update readme
Browse files
README.md
CHANGED
@@ -1,59 +1,36 @@
|
|
1 |
---
|
2 |
-
|
3 |
language:
|
4 |
-
|
5 |
- ca
|
6 |
-
|
7 |
license: apache-2.0
|
8 |
-
|
9 |
tags:
|
10 |
-
|
11 |
- "catalan"
|
12 |
-
|
13 |
-
- "part of speech tagging"
|
14 |
-
|
15 |
-
- "pos"
|
16 |
-
|
17 |
-
- "CaText"
|
18 |
-
|
19 |
-
- "Catalan Textual Corpus"
|
20 |
-
|
21 |
datasets:
|
22 |
-
|
23 |
-
- "
|
24 |
-
|
25 |
metrics:
|
26 |
-
|
27 |
-
-
|
28 |
-
|
29 |
-
inference:
|
30 |
-
parameters:
|
31 |
-
aggregation_strategy: "first"
|
32 |
-
|
33 |
-
model-index:
|
34 |
-
- name: roberta-base-ca-cased-pos
|
35 |
-
results:
|
36 |
-
- task:
|
37 |
-
type: token-classification
|
38 |
-
dataset:
|
39 |
-
type: universal_dependencies
|
40 |
-
name: Ancora-ca-POS
|
41 |
-
metrics:
|
42 |
-
- name: F1
|
43 |
-
type: f1
|
44 |
-
value: 0.9893832385244624
|
45 |
-
|
46 |
widget:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
47 |
|
48 |
-
- text: "Em dic Lluïsa i visc a Santa Maria del Camí."
|
49 |
-
|
50 |
-
- text: "L'Aina, la Berta i la Norma són molt amigues."
|
51 |
-
|
52 |
-
- text: "El Martí llegeix el Cavall Fort."
|
53 |
|
54 |
---
|
55 |
|
56 |
-
# Catalan BERTa (roberta-base-ca) finetuned for
|
57 |
|
58 |
## Table of Contents
|
59 |
- [Model Description](#model-description)
|
@@ -72,11 +49,11 @@ widget:
|
|
72 |
|
73 |
## Model description
|
74 |
|
75 |
-
The **roberta-base-ca-cased-
|
76 |
|
77 |
## Intended Uses and Limitations
|
78 |
|
79 |
-
**roberta-base-ca-cased-
|
80 |
|
81 |
## How to Use
|
82 |
|
@@ -84,18 +61,18 @@ Here is how to use this model:
|
|
84 |
|
85 |
```python
|
86 |
from transformers import pipeline
|
87 |
-
|
88 |
-
|
89 |
-
|
90 |
-
|
91 |
-
|
92 |
-
|
93 |
-
pprint(pos_results)
|
94 |
```
|
|
|
95 |
## Training
|
96 |
|
97 |
### Training data
|
98 |
-
We used the
|
99 |
|
100 |
### Training Procedure
|
101 |
The model was trained with a batch size of 16 and a learning rate of 5e-5 for 5 epochs. We then selected the best checkpoint using the downstream task metric in the corresponding development set and then evaluated it on the test set.
|
@@ -106,16 +83,16 @@ The model was trained with a batch size of 16 and a learning rate of 5e-5 for 5
|
|
106 |
|
107 |
This model was finetuned maximizing F1 score.
|
108 |
|
109 |
-
|
110 |
-
We evaluated the _roberta-base-ca-cased-
|
111 |
|
112 |
|
113 |
-
| Model |
|
114 |
-
|
|
115 |
-
| roberta-base-ca-cased-
|
116 |
-
| mBERT |
|
117 |
-
| XLM-RoBERTa |
|
118 |
-
| WikiBERT-ca |
|
119 |
|
120 |
For more details, check the fine-tuning and evaluation scripts in the official [GitHub repository](https://github.com/projecte-aina/club).
|
121 |
|
@@ -123,7 +100,7 @@ For more details, check the fine-tuning and evaluation scripts in the official [
|
|
123 |
|
124 |
[Apache License, Version 2.0](https://www.apache.org/licenses/LICENSE-2.0)
|
125 |
|
126 |
-
## Citation Information
|
127 |
If you use any of these resources (datasets or models) in your work, please cite our latest paper:
|
128 |
```bibtex
|
129 |
@inproceedings{armengol-estape-etal-2021-multilingual,
|
@@ -148,11 +125,9 @@ If you use any of these resources (datasets or models) in your work, please cite
|
|
148 |
```
|
149 |
|
150 |
### Funding
|
151 |
-
|
152 |
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).
|
153 |
|
154 |
|
155 |
-
|
156 |
## Contributions
|
157 |
|
158 |
-
[N/A]
|
1 |
---
|
|
|
2 |
language:
|
|
|
3 |
- ca
|
|
|
4 |
license: apache-2.0
|
|
|
5 |
tags:
|
|
|
6 |
- "catalan"
|
7 |
+
- "qa"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
8 |
datasets:
|
9 |
+
- "xquad-ca"
|
10 |
+
- "viquiquad"
|
|
|
11 |
metrics:
|
12 |
+
- "f1"
|
13 |
+
- "exact match"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
14 |
widget:
|
15 |
+
- text: "Quan va començar el Super3?"
|
16 |
+
context: "El Super3 o Club Super3 és un univers infantil català creat a partir d'un programa emès per Televisió de Catalunya des del 1991. Està format per un canal de televisió, la revista Súpers!, la Festa dels Súpers i un club que té un milió i mig de socis."
|
17 |
+
|
18 |
+
- text: "Quants eren els germans Marx?"
