Instructions to use creat89/NER_FEDA_Cyrillic2 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use creat89/NER_FEDA_Cyrillic2 with Transformers:
# Load model directly from transformers import AutoTokenizer, BERT_model_multidata tokenizer = AutoTokenizer.from_pretrained("creat89/NER_FEDA_Cyrillic2") model = BERT_model_multidata.from_pretrained("creat89/NER_FEDA_Cyrillic2", device_map="auto") - Notebooks
- Google Colab
- Kaggle
| results precision recall f1-score support | |
| DATE 0.9496 0.9496 0.9496 119 | |
| EVT 0.9907 0.9550 0.9725 333 | |
| GEOPOLIT 0.9713 0.9769 0.9741 346 | |
| LOC 0.9347 0.9444 0.9395 19979 | |
| MEDIA 0.9291 0.9704 0.9493 135 | |
| MISC 0.7381 0.7045 0.7209 88 | |
| ORG 0.8793 0.8718 0.8755 14102 | |
| PER 0.9529 0.9501 0.9515 14704 | |
| PRO 0.7942 0.8455 0.8190 356 | |
| micro avg 0.9238 0.9249 0.9244 50162 | |
| macro avg 0.9044 0.9076 0.9058 50162 | |
| weighted avg 0.9237 0.9249 0.9243 50162 | |