Fairseq
Catalan
Spanish
File size: 8,734 Bytes
4ab520c
 
 
3af3027
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3d4bce7
3af3027
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
51076bc
3af3027
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
f357c4e
3af3027
 
 
e205871
3af3027
 
 
f357c4e
 
 
 
 
 
 
78fb9c7
 
3af3027
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
---
license: cc-by-4.0
---
## Aina Project's Catalan-Spanish  machine translation model

## Table of Contents
- [Model Description](#model-description)
- [Intended Uses and Limitations](#intended-use)
- [How to Use](#how-to-use)
- [Training](#training)
  - [Training data](#training-data)
  - [Training procedure](#training-procedure)
    - [Data Preparation](#data-preparation)
    - [Tokenization](#tokenization)
    - [Hyperparameters](#hyperparameters)
- [Evaluation](#evaluation)
   - [Variable and Metrics](#variable-and-metrics)
   - [Evaluation Results](#evaluation-results)
- [Additional Information](#additional-information)
  - [Author](#author)
  - [Contact Information](#contact-information)
  - [Copyright](#copyright)
  - [Licensing Information](#licensing-information)
  - [Funding](#funding)
  - [Disclaimer](#disclaimer)
  
## Model description

This model was trained from scratch using the [Fairseq toolkit](https://fairseq.readthedocs.io/en/latest/) on a combination of Catalan-Spanish datasets, up to 92 million sentences. Additionally, the model is evaluated on several public datasecomprising 5 different domains (general, adminstrative, technology, biomedical, and news).  

## Intended uses and limitations

You can use this model for machine translation from Catalan to Spanish. 

## How to use

### Usage
Required libraries:

```bash
pip install ctranslate2 pyonmttok
```

Translate a sentence using python 
```python
import ctranslate2
import pyonmttok
from huggingface_hub import snapshot_download
model_dir = snapshot_download(repo_id="projecte-aina/mt-aina-ca-es", revision="main")

tokenizer=pyonmttok.Tokenizer(mode="none", sp_model_path = model_dir + "/spm.model")
tokenized=tokenizer.tokenize("Benvingut al projecte Aina!")

translator = ctranslate2.Translator(model_dir)
translated = translator.translate_batch([tokenized[0]])
print(tokenizer.detokenize(translated[0][0]['tokens']))
```

## Training

### Training data

The was trained on a combination of the following datasets:

| Dataset           | Sentences      | Tokens            |
|-------------------|----------------|-------------------|
| DOCG v2           | 8.472.786      | 188.929.206       |
| El Periodico      | 6.483.106      | 145.591.906       |
| EuroParl          | 1.876.669      | 49.212.670        |
| WikiMatrix        | 1.421.077      | 34.902.039        |
| Wikimedia         | 335.955        | 8.682.025         |
| QED               | 71.867         | 1.079.705         |
| TED2020 v1        | 52.177         | 836.882           |
| CCMatrix v1       | 56.103.820     | 1.064.182.320     |
| MultiCCAligned v1 | 2.433.418      | 48.294.144        |
| ParaCrawl         | 15.327.808     | 334.199.408       |
| **Total**         | **92.578.683** | **1.875.910.305** |

### Training procedure

### Data preparation

 All datasets are concatenated and filtered using the [mBERT Gencata parallel filter](https://huggingface.co/projecte-aina/mbert-base-gencata) and cleaned using the clean-corpus-n.pl script from [moses](https://github.com/moses-smt/mosesdecoder), allowing sentences between 5 and 150 words.

 Before training, the punctuation is normalized using a modified version of the join-single-file.py script from [SoftCatalà](https://github.com/Softcatala/nmt-models/blob/master/data-processing-tools/join-single-file.py)


#### Tokenization

 All data is tokenized using sentencepiece, with 50 thousand token sentencepiece model  learned from the combination of all filtered training data. This model is included.  