|
19 |
+
context: "Els germans Marx van ser un grup de còmics dels Estats Units que originàriament estava compost per cinc germans (entre parèntesis els noms artístics): Leonard (Chico), Adolph (Harpo), Julius (Groucho), Milton (Gummo) i Herbert (Zeppo)."
|
20 |
+
|
21 |
+
- text: "On van ser els Jocs Olímpics de 1992?"
|
22 |
+
context: "Els Jocs Olímpics d'estiu de 1992, oficialment Jocs Olímpics de la XXV Olimpíada, es van celebrar a la ciutat de Barcelona entre els dies 25 de juliol i 9 d'agost de 1992. Hi participaren 9.356 atletes (6.652 homes i 2.704 dones) de 169 comitès nacionals, que competiren en 32 esports i 286 especialitats."
|
23 |
+
|
24 |
+
- text: "Qui va dissenyar la Sagrada Família?"
|
25 |
+
context: "El Temple Expiatori de la Sagrada Família, conegut habitualment com la Sagrada Família, és una basílica catòlica situada a la ciutat de Barcelona. És un dels exemples més coneguts del modernisme català i un edifici únic al món, que ha esdevingut tot un símbol de la ciutat. Obra inacabada de l'arquitecte català Antoni Gaudí, és al barri de la Sagrada Família, al districte de l'Eixample de la ciutat."
|
26 |
+
|
27 |
+
- text: "Quin és el tercer volcà més gran de la Terra?"
|
28 |
+
context: "El Teide (o Pic del Teide) és un estratovolcà i muntanya de Tenerife, Illes Canàries (28.27 N, 16.6 O). Amb una altitud de 3718 m sobre el nivell del mar i amb aproximadament uns 7000 m sobre el llit marí adjacent, és la muntanya més alta d'Espanya, la muntanya més alta de totes les illes atlàntiques i el tercer volcà més gran de la Terra."
|
29 |
|
|
|
|
|
|
|
|
|
|
|
30 |
|
31 |
---
|
32 |
|
33 |
+
# Catalan BERTa (roberta-base-ca) finetuned for Question Answering.
|
34 |
|
35 |
## Table of Contents
|
36 |
- [Model Description](#model-description)
|
49 |
|
50 |
## Model description
|
51 |
|
52 |
+
The **roberta-base-ca-cased-qa** is a Question Answering (QA) model for the Catalan language fine-tuned from the roberta-base-ca model, a [RoBERTa](https://arxiv.org/abs/1907.11692) base model pre-trained on a medium-size corpus collected from publicly available corpora and crawlers.
|
53 |
|
54 |
## Intended Uses and Limitations
|
55 |
|
56 |
+
**roberta-base-ca-cased-qa** model can be used for extractive question answering. The model is limited by its training dataset and may not generalize well for all use cases.
|
57 |
|
58 |
## How to Use
|
59 |
|
61 |
|
62 |
```python
|
63 |
from transformers import pipeline
|
64 |
+
nlp = pipeline("question-answering", model="projecte-aina/roberta-base-ca-cased-qa")
|
65 |
+
text = "Quan va començar el Super3?"
|
66 |
+
context = "El Super3 o Club Super3 és un univers infantil català creat a partir d'un programa emès per Televisió de Catalunya des del 1991. Està format per un canal de televisió, la revista Súpers!, la Festa dels Súpers i un club que té un milió i mig de socis."
|
67 |
+
|
68 |
+
qa_results = nlp(text, context)
|
69 |
+
print(qa_results)
|
|
|
70 |
```
|
71 |
+
|
72 |
## Training
|
73 |
|
74 |
### Training data
|
75 |
+
We used the QA dataset in Catalan called [CatalanQA](https://huggingface.co/datasets/projecte-aina/catalanqa) for training and evaluation, and the [XQuAD-ca](https://huggingface.co/datasets/projecte-aina/xquad-ca) test set for evaluation.
|
76 |
|
77 |
### Training Procedure
|
78 |
The model was trained with a batch size of 16 and a learning rate of 5e-5 for 5 epochs. We then selected the best checkpoint using the downstream task metric in the corresponding development set and then evaluated it on the test set.
|
83 |
|
84 |
This model was finetuned maximizing F1 score.
|
85 |
|
86 |
+
### Evaluation results
|
87 |
+
We evaluated the _roberta-base-ca-cased-qa_ on the CatalanQA and XQuAD-ca test sets against standard multilingual and monolingual baselines:
|
88 |
|
89 |
|
90 |
+
| Model | ViquiQuAD (F1/EM) | XQuAD-ca (F1/EM) |
|
91 |
+
| ------------|:-------------:| -----:|
|
92 |
+
| roberta-base-ca-cased-qa | **86.99/73.25** | **67.81/49.43** |
|
93 |
+
| mBERT | 86.97/72.22 | 67.15/46.51 |
|
94 |
+
| XLM-RoBERTa | 85.50/70.47 | 67.10/46.42 |
|
95 |
+
| WikiBERT-ca | 85.45/70.75 | 65.21/36.60 |
|
96 |
|
97 |
For more details, check the fine-tuning and evaluation scripts in the official [GitHub repository](https://github.com/projecte-aina/club).
|
98 |
|
100 |
|
101 |
[Apache License, Version 2.0](https://www.apache.org/licenses/LICENSE-2.0)
|
102 |
|
103 |
+
## Citation Information
|
104 |
If you use any of these resources (datasets or models) in your work, please cite our latest paper:
|
105 |
```bibtex
|
106 |
@inproceedings{armengol-estape-etal-2021-multilingual,
|
125 |
```
|
126 |
|
127 |
### Funding
|
|
|
128 |
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).
|
129 |
|
130 |
|
|
|
131 |
## Contributions
|
132 |
|
133 |
+
[N/A]
|