T5_Paraphrase_Paws / README.md
1
---
2
language: "en"
3
tags:
4
- paraphrase-generation
5
- text-generation
6
- Conditional Generation
7
inference: false
8
---
9
10
# Paraphrase-Generation
11
12
## Model description
13
14
T5 Model for generating paraphrases of english sentences. Trained on the [Google PAWS](https://github.com/google-research-datasets/paws) dataset.
15
16
## How to use
17
18
PyTorch and TF models available
19
20
```python
21
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
22
23
tokenizer = AutoTokenizer.from_pretrained("Vamsi/T5_Paraphrase_Paws")  
24
model = AutoModelForSeq2SeqLM.from_pretrained("Vamsi/T5_Paraphrase_Paws")
25
26
sentence = "This is something which i cannot understand at all"
27
28
text =  "paraphrase: " + sentence + " </s>"
29
30
encoding = tokenizer.encode_plus(text,pad_to_max_length=True, return_tensors="pt")
31
input_ids, attention_masks = encoding["input_ids"].to("cuda"), encoding["attention_mask"].to("cuda")
32
33
34
outputs = model.generate(
35
    input_ids=input_ids, attention_mask=attention_masks,
36
    max_length=256,
37
    do_sample=True,
38
    top_k=120,
39
    top_p=0.95,
40
    early_stopping=True,
41
    num_return_sequences=5
42
)
43
44
for output in outputs:
45
    line = tokenizer.decode(output, skip_special_tokens=True,clean_up_tokenization_spaces=True)
46
    print(line)
47
48
49
```
50
51
For more reference on training your own T5 model or using this model, do check out [Paraphrase Generation](https://github.com/Vamsi995/Paraphrase-Generator).
52