#### Hyperparameters

The model is based on the Transformer-XLarge proposed by [Subramanian et al.](https://aclanthology.org/2021.wmt-1.18.pdf)
The following hyperparamenters were set on the Fairseq toolkit:

| Hyperparameter                     | Value                            |
|------------------------------------|----------------------------------|
| Architecture                       | transformer_vaswani_wmt_en_de_bi |
| Embedding size                     | 1024                             |
| Feedforward size                   | 4096                             |
| Number of heads                    | 16                               |
| Encoder layers                     | 24                               |
| Decoder layers                     | 6                                |
| Normalize before attention         | True                             |
| --share-decoder-input-output-embed | True                             |
| --share-all-embeddings             | True                             |
| Effective batch size               | 96.000                           |
| Optimizer                          | adam                             |
| Adam betas                         | (0.9, 0.980)                     |
| Clip norm                          | 0.0                              |
| Learning rate                      | 1e-3                             |
| Lr. schedurer                      | inverse sqrt                     |
| Warmup updates                     | 4000                             |
| Dropout                            | 0.1                              |
| Label smoothing                    | 0.1                              |

The model was trained using shards of 10 million sentences, for a total of 13.000 updates. Weights were saved every 1000 updates and reported results are the average of the last 6 checkpoints. 

## Evaluation

### Variable and metrics

We use the BLEU score for evaluation on test sets: [Flores-101](https://github.com/facebookresearch/flores), [TaCon](https://elrc-share.eu/repository/browse/tacon-spanish-constitution-mt-test-set/84a96138b98611ec9c1a00155d02670628f3e6857b0f422abd82abc3795ec8c2/), [United Nations](https://zenodo.org/record/3888414#.Y33-_tLMIW0), [Cybersecurity](https://elrc-share.eu/repository/browse/cyber-mt-test-set/2bd93faab98c11ec9c1a00155d026706b96a490ed3e140f0a29a80a08c46e91e/), [wmt19 biomedical test set](), [wmt13 news test set](https://elrc-share.eu/repository/browse/catalan-wmt2013-machine-translation-shared-task-test-set/84a96139b98611ec9c1a00155d0267061a0aa1b62e2248e89aab4952f3c230fc/)

### Evaluation results

Below are the evaluation results on the machine translation from Catalan to Spanish compared to [Softcatalà](https://www.softcatala.org/) and [Google Translate](https://translate.google.es/?hl=es):

| Test set             | SoftCatalà | Google Translate | mt-aina-ca-es |
|----------------------|------------|------------------|---------------|
| Spanish Constitution | 70,7       | **77,1**         | 75,5          |
| United Nations       | 78,1       | 84,3             | **86,3**      |
| Flores 101 dev       | 23,5       | 24               | **24,1**      |
| Flores 101 devtest   | 24,1       | 24,2             | **24,4**      |
| Cybersecurity        | 67,3       | **76,9**         | 75,1          |
| wmt 19 biomedical    | 60,4       | 62,7             | **63,0**      |
| wmt 13 news          | 22,5       | 23,1             | **23,4**      |
| aina_aapp_ca-es      | 80,9       | 81,4             | **82,8**      |
| Average              | 53,4       | 56,7             | **56,8**      |


## Additional information

### Author
Text Mining Unit (TeMU) at the Barcelona Supercomputing Center ([email protected])

### Contact information
For further information, send an email to [email protected]

### Copyright
Copyright (c) 2022 Text Mining Unit at Barcelona Supercomputing Center 


### Licensing Information
[Apache License, Version 2.0](https://www.apache.org/licenses/LICENSE-2.0)

### Funding
This work was funded by the [Departament de la Vicepresidència i de Polítiques Digitals i Territori de la Generalitat de Catalunya](https://politiquesdigitals.gencat.cat/ca/inici/index.html#googtrans(ca|en) within the framework of [Projecte AINA](https://politiquesdigitals.gencat.cat/ca/economia/catalonia-ai/aina).


## Disclaimer
<details>
<summary>Click to expand</summary>

The models published in this repository are intended for a generalist purpose and are available to third parties. These models may have bias and/or any other undesirable distortions.

When third parties, deploy or provide systems and/or services to other parties using any of these models (or using systems based on these models) or become users of the models, they should note that it is their responsibility to mitigate the risks arising from their use and, in any event, to comply with applicable regulations, including regulations regarding the use of Artificial Intelligence.

In no event shall the owner and creator of the models (BSC – Barcelona Supercomputing Center) be liable for any results arising from the use made by third parties of these models.