INNER CODE UNIT · Python

_train_model

dragondrop-cloud/cloud-concierge · nlpengine/main.py:192

def _train_model(
    nlp: spacy.Language,
    textcat_multilabel_model,
    training_data: List[Example],
    evaluation_data: List[Example],
) -> dict:
    """
    Divides the training data into sets of 10 and updates the model using them.

    After each training batch, output the loss, as well as evaluate the accuracy on the evaluation
    data set set.

    Outputs a dict of the final loss and recall, precision and f1-score on the evaluation data set.
    """
    optimizer = nlp.create_optimizer()
    training_batches = np.array_split(training_data, 10)

    for i, batch in enumerate(training_batches):

View source record →