---
license: apache-2.0
language:
- uz
pipeline_tag: text-generation
tags:
- grammar
- uzbek
---

# Uzbek Grammar Correction Model

Ushbu model o'zbek tilidagi matnlarning grammatik xatolarini tuzatish uchun mo'ljallangan.

## Muallif (Author)
Ushbu model **Davronov Rifqat** tomonidan yaratilgan va o'qitilgan.

## Model tavsifi
Model `rifkat/grammar` repozitoriyasidan olingan va `Seq2Seq` arxitekturasiga asoslangan. U matndagi imlo va grammatik xatolarni avtomatik tahrirlash imkonini beradi.

## Foydalanish (Usage)

Modelni ishlatish uchun quyidagi koddan foydalanishingiz mumkin:

```python
from transformers import Text2TextGenerationPipeline, AutoModelForSeq2SeqLM, AutoTokenizer

model_name = "rifkat/grammar"
model = AutoModelForSeq2SeqLM.from_pretrained(model_name)
tokenizer = AutoTokenizer.from_pretrained(model_name)

pipeline = Text2TextGenerationPipeline(model=model, tokenizer=tokenizer)

text = "SamDUning Kattaqorgon filialining foallar zalida shiftdagi qulagani beton plitaning holati bo‘yicha JKning 258-moddasi bilan jinoyat ishi qo‘zg‘atildi."

result = pipeline(
    text, 
    min_length=10,
    max_length=100,
    do_sample=False,
    early_stopping=False,
    num_beams=1,
    temperature=1.0,
    top_k=50,
    no_repeat_ngram_size=0,
    top_p=1.0
)

print(result[0]['generated_text'])
```

Inference parametrlari
min_length: 10

max_length: 100

num_beams: 1

do_sample: False