2018-04-05

A best practice for tensorflow project template architecture.

image.png

Fllder Structure

├──  base
│   ├── base_model.py   - this file contains the abstract class of the model.
│   └── base_train.py   - this file contains the abstract class of the trainer.
│
│
├── model               - this folder contains any model of your project.
│   └── example_model.py
│
│
├── trainer             - this folder contains trainers of your project.
│   └── example_trainer.py
│   
├──  mains              - here's the main(s) of your project (you may need more than one main).
│    └── example_main.py  - here's an example of main that is responsible for the whole pipeline.

│  
├──  data _loader  
│    └── data_generator.py  - here's the data_generator that is responsible for all data handling.
│ 
└── utils
     ├── logger.py
     └── any_other_utils_you_need