Restaurant Search NER with DistilBERT
Fine-tuned DistilBERT for Named Entity Recognition in restaurant search queries.
Performance
- F1 Score: 79.85%
- Precision: 78.30%
- Recall: 81.46%
- Accuracy: 91.96%
Training Results
| Epoch | Training Loss | Validation Loss | Precision | Recall | F1 | Accuracy |
|---|---|---|---|---|---|---|
| 1 | 0.633 | 0.306 | 0.731 | 0.794 | 0.762 | 0.909 |
| 2 | 0.249 | 0.286 | 0.770 | 0.809 | 0.789 | 0.916 |
| 3 | 0.203 | 0.284 | 0.783 | 0.815 | 0.799 | 0.920 |
Entity Types
Rating, Location, Amenity, Cuisine, Dish, Hours, Price, Restaurant_Name
Usage
from transformers import AutoTokenizer, AutoModelForTokenClassification
tokenizer = AutoTokenizer.from_pretrained('Aakash22134/Aakash22134Restarant-Search-Distaled_Bert')
model = AutoModelForTokenClassification.from_pretrained('Aakash22134/Aakash22134Restarant-Search-Distaled_Bert')