https://github.com/archinetai/emergent-pytorch

A new invertible and distributed training strategy for PyTorch.
https://github.com/archinetai/emergent-pytorch

Last synced: about 2 months ago
JSON representation

A new invertible and distributed training strategy for PyTorch.

• Host: GitHub
• URL: https://github.com/archinetai/emergent-pytorch
• Owner: archinetai
• License: mit
• Created: 2022-06-08T10:01:53.000Z (almost 3 years ago)
• Default Branch: main
• Last Pushed: 2022-06-11T12:39:22.000Z (almost 3 years ago)
• Last Synced: 2025-03-16T06:36:20.513Z (3 months ago)
• Language: Python
• Homepage:
• Size: 34.2 KB
• Stars: 1
• Watchers: 3
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        


Something interesting

## Install

```bash

$ pip install emergent-pytorch

```

[![PyPI - Python Version](https://img.shields.io/pypi/v/emergent-pytorch?style=flat&colorA=0f0f0f&colorB=0f0f0f)](https://pypi.org/project/emergent-pytorch/)

## Usage

TODO

## TODO 

- [ ] add vector quantization unit 

- [ ] test with recursion/repetition like diffusion