PyTorch tools

Some tools to help with PyTorch.

Example

Currently this repo revolves around the Experiment class which wraps a PyTorch model with a Keras inspired API.

First, create an instance of a PyTorch model, then wrap the model in an experiement

model = Model()
exp = Experiment(model)

Note: Experiment supports Visdom logging via Experiment(model, viz_logging=True). A visdom server must be running (python -m visdom.server)

Then create a criterion (loss function) and an optimizer and compile the model.

criterion = nn.CrossEntropyLoss()
optimizer = torch.optim.Adam(model.parameters())

exp.compile(optimizer, criterion)

The model can then be fit and evaluated

exp.fit(train_loader, n_epoch=10)
exp.evaluate(test_loader)

note that train_loader and test_loader are PyTorch Dataloaders

.fit also supports validation sets

exp.fit(train_loader, n_epoch=10, valid_loader=test_loader, valid_freq=2)