Portuguese
xgboost
text-quality
portuguese
File size: 3,084 Bytes
04ab6bc
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
36134f6
 
 
 
 
 
 
 
04ab6bc
 
 
 
 
 
 
 
 
5cc595f
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
76
77
78
79
80
81
---
license: apache-2.0
datasets:
- TucanoBR/GigaVerbo-Text-Filter
language:
- pt
metrics:
- accuracy
library_name: xgboost
tags:
- text-quality
- portuguese
---
# XGBClassifier-text-filter

XGBClassifier-text-filter is a text-quality filter built on top of the [`xgboost`](https://xgboost.readthedocs.io/en/stable/) library. It uses the embeddings generated by [sentence-transformers/LaBSE](https://huggingface.co/sentence-transformers/LaBSE) as a feature vector.

This repository has the [source code](https://github.com/Nkluge-correa/Tucano) used to train this model.

## Usage

Here's an example of how to use the XGBClassifier-text-filter:

```python
from transformers import AutoTokenizer, AutoModel
from xgboost import XGBClassifier
import torch.nn.functional as F
import torch

def mean_pooling(model_output, attention_mask):
    token_embeddings = model_output[0]
    input_mask_expanded = attention_mask.unsqueeze(-1).expand(token_embeddings.size()).float()
    return torch.sum(token_embeddings * input_mask_expanded, 1) / torch.clamp(input_mask_expanded.sum(1), min=1e-9)

tokenizer = AutoTokenizer.from_pretrained("sentence-transformers/LaBSE")
embedding_model = AutoModel.from_pretrained("sentence-transformers/LaBSE")
device = ("cuda" if torch.cuda.is_available() else "cpu")
embedding_model.to(device)

bst = XGBClassifier({'device': device})
bst.load_model('/path/to/XGBClassifier-text-classifier.json')

def score_text(text, model):

    encoded_input = tokenizer(text, padding=True, truncation=True, return_tensors='pt').to(device)

    with torch.no_grad():
        model_output = embedding_model(**encoded_input)

    sentence_embedding = mean_pooling(model_output, encoded_input['attention_mask'])

    embedding = F.normalize(sentence_embedding, p=2, dim=1).numpy()
    score = model.predict(embedding)[0]

    return score

score_text("Os tucanos são aves que correspondem à família Ramphastidae, vivem nas florestas tropicais da América Central e América do Sul. A família inclui cinco gêneros e mais de quarenta espécies diferentes. Possuem bicos notavelmente grandes e coloridos, que possuem a função de termorregulação para as muitas espécies que passam muito tempo na copa da floresta exposta ao sol tropical quente.", bst)
```

## Cite as 🤗

```latex
@misc{correa2024tucanoadvancingneuraltext,
      title={{Tucano: Advancing Neural Text Generation for Portuguese}}, 
      author={Corr{\^e}a, Nicholas Kluge and Sen, Aniket and Falk, Sophia and Fatimah, Shiza},
      year={2024},
      eprint={2411.07854},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2411.07854}, 
}
```

## Aknowlegments

We gratefully acknowledge the granted access to the [Marvin cluster](https://www.hpc.uni-bonn.de/en/systems/marvin) hosted by [University of Bonn](https://www.uni-bonn.de/en) along with the support provided by its High Performance Computing \& Analytics Lab.

## License

XGBClassifier-text-filter is licensed under the Apache License, Version 2.0. For more details, see the [LICENSE](LICENSE) file.