API Reference

class terrier.apps.Terrier(**kwargs)

Transposable Element Repeat Result classifIER

bibliography = Method(func=<function TorchApp.bibliography>, methods_to_call=[], main=False, tool=False, flag=True, shortcut='', signature_ready=False)

bibtex = Method(func=<function TorchApp.bibtex>, methods_to_call=[], main=False, tool=False, flag=True, shortcut='', signature_ready=False)

cache_dir() → Path

Returns a directory under the user’s cache folder for storing downloads, e.g. ~/.cache/torchapps/<AppName>/.

callbacks = Method(func=<function TorchApp.callbacks>, methods_to_call=('extra_callbacks',), main=False, tool=False, flag=False, shortcut='', signature_ready=False)

checkpoint(checkpoint: Path = None) → str

Returns a path (local or remote) to a checkpoint. By default, you must override or pass –checkpoint.

checkpoint_callback = Method(func=<function TorchApp.checkpoint_callback>, methods_to_call=('monitor',), main=False, tool=False, flag=False, shortcut='', signature_ready=False)

comparison_plot = Method(func=<function Terrier.comparison_plot>, methods_to_call=[], main=False, tool=True, flag=False, shortcut='', signature_ready=False)

data = Method(func=<function Terrier.data>, methods_to_call=('super',), main=False, tool=False, flag=False, shortcut='', signature_ready=False)

extra_callbacks = Method(func=<function TorchApp.extra_callbacks>, methods_to_call=[], main=False, tool=False, flag=False, shortcut='', signature_ready=False)

extra_hyperparameters = Method(func=<function Corgi.extra_hyperparameters>, methods_to_call=[], main=False, tool=False, flag=False, shortcut='', signature_ready=False)

lightning_module = Method(func=<function TorchApp.lightning_module>, methods_to_call=('model', 'loss_function', 'extra_hyperparameters', 'input_count', 'metrics', 'module_class'), main=False, tool=False, flag=False, shortcut='', signature_ready=False)

load_checkpoint = Method(func=<function TorchApp.load_checkpoint>, methods_to_call=('checkpoint',), main=False, tool=False, flag=False, shortcut='', signature_ready=False)

model = Method(func=<function Corgi.model>, methods_to_call=('module_class',), main=False, tool=False, flag=False, shortcut='', signature_ready=False)

node_to_str(node: SoftmaxNode) → str

Converts the node to a string

one_batch = Method(func=<function TorchApp.one_batch>, methods_to_call=('setup_and_data',), main=False, tool=False, flag=False, shortcut='', signature_ready=False)

one_batch_loss = Method(func=<function TorchApp.one_batch_loss>, methods_to_call=('one_batch', 'lightning_module'), main=False, tool=True, flag=False, shortcut='', signature_ready=False)

one_batch_output_size = Method(func=<function TorchApp.one_batch_output_size>, methods_to_call=('setup_and_data', 'lightning_module'), main=False, tool=True, flag=False, shortcut='', signature_ready=False)

one_batch_size = Method(func=<function TorchApp.one_batch_size>, methods_to_call=('one_batch',), main=False, tool=True, flag=False, shortcut='', signature_ready=False)

package_name() → str

Returns the name of the package.

predict = Method(func=<function TorchApp.predict>, methods_to_call=('load_checkpoint', 'prediction_trainer', 'prediction_dataloader', 'output_results'), main=True, tool=True, flag=False, shortcut='', signature_ready=False)

prediction_trainer = Method(func=<function TorchApp.prediction_trainer>, methods_to_call=[], main=False, tool=False, flag=False, shortcut='', signature_ready=False)

process_location(location: str, reload: bool = False) → Path

Turn a URL or filesystem path into a Path that points to a local file. If it’s a URL, download (and cache) via cached_download().

profiler = Method(func=<function TorchApp.profiler>, methods_to_call=[], main=False, tool=False, flag=False, shortcut='', signature_ready=False)

setup_and_data = Method(func=<function TorchApp.setup_and_data>, methods_to_call=('setup', 'data', 'validation_dataloader'), main=False, tool=False, flag=False, shortcut='', signature_ready=False)

train = Method(func=<function TorchApp.train>, methods_to_call=('setup_and_data', 'validation_dataloader', 'lightning_module', 'trainer'), main=False, tool=True, flag=False, shortcut='', signature_ready=False)

trainer = Method(func=<function TorchApp.trainer>, methods_to_call=('callbacks', 'profiler', 'project_name'), main=False, tool=False, flag=False, shortcut='', signature_ready=False)

tune = Method(func=<function TorchApp.tune>, methods_to_call=('train', 'project_name'), main=False, tool=True, flag=False, shortcut='', signature_ready=False)

tuning_params()

Inspect the signature of self.tune(…) and return a dict of only those Param() arguments where Param.tune == True. (So that an external tuner can see which hyperparameters are tunable.)

validate = Method(func=<function TorchApp.validate>, methods_to_call=('setup_and_data', 'load_checkpoint', 'trainer'), main=False, tool=True, flag=False, shortcut='', signature_ready=False)

version = Method(func=<function TorchApp.version>, methods_to_call=(), main=False, tool=False, flag=True, shortcut='-v', signature_ready=False)