seleznevchamp1/rubert-tiny2-NER-no_aug

---

language: ru tags: - NER - BERT - ruBERT - job-postings license: mit datasets: - custom: hh.ru metrics: - f1

ruBERT-tiny2 for NER on Russian Job Postings

Compact transformer model fine-tuned for extracting Skill and Knowledge entities from Russian job descriptions.

Base model: cointegrated/ruBert-tiny2
Training: fine-tuned on 250 manually annotated vacancies (no augmentation).
Labels: O, B-Skill, I-Skill, B-Knowledge, I-Knowledge
F1: 0.38

Usage

from transformers import AutoTokenizer, AutoModelForTokenClassification, pipeline
model = AutoModelForTokenClassification.from_pretrained("seleznevchamp1/rubert-tiny2-NER-no_aug")
tok = AutoTokenizer.from_pretrained("seleznevchamp1/rubert-tiny2-NER-no_aug")
nlp = pipeline("ner", model=model, tokenizer=tok)
print(nlp("Требуется знание Python и умение управлять командой"))