lightningrod-ai 0.1.17

Python SDK for dataset generation on LightningRod platform ⚡

Lightning Rod Python SDK

The Lightning Rod SDK provides a simple Python API for generating custom forecasting datasets to train your LLMs. Transform news articles, documents, and other real-world data into high-quality training samples automatically.

Based on our research: Future-as-Label: Scalable Supervision from Real-World Outcomes

Documentation: docs.lightningrod.ai

👋 Quick Start

1. Install the SDK

pip install lightningrod-ai

2. Get your API key

Sign up at dashboard.lightningrod.ai to get your API key and $50 of free credits.

3. Generate your first dataset

Generate 1000+ forecasting questions in minutes - from raw sources to labeled dataset, automatically. ⚡

from lightningrod import LightningRod, BinaryAnswerType, QuestionPipeline, NewsSeedGenerator, ForwardLookingQuestionGenerator, WebSearchLabeler

lr = LightningRod(api_key="your-api-key")

binary_answer = BinaryAnswerType()

pipeline = QuestionPipeline(
    seed_generator=NewsSeedGenerator(
        start_date=datetime.now() - timedelta(days=90),
        end_date=datetime.now(),
        search_query=["Trump"],
    ),
    question_generator=ForwardLookingQuestionGenerator(
        instructions="Generate binary forecasting questions about Trump's actions and decisions.",
        examples=[
            "Will Trump impose 25% tariffs on all goods from Canada by February 1, 2025?",
            "Will Pete Hegseth be confirmed as Secretary of Defense by February 15, 2025?",
        ]
    ),
    labeler=WebSearchLabeler(answer_type=binary_answer),
)

dataset = lr.transforms.run(pipeline, max_questions=3000)
dataset.flattened(binary_answer)  # Ready-to-use data for your training pipelines

We use this to generate the Future-as-Label training dataset for our research paper.

🆕 New: Foresight-v3 Forecasting Model

We've released foresight-v3, our latest forecasting model. Use it via the OpenAI-compatible API for probability estimates on forecasting questions:

from openai import OpenAI

client = OpenAI(
    api_key="your-api-key",
    base_url="https://api.lightningrod.ai/api/public/v1"
)

response = client.chat.completions.create(
    model="LightningRodLabs/foresight-v3",
    messages=[{"role": "user", "content": "Will the Fed cut rates by 25bp in March 2025?"}]
)
print(response.choices[0].message.content)

See the API docs for full details, or try the Foresight-v3 notebook.

✨ Examples

We have some example notebooks to help you get started! If you have trouble using the SDK, please submit an issue on Github.

Tutorials

Example Name	Path	Google Colab Link
Quick Start	notebooks/01_quick_start.ipynb	Open in Colab
News Datasource	notebooks/02_news_datasource.ipynb	Open in Colab
Custom Documents	notebooks/03_custom_documents_datasource.ipynb	Open in Colab
Binary Answer Type	notebooks/04_binary_answer_type.ipynb	Open in Colab
Continuous Answer Type	notebooks/05_continuous_answer_type.ipynb	Open in Colab
Multiple Choice Answer Type	notebooks/06_multiple_choice_answer_type.ipynb	Open in Colab
Free Response Answer Type	notebooks/07_free_response_answer_type.ipynb	Open in Colab
BigQuery Datasource	notebooks/08_bigquery_datasource.ipynb	Open in Colab
Foresight-v3 Model	notebooks/09_foresight_model.ipynb	Open in Colab

End-to-end

Example Name	Path	Google Colab Link
Golf Forecasting	notebooks/e2e/golf_forecasting.ipynb	Open in Colab
Trump Forecasting	notebooks/e2e/trump_forecasting.ipynb	Open in Colab
Document Classification	notebooks/e2e/document_classification.ipynb	Open in Colab
Polymarket Backtesting	notebooks/e2e/polymarket_backtesting.ipynb	Open in Colab
Model Consensus	notebooks/e2e/model_consensus.ipynb	Open in Colab
Survival LLM	notebooks/e2e/survival_llm.ipynb	Open in Colab

For full documentation, see docs.lightningrod.ai. For the SDK API reference in this repo, see API.md.