Project Overview

Introduction

Welcome to ast-js-language-model, an initiative aimed at transcending the boundaries set by today's most well-known systems. While these existing systems have made significant contributions, they are often inflated in terms of complexity and resource consumption. This project is dedicated to achieving similar or even superior results using only a fraction of the resources, with an emphasis on simplicity and functionality.

Goals

1. Surpass Existing Standards: Challenge and surpass the levels reached by current prominent systems.
2. Efficient Resource Utilization: Achieve high-quality outcomes using just 1/10,000 of the resources typically required.
3. Simplicity and Effectiveness: Focus on straightforward solutions that work without unnecessary complexity.
4. Collaboration and Community: Foster a collaborative environment where everyone's contribution is valued.

Philosophy

This project recognizes that automation doesn't have to be overly complex or resource-intensive. While we may not claim to be the best in the field, we are committed to making the best use of what we have. We believe that with everyone's collaboration, we can indeed create something of high quality, all while keeping things simple and efficient.

Getting Involved

We invite you to join us in this exciting endeavor. Whether you're an expert or a beginner, your collaboration can make a difference. Explore the various sections of this wiki to learn more about the project and find out how you can contribute.

Conclusion

Don't forget what is simple and works. Together, we can redefine what's possible in automation and create something truly remarkable. Thank you for being a part of this journey.

language model implemetation for js

1. yarn install
2. node build_tokens.js
3. node train.js

allows running models with length>4096 on 8gb video cards, it's just an example of operation. to activate the video card, just remove the comments in the header of train.js