DistilCamemBERT-NER
| Property | Value |
|---|---|
| Parameter Count | 67.5M parameters |
| License | MIT |
| Language | French |
| Task | Named Entity Recognition |
What is distilcamembert-base-ner?
DistilCamemBERT-NER is a specialized French language model designed for Named Entity Recognition (NER), built upon the DistilCamemBERT architecture. This model represents a significant advancement in French NLP, offering twice the inference speed of traditional CamemBERT-based models while maintaining impressive accuracy levels.
Implementation Details
The model was trained on the wikiner_fr dataset, comprising approximately 170,000 labeled sentences. It achieves remarkable performance metrics, including a global F1 score of 98.18%, with particularly strong results in personality (96.82%) and location (93.82%) recognition.
- Optimized for production environments with reduced inference costs
- Supports five entity categories: PER, LOC, ORG, MISC, and O
- Mean inference time of 43.44ms on standard hardware
Core Capabilities
- High-accuracy entity recognition across all categories
- Efficient processing with 2x speed improvement
- Support for ONNX runtime optimization
- Comprehensive French language entity detection
Frequently Asked Questions
Q: What makes this model unique?
The model stands out for its optimal balance between performance and efficiency, achieving comparable accuracy to larger models while requiring only half the inference time. This makes it particularly suitable for production environments where resource optimization is crucial.
Q: What are the recommended use cases?
The model is ideal for French text analysis tasks requiring named entity recognition, including document processing, information extraction, and content analysis. It's particularly effective in scenarios requiring real-time processing or handling large volumes of text.
