opendatagen 0.0.26

Data preparation system to build controllable AI system

⬜️ Open Datagen ⬜️

Open Datagen is a Data Preparation Tool designed to build Controllable AI Systems

It offers improvements for:

RAG: Generate large Q&A datasets to improve your Retrieval strategies.

Evals: Create unique, “unseen” datasets to robustly test your models and avoid overfitting.

Fine-Tuning: Produce large, low-bias, and high-quality datasets to get better models after the fine-tuning process.

Guardrails: Generate red teaming datasets to strengthen the security and robustness of your Generative AI applications against attack.

Additional Features

• Use external sources to generate high-quality synthetic data (Local files, Hugging Face datasets and Internet)

• Data anonymization

• Open-source model support + local inference

• (SOON) Open DataGen Agent

• (SOON) Web or Desktop app 🤔

Installation

pip install --upgrade opendatagen

Setting up your API keys

export OPENAI_API_KEY='your_openai_api_key' #(using openai>=1.2)
export MISTRAL_API_KEY='your_mistral_api_key'
export TOGETHER_API_KEY='your_together_api_key'
export HUGGINGFACE_API_KEY='your_huggingface_api_key'
export SERPLY_API_KEY='your_serply_api_key' #Google Search API 

Usage

Example: Generate a low-biased dataset to improve factuality of an LLM.

template.json:

{
    "factuality": {
        "description": "Factuality",
        "prompt": "Given the following text:\n\n'''{wikipedia_content}'''\n\nAnswer to this factually checkable question:\n'''{question}'''.",
        "completion": "Answer: '''{answer}'''. Rate the answer out of 10: {score}",
        "prompt_variation_number": 0,
        "variables": {
            "wikipedia_content": {
                "name": "Wikipedia content",
                "generation_number": 1,
                "get_value_from_huggingface": {
                    "dataset_name": "20220301.en",
                    "dataset_path": "wikipedia",
                    "column_name": "text",
                    "max_tokens": 512
                }
            },
            "question": {
                "name": "Factually checkable question",
                "generation_number": 3,
                "models": [
                    {
                        "openai_chat_model": {
                            "name": "gpt-3.5-turbo-1106",
                            "temperature": [0, 1],
                            "max_tokens": 128
                        }
                    }
                ]
            }, 
            "answer": {
                "name": "Short answer to the question",
                "generation_number": 1,
                "models": [
                    {
                        "openai_instruct_model": {
                            "name": "gpt-3.5-turbo-instruct",
                            "temperature": [0, 1],
                            "max_tokens": 128,
                            "start_with": ["Answer:"]
                        }
                    }
                ]

            },
            "score": {
                "name": "Score",
                "generation_number": 1,
                "note": ["You must answer with an integer."],
                "models": [
                    {
                        "openai_chat_model": {
                            "name": "gpt-3.5-turbo-1106",
                            "temperature": [0, 1],
                            "max_tokens": 5
                        }
                    }
                ]
            }

        }
    }
}

Python code to generate the dataset:

from opendatagen.template import TemplateManager
from opendatagen.data_generator import DataGenerator

output_path = "factuality.csv"
template_name = "factuality"
manager = TemplateManager(template_file_path="template.json")
template = manager.get_template(template_name=template_name)

if template:

    generator = DataGenerator(template=template)

    data = generator.generate_data(output_path=output_path)

    print(data)

Using this template you will:

1. Get text content from the Wikipedia dataset hosted on HuggingFace
2. Generate 3 questions about this content
3. Generate an short answer
4. Rate the answer

Contribution

We welcome contributions to Open Datagen! Whether you're looking to fix bugs, add templates, new features, or improve documentation, your help is greatly appreciated.

Note

Please note that opendatagen is initially powered by OpenAI's models. Be aware of potential biases and use the note field to guide outputs.

Acknowledgements

We would like to express our gratitude to the following open source projects and individuals that have inspired and helped us:

• Textbooks are all you need (Read the paper)

• Evol-Instruct Paper (Read the paper) by WizardLM_AI

• Textbook Generation by VikParuchuri

Connect

If you need help for your Generative AI strategy, implementation, and infrastructure, reach us on

Linkedin: @Thomas. Twitter: @thoddnn.