
        {"math_concept": "algebra", "grade_level": "8th grade"},
        {"math_concept": "geometry", "grade_level": "6th grade"},
    ]
)

# Run the experiment (handles all response generation and rating)
experiment = client.run_experiment(
    name="Math Teaching Evaluation",
    prompt_template=template,
    collection=collection,
    description="Evaluating math explanations across different concepts and grade levels",
    rating_mode=RatingMode.DETAILED,  # Get reasoning with ratings
)

# Print results for each response
for response in experiment.responses():
    variables = response.prompt.template_variables.input_values
    print(f"\nConcept: {variables['math_concept']}, Grade: {variables['grade_level']}")
    print(f"Response: {response.response_str[:100]}...")

    for rating in response.ratings:
        print(f"  • {rating.criterion.criterion_str}: {rating.rating}")
        if rating.reasoning:
            print(f"    Reasoning: {rating.reasoning}")

Evaluating External Agents

To evaluate responses from external systems (LangChain agents, OpenAI Assistants, custom APIs):

from elluminate import Client
from elluminate.schemas import RatingValue

client = Client()

# Set up template and collection
template, _ = client.get_or_create_prompt_template(
    name="Agent Evaluation",
    messages=[{"role": "user", "content": "Answer: {{question}}"}],
)
template.get_or_generate_criteria()

collection, _ = client.get_or_create_collection(
    name="Agent Test Cases",
    defaults={"variables": [{"question": "What is Python?"}]},
)

# Create experiment WITHOUT auto-generation
experiment = client.create_experiment(
    name="External Agent Eval",
    prompt_template=template,
    collection=collection,
)

# Get responses from your external agent
external_responses = ["Python is a high-level programming language..."]
template_vars = list(collection.items())

# Upload responses and rate them
experiment.add_responses(responses=external_responses, template_variables=template_vars)
experiment.rate_responses()

# Analyze results
for response in experiment.responses():
    passed = sum(1 for r in response.ratings if r.rating == RatingValue.YES)
    print(f"Pass rate: {passed}/{len(response.ratings)}")

Additional Resources

• General Documentation
• Key Concepts Guide
• API Documentation