File size: 2,588 Bytes
eab98b6
 
 
4a65b37
eab98b6
 
 
 
4a65b37
eab98b6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
---
language: vi
datasets:
- nyamuda/samsum
tags:
- summarization
license: mit
widget:
- text: ViFortuneAI.
---

# ViT5-Base Finetuned on `vietnews` Abstractive Summarization (No prefix needed)


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

# Load model và tokenizer
model_name = "ViFortune-AI/ViT5Summer"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForSeq2SeqLM.from_pretrained(model_name)
model.cuda()

# DỮ LIỆU ĐẦU VÀO CỦA BẠN: nguyên văn hội thoại (giống trong dataset)
sentence = "Bạn đã thanh toán cho cà phê không?>> Hmm... tôi nghĩ không phải là vậy, nhưng nó cũng không sao, tôi sẽ thanh toán anh ta mai nhé."

# ✅ KHÔNG thêm "summarize:", KHÔNG thêm "</s>"
encoding = tokenizer(
    sentence,
    return_tensors="pt",
    max_length=512,
    truncation=True,
    padding=False  # hoặc "max_length" nếu muốn
)

input_ids = encoding["input_ids"].to("cuda")
attention_mask = encoding["attention_mask"].to("cuda")

# Generate
outputs = model.generate(
    input_ids=input_ids,
    attention_mask=attention_mask,
    max_length=256,
    min_length=10,
    num_beams=4,
    early_stopping=True,
    no_repeat_ngram_size=2,
    length_penalty=1.0
)

# Decode
for output in outputs:
    summary = tokenizer.decode(output, skip_special_tokens=True, clean_up_tokenization_spaces=True)
    print("Tóm tắt:", summary)
```

## Citation
```
@inproceedings{phan-etal-2022-vit5,
    title = "{V}i{T}5: Pretrained Text-to-Text Transformer for {V}ietnamese Language Generation",
    author = "Phan, Long and Tran, Hieu and Nguyen, Hieu and Trinh, Trieu H.",
    booktitle = "Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies: Student Research Workshop",
    year = "2022",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2022.naacl-srw.18",
    pages = "136--142",
}
```