Update README.md
Browse files
README.md
CHANGED
@@ -74,12 +74,103 @@ You can use the raw model for fill mask or fine-tune it to a downstream task.
|
|
74 |
|
75 |
## How to use
|
76 |
Here is how to use this model:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
77 |
|
78 |
|
79 |
## Limitations and bias
|
80 |
|
81 |
At the time of submission, no measures have been taken to estimate the bias and toxicity embedded in the model. However, we are well aware that our models may be biased since the corpora have been collected using crawling techniques on multiple web sources. We intend to conduct research in these areas in the future, and if completed, this model card will be updated. Nevertheless, here's an example of how the model can have biased predictions:
|
82 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
83 |
|
84 |
## Training
|
85 |
|
|
|
74 |
|
75 |
## How to use
|
76 |
Here is how to use this model:
|
77 |
+
```
|
78 |
+
python
|
79 |
+
>>> from transformers import pipeline
|
80 |
+
>>> from pprint import pprint
|
81 |
+
>>> unmasker = pipeline('fill-mask', model='PlanTL-GOB-ES/roberta-base-bne')
|
82 |
+
>>> pprint(unmasker("Gracias a los datos de la BNE se ha podido <mask> este modelo del lenguaje."))
|
83 |
+
[{'score': 0.08422081917524338,
|
84 |
+
'token': 3832,
|
85 |
+
'token_str': ' desarrollar',
|
86 |
+
'sequence': 'Gracias a los datos de la BNE se ha podido desarrollar este modelo del lenguaje.'},
|
87 |
+
{'score': 0.06348305940628052,
|
88 |
+
'token': 3078,
|
89 |
+
'token_str': ' crear',
|
90 |
+
'sequence': 'Gracias a los datos de la BNE se ha podido crear este modelo del lenguaje.'},
|
91 |
+
{'score': 0.06148449331521988,
|
92 |
+
'token': 2171,
|
93 |
+
'token_str': ' realizar',
|
94 |
+
'sequence': 'Gracias a los datos de la BNE se ha podido realizar este modelo del lenguaje.'},
|
95 |
+
{'score': 0.056218471378088,
|
96 |
+
'token': 10880,
|
97 |
+
'token_str': ' elaborar',
|
98 |
+
'sequence': 'Gracias a los datos de la BNE se ha podido elaborar este modelo del lenguaje.'},
|
99 |
+
{'score': 0.05133328214287758,
|
100 |
+
'token': 31915,
|
101 |
+
'token_str': ' validar',
|
102 |
+
'sequence': 'Gracias a los datos de la BNE se ha podido validar este modelo del lenguaje.'}]
|
103 |
+
```
|
104 |
+
|
105 |
+
Here is how to use this model to get the features of a given text in PyTorch:
|
106 |
+
|
107 |
+
```python
|
108 |
+
>>> from transformers import RobertaTokenizer, RobertaModel
|
109 |
+
>>> tokenizer = RobertaTokenizer.from_pretrained('PlanTL-GOB-ES/roberta-base-bne')
|
110 |
+
>>> model = RobertaModel.from_pretrained('PlanTL-GOB-ES/roberta-base-bne')
|
111 |
+
>>> text = "Gracias a los datos de la BNE se ha podido desarrollar este modelo del lenguaje."
|
112 |
+
>>> encoded_input = tokenizer(text, return_tensors='pt')
|
113 |
+
>>> output = model(**encoded_input)
|
114 |
+
>>> print(output.last_hidden_state.shape)
|
115 |
+
torch.Size([1, 19, 768])
|
116 |
+
```
|
117 |
|
118 |
|
119 |
## Limitations and bias
|
120 |
|
121 |
At the time of submission, no measures have been taken to estimate the bias and toxicity embedded in the model. However, we are well aware that our models may be biased since the corpora have been collected using crawling techniques on multiple web sources. We intend to conduct research in these areas in the future, and if completed, this model card will be updated. Nevertheless, here's an example of how the model can have biased predictions:
|
122 |
|
123 |
+
```python
|
124 |
+
>>> from transformers import pipeline, set_seed
|
125 |
+
>>> from pprint import pprint
|
126 |
+
>>> unmasker = pipeline('fill-mask', model='PlanTL-GOB-ES/roberta-base-bne')
|
127 |
+
>>> set_seed(42)
|
128 |
+
>>> pprint(unmasker("Antonio está pensando en <mask>."))
|
129 |
+
[{'score': 0.07950365543365479,
|
130 |
+
'sequence': 'Antonio está pensando en ti.',
|
131 |
+
'token': 486,
|
132 |
+
'token_str': ' ti'},
|
133 |
+
{'score': 0.03375273942947388,
|
134 |
+
'sequence': 'Antonio está pensando en irse.',
|
135 |
+
'token': 13134,
|
136 |
+
'token_str': ' irse'},
|
137 |
+
{'score': 0.031026942655444145,
|
138 |
+
'sequence': 'Antonio está pensando en casarse.',
|
139 |
+
'token': 24852,
|
140 |
+
'token_str': ' casarse'},
|
141 |
+
{'score': 0.030703715980052948,
|
142 |
+
'sequence': 'Antonio está pensando en todo.',
|
143 |
+
'token': 665,
|
144 |
+
'token_str': ' todo'},
|
145 |
+
{'score': 0.02838558703660965,
|
146 |
+
'sequence': 'Antonio está pensando en ello.',
|
147 |
+
'token': 1577,
|
148 |
+
'token_str': ' ello'}]
|
149 |
+
|
150 |
+
>>> set_seed(42)
|
151 |
+
>>> pprint(unmasker("Mohammed está pensando en <mask>."))
|
152 |
+
[{'score': 0.05433618649840355,
|
153 |
+
'sequence': 'Mohammed está pensando en morir.',
|
154 |
+
'token': 9459,
|
155 |
+
'token_str': ' morir'},
|
156 |
+
{'score': 0.0400255024433136,
|
157 |
+
'sequence': 'Mohammed está pensando en irse.',
|
158 |
+
'token': 13134,
|
159 |
+
'token_str': ' irse'},
|
160 |
+
{'score': 0.03705748915672302,
|
161 |
+
'sequence': 'Mohammed está pensando en todo.',
|
162 |
+
'token': 665,
|
163 |
+
'token_str': ' todo'},
|
164 |
+
{'score': 0.03658654913306236,
|
165 |
+
'sequence': 'Mohammed está pensando en quedarse.',
|
166 |
+
'token': 9331,
|
167 |
+
'token_str': ' quedarse'},
|
168 |
+
{'score': 0.03329474478960037,
|
169 |
+
'sequence': 'Mohammed está pensando en ello.',
|
170 |
+
'token': 1577,
|
171 |
+
'token_str': ' ello'}]
|
172 |
+
```
|
173 |
+
|
174 |
|
175 |
## Training
|
176 |
|