QomSSLab/etghan_tagger_v1
This repository hosts an XLM-RoBERTa token-classification head trained.
Usage
from transformers import AutoTokenizer, AutoModelForTokenClassification, pipeline
model_id = "QomSSLab/etghan_tagger_v1"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForTokenClassification.from_pretrained(model_id)
tagger = pipeline("token-classification", model=model, tokenizer=tokenizer, aggregation_strategy="simple")
text = "مثال از یک ورودی فارسی"
for entity in tagger(text):
print(entity)
Labels
ABSENCEAPPEALABILITYAPPEAL_AUTHORITYAPPEAL_DEADLINECOURT_NAMEDEFENDANT_ADDRESSDEFENDANT_FATHERDEFENDANT_NAME_LEGALDEFENDANT_NAME_NATURALFINALITYJUDGE_NAMEJUDGE_POSITIONJUDGE_SIGNATUREJUDGMENT_DATELEGAL_CHARGELEGAL_REPRESENTATIVEOPLACE_OF_OFFENSEPLAINTIFF_ADDRESSPLAINTIFF_CLAIMPLAINTIFF_FATHERPLAINTIFF_NAME_LEGALPLAINTIFF_NAME_NATURALPRESENCESUBJECT_OF_CLAIMTIME_OF_OFFENSE
Metrics
Validation Metrics
- Precision: 0.7686
- Recall: 0.7340
- F1: 0.7509
- Accuracy: 0.9597
Per-label Breakdown
| Label | Precision | Recall | F1 | Support |
|---|---|---|---|---|
| ABSENCE | 1.0000 | 1.0000 | 1.0000 | 30 |
| APPEALABILITY | 0.9327 | 0.9652 | 0.9487 | 201 |
| APPEAL_AUTHORITY | 0.8130 | 0.9590 | 0.8800 | 195 |
| APPEAL_DEADLINE | 0.8860 | 0.9018 | 0.8938 | 112 |
| COURT_NAME | 0.8455 | 0.6596 | 0.7410 | 141 |
| DEFENDANT_ADDRESS | 0.9474 | 0.7714 | 0.8504 | 70 |
| DEFENDANT_FATHER | 1.0000 | 0.9737 | 0.9867 | 114 |
| DEFENDANT_NAME_LEGAL | 0.6190 | 0.9286 | 0.7429 | 14 |
| DEFENDANT_NAME_NATURAL | 0.9500 | 0.8711 | 0.9088 | 349 |
| FINALITY | 0.9500 | 0.9048 | 0.9268 | 21 |
| JUDGE_NAME | 0.7439 | 0.9683 | 0.8414 | 126 |
| JUDGE_POSITION | 0.8020 | 0.8804 | 0.8394 | 92 |
| JUDGE_SIGNATURE | 1.0000 | 1.0000 | 1.0000 | 0 |
| JUDGMENT_DATE | 0.6296 | 0.7969 | 0.7034 | 64 |
| LEGAL_CHARGE | 0.9709 | 0.9926 | 0.9816 | 404 |
| LEGAL_REPRESENTATIVE | 0.9080 | 0.9367 | 0.9221 | 158 |
| O | 0.9804 | 0.9769 | 0.9786 | 27566 |
| PLACE_OF_OFFENSE | 0.3784 | 0.5600 | 0.4516 | 25 |
| PLAINTIFF_ADDRESS | 0.9846 | 0.9846 | 0.9846 | 65 |
| PLAINTIFF_CLAIM | 0.6336 | 0.7318 | 0.6791 | 645 |
| PLAINTIFF_FATHER | 0.8033 | 0.9515 | 0.8711 | 103 |
| PLAINTIFF_NAME_LEGAL | 0.7436 | 1.0000 | 0.8529 | 29 |
| PLAINTIFF_NAME_NATURAL | 0.9413 | 0.9051 | 0.9229 | 390 |
| PRESENCE | 1.0000 | 0.9855 | 0.9927 | 69 |
| SUBJECT_OF_CLAIM | 0.7070 | 0.6205 | 0.6609 | 556 |
| TIME_OF_OFFENSE | 0.8909 | 0.7538 | 0.8167 | 65 |
- Downloads last month
- 198