Quinten Datalab
commited on
Commit
•
26fe588
1
Parent(s):
c82f21e
Update README.md
Browse filesDetails of the model.
README.md
CHANGED
@@ -5,7 +5,7 @@ language:
|
|
5 |
library_name: transformers
|
6 |
tags:
|
7 |
- Biomedical
|
8 |
-
-
|
9 |
- French-Biomedical
|
10 |
Mask token:
|
11 |
- [MASK]
|
@@ -21,4 +21,70 @@ widget:
|
|
21 |
- text: "La prise de greffe a été systématiquement réalisée au niveau de la face interne de la [MASK] afin de limiter la plaie cicatricielle."
|
22 |
example_title: "Example 5"
|
23 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
24 |
AliBERT: A Pre-trained Language Model for French Biomedical Text
|
|
|
5 |
library_name: transformers
|
6 |
tags:
|
7 |
- Biomedical
|
8 |
+
- Medical
|
9 |
- French-Biomedical
|
10 |
Mask token:
|
11 |
- [MASK]
|
|
|
21 |
- text: "La prise de greffe a été systématiquement réalisée au niveau de la face interne de la [MASK] afin de limiter la plaie cicatricielle."
|
22 |
example_title: "Example 5"
|
23 |
---
|
24 |
+
|
25 |
+
# quinten-datalab/AliBERT-7GB: AliBERT: is a pre-trained language model for French biomedical text. It is trained with masked language model like RoBERTa.
|
26 |
+
|
27 |
+
# Introduction
|
28 |
+
|
29 |
+
AliBERT: is a pre-trained language model for French biomedical text. It is trained with masked language model like RoBERTa.
|
30 |
+
|
31 |
+
Here are the main contributions of our work:
|
32 |
+
A French biomedical language model, a language-specific and domain-specific PLM, which can be used to represent French biomedical text for different downstream tasks.
|
33 |
+
A normalization of a Unigram sub-word tokenization of French biomedical textual input which improves our vocabulary and overall performance of the models trained.
|
34 |
+
AliBERT outperforms other French PLMs in different downstream tasks. It is a foundation model that achieved state-of-the-art results on French biomedical text.
|
35 |
+
|
36 |
+
# Data
|
37 |
+
The pre-training corpus was gathered from different sub-corpora.It is composed of 7GB French biomedical textual documents. Here are the sources used.
|
38 |
+
|
39 |
+
|Dataset name| Quantity| Size |
|
40 |
+
|----|---|---|
|
41 |
+
|Drug database| 23K| 550Mb |
|
42 |
+
|RCP| 35K| 2200Mb|
|
43 |
+
|Articles| 500K| 4300Mb |
|
44 |
+
|Thesis| 300K|300Mb |
|
45 |
+
|Cochrane| 7.6K| 27Mb|
|
46 |
+
|
47 |
+
|
48 |
+
# How to use alibert-quinten/Oncology-NER with HuggingFace
|
49 |
+
|
50 |
+
Load quinten-datalab/AliBERT-7GB fill-mask model and the tokenizer used to train AliBERT:
|
51 |
+
|
52 |
+
```python
|
53 |
+
from transformers import AutoTokenizer, AutoModelForTokenClassification,pipeline
|
54 |
+
|
55 |
+
tokenizer = AutoTokenizer.from_pretrained("quinten-datalab/AliBERT-7GB")
|
56 |
+
|
57 |
+
model = AutoModelForTokenMaskedLM.from_pretrained("quinten-datalab/AliBERT-7GB")
|
58 |
+
|
59 |
+
fill_mask=pipeline("fill-mask",model=model,tokenizer=tokenizer)
|
60 |
+
nlp_AliBERT=fill_mask("La prise de greffe a été systématiquement réalisée au niveau de la face interne de la [MASK] afin de limiter la plaie cicatricielle.")
|
61 |
+
|
62 |
+
[{'score': 0.7724128365516663,
|
63 |
+
'token': 6749,
|
64 |
+
'token_str': 'cuisse',
|
65 |
+
'sequence': 'La prise de greffe a été systématiquement réalisée au niveau de la face interne de la cuisse afin de limiter la plaie cicatricielle.'},
|
66 |
+
{'score': 0.09472355246543884,
|
67 |
+
'token': 4915,
|
68 |
+
'token_str': 'jambe',
|
69 |
+
'sequence': 'La prise de greffe a été systématiquement réalisée au niveau de la face interne de la jambe afin de limiter la plaie cicatricielle.'},
|
70 |
+
{'score': 0.03340734913945198,
|
71 |
+
'token': 2050,
|
72 |
+
'token_str': 'main',
|
73 |
+
'sequence': 'La prise de greffe a été systématiquement réalisée au niveau de la face interne de la main afin de limiter la plaie cicatricielle.'},
|
74 |
+
{'score': 0.030924487859010696,
|
75 |
+
'token': 844,
|
76 |
+
'token_str': 'face',
|
77 |
+
'sequence': 'La prise de greffe a été systématiquement réalisée au niveau de la face interne de la face afin de limiter la plaie cicatricielle.'},
|
78 |
+
{'score': 0.012518334202468395,
|
79 |
+
'token': 3448,
|
80 |
+
'token_str': 'joue',
|
81 |
+
'sequence': 'La prise de greffe a été systématiquement réalisée au niveau de la face interne de la joue afin de limiter la plaie cicatricielle.'}]
|
82 |
+
```
|
83 |
+
|
84 |
+
## Metrics and results
|
85 |
+
The model has been evaluted in the following downstream tasks
|
86 |
+
|
87 |
+
## Biomedical Named Entity Recognition (NER)
|
88 |
+
|
89 |
+
##
|
90 |
AliBERT: A Pre-trained Language Model for French Biomedical Text
|