---
language: vi
datasets:
- cc100
tags:
- summarization

license: mit

widget:
- text: "Input text."
---

# ViT5-large Finetuned on `vietnews` Abstractive Summarization


State-of-the-art pretrained Transformer-based encoder-decoder model for Vietnamese.
[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/vit5-pretrained-text-to-text-transformer-for/abstractive-text-summarization-on-vietnews)](https://paperswithcode.com/sota/abstractive-text-summarization-on-vietnews?p=vit5-pretrained-text-to-text-transformer-for)


## How to use
For more details, do check out [our Github repo](https://github.com/vietai/ViT5) and [eval script](https://github.com/vietai/ViT5/blob/main/eval/Eval_vietnews_sum.ipynb). 

```python
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
​
tokenizer = AutoTokenizer.from_pretrained("VietAI/vit5-large-vietnews-summarization")  
model = AutoModelForSeq2SeqLM.from_pretrained("VietAI/vit5-large-vietnews-summarization")
model.cuda()
​
sentence = "Input text"
text =  "vietnews: " + sentence + " </s>"
encoding = tokenizer(text, return_tensors="pt")
input_ids, attention_masks = encoding["input_ids"].to("cuda"), encoding["attention_mask"].to("cuda")
outputs = model.generate(
    input_ids=input_ids, attention_mask=attention_masks,
    max_length=128,
    early_stopping=True
)
for output in outputs:
    line = tokenizer.decode(output, skip_special_tokens=True, clean_up_tokenization_spaces=True)
    print(line)
```