wyra 0.0.6

Fine Tuning Data Maker

Wyra

Wyra is a Python library designed to create and format data for fine-tuning OpenAI models, specifically tailored for Azure OpenAI services. It simplifies the process of converting text content into the JSON Lines (JSONL) format required for fine-tuning conversational AI models.

"Wyra" is a term used in indigenous languages to refer to a bird. Just like a "bird", this library will help you "fly" through the implementation of fine-tuning. Enjoy!

Features

• Easy usability: Any text in multiple languages ​​can be used.
• Easy Integration: Seamlessly integrates with Azure OpenAI services.
• Secure Data Handling: Utilizes encryption for sensitive data.
• Automated Formatting: Converts text content into JSONL format effortlessly.
• Customizable Prompts: Allows for the creation of flexible fine-tuning datasets for various use cases.

Installation

To install Wyra, use pip:

pip install wyra

Usage

Here's a basic example of how to use Wyra:

# Import necessary libraries
import json
from wyra import FineTuningDataMaker

# Initialize the Fine-Tuning data maker
creator = FineTuningDataMaker()

# Sample content to format
content = "Your text content here (You can use any text in any language you want)."

# Create and format data
formatted_data = creator.format_data(content)

# Save JSONL data to a file
with open('formatted_data.jsonl', 'w') as file:
    file.write(formatted_data)

Multiple Texts

Here's an example of how to use Wyra with multiple texts:

# Import necessary libraries
import json
from wyra import FineTuningDataMaker

# Initialize the Fine-Tuning data maker
creator = FineTuningDataMaker()

# Sample contents to format
contents = [
    "First text content here.",
    "Second text content here.",
    "Third text content here."
]

# Create and format data for multiple contents
formatted_data = [creator.format_data(content) for content in contents]

# Save JSONL data to a file
with open('formatted_data.jsonl', 'w') as file:
    for data in formatted_data:
        file.write(data + '\n')

License

This project is licensed under the MIT License. See the LICENSE file for details.

Contact

For any questions or feedback, this is my Github GitHub repositories.

Other AI Projects

Check out my GitHub Pages: GitHub IO.