Welcome to the Predacons Project!

Hello! We're the team behind the Predacons project. We're passionate about creating a powerful, efficient, and user-friendly machine learning model training tool. Our project is open-source and we're committed to keeping it that way.

What is Predacons?

Predacons is a Python-based project that simplifies the process of training, fine-tuning and using large language models. It provides a set of functions to read and clean data, train models, and manage the training process. It's designed to be easy to use, yet flexible enough to handle complex training scenarios.
it automatically optimizes the fine-tuning process to meet the vram cap by automatically suggesting and applying quantization and peft_config while maintaining ability to use any configuration of model and training parameter it user chooses to do so. It also provides speculative decoding and fast generation that speeds up the output generation by up to 6 time the native approach provided by pytorch and hugging face transformers.

Why Sponsor Us?

Your sponsorship will help us continue to maintain and improve Predacons. With your support, we can dedicate more time to fixing bugs, adding new features, and providing support to our users. Your sponsorship also helps us keep Predacons free and open-source.

What We'll Do With Your Sponsorship

All funds from sponsorships will go directly towards:

• Maintaining the Predacons codebase
• Developing new features and improvements
• Providing support to our users
• Keeping Predacons free and open-source

We're grateful for any support you can provide. Thank you for considering sponsoring the Predacons project!