---

language:
- ru
tags:
- summarization
license: apache-2.0

---

# RuT5TelegramHeadlines

## Model description

Based on [rut5-base](https://huggingface.co/cointegrated/rut5-base) model

## Intended uses & limitations

#### How to use

```python
from transformers import AutoTokenizer, T5ForConditionalGeneration

model_name = "IlyaGusev/rut5_telegram_headlines"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = T5ForConditionalGeneration.from_pretrained(model_name)

article_text = "..."

input_ids = tokenizer(
    [article_text],
    max_length=600,
    add_special_tokens=True,
    padding="max_length",
    truncation=True,
    return_tensors="pt"
)["input_ids"]

output_ids = model.generate(
    input_ids=input_ids,
    no_repeat_ngram_size=4
)[0]

headline = tokenizer.decode(output_ids, skip_special_tokens=True)
print(headline)
```

## Training data

- Dataset: [ru_all_split.tar.gz](https://www.dropbox.com/s/ykqk49a8avlmnaf/ru_all_split.tar.gz)

## Training procedure

- Training script: [train.py](https://github.com/IlyaGusev/summarus/blob/master/external/hf_scripts/train.py)