Preparation
Basic Usage
nerblackbox provides a Python API with four main classes
Store, Dataset, Training and Model.
It is complemented by a CLI.
basic usage
from nerblackbox import Store, Dataset, Training, Model
nerblackbox --help
Store
As a mandatory first step, a store has to be created. The store is a directory that contains all the data (datasets, training configurations, results, model checkpoints) that nerblackbox needs access to. It is handled by the Store class:
create store
Store.create()
By default, the store is located at ./store and has the following subdirectories:
store/
└── datasets
└── training_configs
└── pretrained_models
└── results
If wanted, the path of the store can be adjusted (before creation) like this:
adjust store path
Store.set_path("<store_path>")