flair/ner-french

This is the standard 4-class NER model for French that ships with Flair . The F1-Score: 90,61 (WikiNER) Predicts 4 tags: "George Washington est allé à Washington" and "Washington" The model is based on Flair embeddings and LSTM-CRF . The following Flair script was used to train this model: Flair (pip install flair) and run training with the model: F1/ner-french (default model) The paper is available in the issue of Flair and the Flair tracker is available here here . The model was trained using Flair with the standard French NER (,