French model trained to recognize named entities from archaeological reports.
| Feature | Description |
|---|---|
| Name | fr_arches_ner |
| Version | 0.0.0 |
| spaCy | >=3.6.1,<3.7.0 |
| Default Pipeline | tok2vec, ner, entity_punctuation_removal |
| Components | tok2vec, ner, entity_punctuation_removal |
| Vectors | 500000 keys, 500000 unique vectors (300 dimensions) |
| Sources | 21 archaeological reports from the Inrap. |
| License | cc-by-nc 2.0 |
| Author | Institut national de recherches archéologiques préventives |
Label Scheme
View label scheme (15 labels for 1 components)
| Component | Labels |
|---|---|
ner |
CHRONOLOGIE, DECOR, EDIFICE, ESPECE, GPE, ID, LIEUDIT_SITE, LOC, MATERIAU, MOBILIER, ORG, PERSONNE, PEUPLE_CULTURE, STRUCTURE, TECHNIQUE_STYLE |
Accuracy
| Type | Score |
|---|---|
ENTS_F |
69.62 |
ENTS_P |
67.78 |
ENTS_R |
71.57 |
TOK2VEC_LOSS |
63436.09 |
NER_LOSS |
246059.83 |
- Downloads last month
- 6
Inference Providers
NEW
This model isn't deployed by any Inference Provider.
🙋
Ask for provider support
Space using a-menu/fr_arches_ner 1
Evaluation results
- NER Precisionself-reported0.678
- NER Recallself-reported0.716
- NER F Scoreself-reported0.696