Skip to content

0-5788719150923125/vtx

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

vtx

"What is this place?" you ask.

Ryan shakes his head and laughs, %FLUXTAPOSITION% "You wouldn't understand."

Adam

"Go play in the lab."

mission

To create a simple, opinionated, and declarative framework for machine learning experiments. Our goal is to standardize AGI (Autonomous Generative Influence) via a decentralized ensemble of toy models on commodity hardware (i.e. your desktop computer).

features

  • Automated pipelines for development, training, and inference lifecycles across a variety of AI architectures.
  • A portable, autonomous AI agent used for distributed, unsupervised learning.
  • Easy to install, runs anywhere.
  • Integration with all of your platforms.
  • Local-first, offline APIs for text generation, image interrogation, and pattern mining.
  • Designed for deployment into auto-scaling, distributed architectures.
  • Swarm intelligence via sparse networks.
  • Rolling release cycles.
  • A hypothesis, a language, a story.
  • And so much more!

interface

tools

data

instructions

  • Install Git and Docker.
  • Clone this repo to a local directory: git clone --recurse-submodules https://github.com/0-5788719150923125/vtx.git path/to/my/directory

To clone faster, consider adding --depth 1 to clone only the most recent commit, reducing download time.

  • Use VSCode tasks with ctrl + shift + p, the controller.sh script (on Linux/Mac) or the controller.ps1 script (Windows) to control the AI.
  • Place configurations into a config.yml file. (example 1, example 2)
  • Put credentials into a .env file at the root of this project. (example .env file)
  • The "up" task is used to bring your lab online.
  • The "fetch" and "prepare" tasks are used for data retrieval and preparation.
  • The "train" task is used to train your models.
  • Checkout, fix, or update git submodules with init, repair, or update tasks.
  • Ask me questions.

examples

You will find many other examples about how to put the VTX into practice at: ./examples

For complete documentation, please visit this link.

troubleshooting

  • If, for some reason, a git submodule directory is empty, please try to cd into the module directory, and run this command: git submodule update --init
  • For Windows users, VSCode tasks will not work until you unblock the appropriate Powershell script: Unblock-File -Path .\controller.ps1

contact