Hugging Face's logo

osiria
/
roberta-large-italian

Fill-Mask
Transformers
PyTorch
Italian
roberta
Inference Endpoints
Model card Files Community
osiria commited on
Commit
8b5edef
•
1 Parent(s): 6ce09c1

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +79 -0
README.md CHANGED
@@ -1,3 +1,82 @@
1
  ---
2
  license: mit
 
 
 
 
 
 
 
 
 
3
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
  license: mit
3
+ language:
4
+ - it
5
+ widget:
6
+ - text: "Milano è una <mask> italiana"
7
+ example_title: "Example 1"
8
+ - text: "Leopardi è stato uno dei più grandi <mask> del classicismo italiano"
9
+ example_title: "Example 2"
10
+ - text: "L'Italia è uno <mask> dell'Unione Europea"
11
+ example_title: "Example 3"
12
  ---
13
+ --------------------------------------------------------------------------------------------------
14
+
15
+ <body>
16
+ <span class="vertical-text" style="background-color:lightgreen;border-radius: 3px;padding: 3px;"> </span>
17
+ <br>
18
+ <span class="vertical-text" style="background-color:orange;border-radius: 3px;padding: 3px;">  </span>
19
+ <br>
20
+ <span class="vertical-text" style="background-color:lightblue;border-radius: 3px;padding: 3px;">    Model: RoBERTa Large</span>
21
+ <br>
22
+ <span class="vertical-text" style="background-color:tomato;border-radius: 3px;padding: 3px;">    Lang: IT</span>
23
+ <br>
24
+ <span class="vertical-text" style="background-color:lightgrey;border-radius: 3px;padding: 3px;">  </span>
25
+ <br>
26
+ <span class="vertical-text" style="background-color:#CF9FFF;border-radius: 3px;padding: 3px;"> </span>
27
+ </body>
28
+
29
+ --------------------------------------------------------------------------------------------------
30
+
31
+ <h3>Model description</h3>
32
+
33
+ This is a <b>RoBERTa Large</b> <b>[1]</b> model for the <b>Italian</b> language, obtained using <b>XLM-RoBERTa-Large</b> <b>[2]</b> ([xlm-roberta-large](https://huggingface.co/xlm-roberta-large)) as a starting point and focusing it on the italian language by modifying the embedding layer
34
+ (as in <b>[3]</b>, computing document-level frequencies over the <b>Wikipedia</b> dataset)
35
+
36
+ The resulting model has 356M parameters, a vocabulary of 50.670 tokens, and a size of ~1.42 GB.
37
+
38
+ <h3>Quick usage</h3>
39
+
40
+ ```python
41
+ from transformers import RobertaTokenizerFast, RobertaForMaskedLM
42
+
43
+ tokenizer = RobertaTokenizerFast.from_pretrained("osiria/roberta-large-italian")
44
+ model = RobertaForMaskedLM.from_pretrained("osiria/roberta-large-italian")
45
+
46
+ pipe = pipeline("fill-mask", model=model, tokenizer=tokenizer)
47
+
48
+ pipe("Milano è una <mask> italiana")
49
+
50
+ [{'score': 0.9284337759017944,
51
+ 'token': 7786,
52
+ 'token_str': 'città',
53
+ 'sequence': 'Milano è una città italiana'},
54
+ {'score': 0.03296631574630737,
55
+ 'token': 26960,
56
+ 'token_str': 'capitale',
57
+ 'sequence': 'Milano è una capitale italiana'},
58
+ {'score': 0.015821034088730812,
59
+ 'token': 8043,
60
+ 'token_str': 'provincia',
61
+ 'sequence': 'Milano è una provincia italiana'},
62
+ {'score': 0.007335659582167864,
63
+ 'token': 18841,
64
+ 'token_str': 'regione',
65
+ 'sequence': 'Milano è una regione italiana'},
66
+ {'score': 0.006183209829032421,
67
+ 'token': 50152,
68
+ 'token_str': 'cittadina',
69
+ 'sequence': 'Milano è una cittadina italiana'}]
70
+ ```
71
+
72
+ <h3>References</h3>
73
+
74
+ [1] https://arxiv.org/abs/1907.11692
75
+
76
+ [2] https://arxiv.org/abs/1911.02116
77
+
78
+ [3] https://arxiv.org/abs/2010.05609
79
+
80
+ <h3>License</h3>
81
+
82
+ The model is released under <b>MIT</b> license