A framework for quick start to write codes for Pytorch Coding.

The goal of this project is to simplify the coding for deep learning of pytorch, and let developers more focus on data processing and model designing.

Meanwhile, this is a canonical deep learning code framework for Pytorch.

version 1.0.

python >= 3.7

pytorch == 1.13.1+cu117 (An option. There is no special requirements for pytorch or other packages)

|-- PtCoding
    |-- main.py
    |-- README.md
    |-- data
    |-- logs
    |-- src
        |-- parse_args.py
        |-- train.py
        |-- test.py
        |-- utils.py
        |-- data_load
            |-- data_loader.py
            |-- MyData.py
        |-- model
            |-- model_process.py
            |-- MyModel

• Rewrite ./src/data_load/MyData.py to define your own data.

• Rewrite ./src/data_load/data_loader.py to load your data into class [Dataset].

• Rerite ./src/model/MyModel.py to define your own model.

• Rewrite ./src/model/model_process.py to define the process of train, valid, and test in an epoch.

• Rewrite ./train.py and ./test.py to define the train process (train and valid) and test process (test) in an epoch.

• Rewrite ./src/parse_args.py to set the parameters.

• Rewrite ./main.py to define the whole process.

python main.py