https://github.com/mokkemeguru/tfgenzoo

Library about construction helper for Generative models e.g. Flow-based Model with Tensorflow 2.x.
https://github.com/mokkemeguru/tfgenzoo

flow flow-based flow-based-model tensorflow

Last synced: 3 months ago
JSON representation

Library about construction helper for Generative models e.g. Flow-based Model with Tensorflow 2.x.

• Host: GitHub
• URL: https://github.com/mokkemeguru/tfgenzoo
• Owner: MokkeMeguru
• Created: 2019-12-30T02:55:40.000Z (over 6 years ago)
• Default Branch: master
• Last Pushed: 2023-02-16T01:33:31.000Z (over 3 years ago)
• Last Synced: 2025-07-01T12:49:59.900Z (11 months ago)
• Topics: flow, flow-based, flow-based-model, tensorflow
• Language: Python
• Homepage: https://mokkemeguru.github.io/TFGENZOO/
• Size: 21.6 MB
• Stars: 12
• Watchers: 4
• Forks: 2
• Open Issues: 24
• Metadata Files:
  • Readme: readme.md

Awesome Lists containing this project

README

          
# TFGENZOO (Generative Model x Tensorflow 2.x)

![img](https://github.com/MokkeMeguru/TFGENZOO/workflows/tensorflow%20test/badge.svg?branch=master)

![img](https://img.shields.io/badge/License-MIT-yellow.svg)

![img](https://img.shields.io/badge/python-3.7-blue.svg)

![img](https://img.shields.io/badge/tensorflow-%3E%3D2.2.0-brightgreen.svg)

![img](https://badge.fury.io/py/TFGENZOO.svg)

# What’s this repository?

This is a repository for some researcher to build some Generative models using Tensorflow 2.x.

I NEED YOUR HELP(please let me know about formula, implementation and anything you worried)

![img](https://raw.githubusercontent.com/MokkeMeguru/TFGENZOO/master/docs/tfgenzoo_header.png)

# Zen of this repository

    We don't want to need flexible architectures.

    We need strict definitions for shapes, parameters, and formulas.

    We should Implement correct codes with well-documented(tested).

# How to use?

## By Install

- pipenv

      pipenv install TFGENZOO==1.2.5

- pip

      pip install TFGENZOO==1.2.5

## Source build for development

1.  clone this repository (If you want to do it, I will push this repository to PYPI)

2.  build this repository `docker-compose build`

3.  run the environment `sh run_script.sh`

4.  connect it via VSCode or Emacs or vi or anything.

# Examples

- [TFGENZOO_EXAMPLE](https://github.com/MokkeMeguru/TFGENZOO_EXAMPLE)

- Simple Tutorials

  - [What is the invertible layer](./tutorials/01_What_is_the_invertible_layer.ipynb)

    The tutorial about Flow-based Model

  - [conditional flow-based model](./tutorials/02_conditional_flow-based_model.ipynb)

    How to add conditional input into Flow-based Model for the image generation.

# Documents

# Roadmap

- [x] Flow-based Model Architecture (RealNVP, Glow)

- [ ] i-ResNet Model Architecture (i-ResNet, i-RevNet)

- [ ] GANs Model Architecture (GANs)

# Remarkable Backlog

Whole backlog is [here](https://github.com/MokkeMeguru/TFGENZOO/wiki/Backlog)

## News [2020/6/16]

New training results [Oxford-flower102](https://www.tensorflow.org/datasets/catalog/oxford_flowers102) with only 8 hours! (Quadro P6000 x 1)

data

NLL(test)

epoch

pretrained

Oxford-flower102

4.590211391448975

1024

---

![img](https://raw.githubusercontent.com/MokkeMeguru/TFGENZOO/master/docs/oxford.png)

see more detail, you can see [my internship’s report](https://docs.google.com/presentation/d/12z6MZizIsytLxUb2ly7vYorFiKruIGZ2ckQ0-By4b6s/edit?usp=sharing) (Japanese only, if you need translated version, please contact me.)

## News [2020/7/11]

Add some tutorial into `./tutorial`

## News [2021/3/30]

I wrote the master's paper about japanese text style transfer.  "AutoEncoder に基づく半教師あり和文スタ

イル変換"

https://drive.google.com/file/d/1KtkLZi6PUvL7msAqbg_KRdEC0pmmpbhf/view?usp=sharing

# Contact

MokkeMeguru ([@MokkeMeguru](https://twitter.com/MeguruMokke)): DM or Mention Please (in Any language).