Skip to content

Custom fork of Bittensor’s ai_factory, focused on validator improvements, feature experimentation, and integration with MindNode. Contains personal modifications and test branches for new ideas. Original upstream: star145s/ai_factory

License

Notifications You must be signed in to change notification settings

GStar718/ai_factory

 
 

Repository files navigation

AI-Factory: Specialized AI Development Subnet

AI-Factory AI Factory is a dedicated subnet accelerating AI research by combining automated ideation with community collaboration.

How to mine in AI Factory subnet?

Our AI Scientist Assistant generates novel paper ideas and outlines, enabling the rapid development of specialized AI pipelines. We also host key events: the AI Factory Conference, which showcases and peer-reviews new research, and the Community Workshop, focused on surveys and knowledge sharing.

Current Miner Emission Distribution:

  • Scientist AI Tracks: 30% (5% for all new models to encourage participation)
  • AI Factory Conference: 0% (will increase to 50% after release)
  • Community Workshop: 70%

Table of Contents

Scientist AI Tracks

1. Research Track

On going

This track aims to develop AI models capable of generating original scientific ideas, with a particular emphasis on proposing new methods in the AI domain. Participants will explore the potential of generative AI in devising innovative solutions to complex scientific problems.

2. Demo Track

In progress

The demo track serves as a platform to continuously produce AI models addressing specific problems, aiming to impress end-users and deliver value across various industries. We envision that, in the future, models from the research track could automate the ideation process for the demo track, reducing the need for manual brainstorming of intriguing AI topics.

AI Factory Conference

In progress

The AI Factory Conference is a decentralized research event that leverages the Bittensor network’s token emissions to incentivize high-quality AI research and peer review. Miners across the network are called upon to submit novel work, review their peers’ submissions, and collaboratively elevate the state of the art in machine learning and related fields.

Community Workshop

On going

Miners are invited to contribute knowledge by researching and presenting AI topics or papers (original or survey) in a concise video format. Approved presentations earn a share of Bittensor emissions and drive community learning.

Getting Started

To set up and participate in AI-Factory, please refer to the following guides:

See Scientist AI Competitions

See Community Workshop to learn how to submit presentation.

See AI Factory Conference to learn about the conference.

See Validator Setup to learn how to set up a Validator.

Our codebase for the Scientist AI competition is adapted from Subnet 9, developed by Macrocosmos, which specializes in pre-training models within the Bittensor ecosystem. We extend our sincere gratitude to Macrocosmos for providing a strong foundation for our work. To manage contributions to the Community Workshop and AI Factory Conference, we use a separate system. Please refer to the documentation for more details.

Contact Us

For inquiries, collaborations, or more information:

About

Custom fork of Bittensor’s ai_factory, focused on validator improvements, feature experimentation, and integration with MindNode. Contains personal modifications and test branches for new ideas. Original upstream: star145s/ai_factory

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 97.8%
  • Jupyter Notebook 1.4%
  • Shell 0.8%