Transformers
PyTorch
Portuguese
File size: 1,705 Bytes
827e9ef
d6a2b22
 
 
 
827e9ef
 
d6a2b22
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
566b8c2
 
 
 
 
 
 
 
 
 
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
---
inference: false
language: pt
datasets:
- assin2
license: mit
---

# DeBERTinha XSmall for Recognizing Textual Entailment

### **Labels**:
* 0 : There is no entailment between premise and hypothesis.   
* 1 : There is entailment between premise and hypothesis.


## Full classification example

```python
from transformers import AutoModelForSequenceClassification, AutoTokenizer, AutoConfig
import numpy as np
import torch
from scipy.special import softmax

model_name = "sagui-nlp/debertinha-ptbr-xsmall-assin2-rte"
s1 = "Os homens estão cuidadosamente colocando as malas no porta-malas de um carro."
s2 = "Os homens estão colocando bagagens dentro do porta-malas de um carro."
model = AutoModelForSequenceClassification.from_pretrained(model_name)
tokenizer = AutoTokenizer.from_pretrained(model_name)
config = AutoConfig.from_pretrained(model_name)
model_input = tokenizer(*([s1], [s2]), padding=True, return_tensors="pt")
with torch.no_grad():
    output = model(**model_input)
    scores = output[0][0].detach().numpy()
    scores = softmax(scores)
    ranking = np.argsort(scores)
    ranking = ranking[::-1]
    for i in range(scores.shape[0]):
        l = config.id2label[ranking[i]]
        s = scores[ranking[i]]
        print(f"{i+1}) Label: {l} Score: {np.round(float(s), 4)}")
```

## Citation 

```
@misc{campiotti2023debertinha,
      title={DeBERTinha: A Multistep Approach to Adapt DebertaV3 XSmall for Brazilian Portuguese Natural Language Processing Task}, 
      author={Israel Campiotti and Matheus Rodrigues and Yuri Albuquerque and Rafael Azevedo and Alyson Andrade},
      year={2023},
      eprint={2309.16844},
      archivePrefix={arXiv},
      primaryClass={cs.CL}
}
```