Using Flair Embeddings? about presidio-research HOT 2 CLOSED

Hi,
The flair-bert-embeddings.pt and flair-embeddings.pt files are the output of training a model using Flair. See https://github.com/microsoft/presidio-research/blob/master/models/flair_train.py for how we trained our Flair models.
You could also use pretrained Flair models, like 'ner', 'ner-fast' and 'ner-ontonotes-fast'. These could be loaded the same way:

from presidio_evaluator.flair_evaluator import FlairEvaluator

flair_evaluator = FlairEvaluator(model_path='ner')
evaluation_results = flair_evaluator.evaluate_all(DATASET)
scores = flair_evaluator.calculate_score(evaluation_results)

I'm not sure why the second example doesn't work. You could use the same pipeline in FlairTrainer to train a model on your data.

from presidio-research.

Thanks, running flair_train.py now. It seems this will generate the files needed.

from presidio-research.