Kito Documentation
Kito is a PyTorch-based deep learning framework thought for researchers. Kito offers effortless training of deep learning models, by automatically handling training loops, optimization, callbacks, distributed training, and more.
Note
This documentation is under active development.
Quick Start
Installation:
pip install pytorch-kito
Basic usage:
from kito import Engine, KitoModule
# Define your model
class MyModel(KitoModule):
def build_inner_model(self):
self.model = nn.Sequential(...)
# Train
module = MyModel("model_name", device, config)
engine = Engine(module, config)
engine.fit(train_loader, val_loader)
Contents
User Guide