CamemBERT-bio-gliner-v0.1 : Zero-shot French Biomedical Named Entity Recognition

CamemBERT-bio-gliner is a Named Entity Recognition (NER) model capable of identifying any french biomedical entity type using a BERT-like encoder. It provides a practical alternative to traditional NER models, which are limited to predefined entities, and Large Language Models (LLMs) that, despite their flexibility, are costly and large for resource-constrained scenarios. CamemBERT-bio is used as a backbone. This model is based on the fantastic work of Urchade Zaratiana on the GLiNER architecture.

Important

This is the v0.1 of the CamemBERT-bio-gliner model. There might be a few quirks or unexpected predictions. So, if you notice anything off or have suggestions for improvements, we'd really appreciate hearing from you!

Installation

To use this model, you must install the GLiNER Python library:

!pip install gliner

Usage

Once you've downloaded the GLiNER library, you can import the GLiNER class. You can then load this model using GLiNER.from_pretrained and predict entities with predict_entities.

from gliner import GLiNER

model = GLiNER.from_pretrained("almanach/camembert-bio-gliner-v0.1")

text = """
Mme A.P. âgée de 52 ans, non tabagique, ayant un diabète de type 2 a été hospitalisée pour une pneumopathie infectieuse. Cette patiente présentait depuis 2 ans des infections respiratoires traités en ambulatoire. L’examen physique a trouvé une fièvre à 38ºc et un foyer de râles crépitants de la base pulmonaire droite.
"""

labels = ["Âge", "Patient", "Maladie", "Symptômes"]

entities = model.predict_entities(text, labels, threshold=0.5, flat_ner=True)

for entity in entities:
    print(entity["text"], "=>", entity["label"])

Mme A.P. => Patient
52 ans => Âge
pneumopathie infectieuse => Maladie
infections respiratoires => Maladie
fièvre => Symptômes
râles crépitants => Symptômes

Links

• Model: https://huggingface.co/almanach/camembert-bio-gliner-v0.1
• Backbone model: https://huggingface.co/almanach/camembert-bio-base
• GLiNER library: https://github.com/urchade/GLiNER
• Developed by: Rian Touchent, Eric Villemonte de La Clergerie
• Logo by: Alix Chagué, Rian Touchent
• License: MIT